Está en la página 1de 615

Editor-in-Chief
John H. Byrne

Department of Neurobiology & Anatomy,
The University of Texas Medical School at Houston,
Houston, Texas, USA

Volume Editors

Volume 1
LEARNING THEORY AND BEHAVIOUR

Volume Editor
Randolf Menzel
Institut für Biologie – Neurobiologie, Freie Universität Berlin, Berlin, Germany

Volume 2
COGNITIVE PSYCHOLOGY OF MEMORY

Volume Editor
Henry L. Roediger III
Department of Psychology, Washington University in St. Louis, St. Louis, Missouri, USA

Volume 3
MEMORY SYSTEMS

Volume Editor
Howard Eichenbaum
Department of Psychology, Boston University, Boston, Massachusetts, USA

Volume 4
MOLECULAR MECHANISMS OF MEMORY

Volume Editor
J. David Sweatt
Department of Neurobiology and McKnight Brain Institute,
University of Alabama at Birmingham, Birmingham, Alabama, USA

FOREWORD

A comprehensive reference work on learning and memory could not be better timed than this. During the
second half of the twentieth century, the study of learning and memory moved from a descriptive
science largely based on the pioneering behavioral analyses of Pavlov, Thorndike, Watson, Skinner, Kamin,
Rescorla, and Wagner to a new mechanistic science of mind that combines these brilliant behavioral studies
with an analysis of the underlying neural mechanisms, first in a regional manner by Milner, Tulving, Mishkin,
Squire, Schachter, and Morris, then on the cellular level, and finally on the molecular level.
The challenges that now face the field are outlined by the five great pioneers in the study of memory – the
editor-in-chief Jack Byrne and the editors of these four extraordinary volumes: Learning Theory and Behavior,
edited by Randolf Menzel; Cognitive Psychology of Memory, edited by Henry Roediger; Memory Systems, edited by
Howard Eichenbaum; and Molecular Mechanisms of Memory, edited by David Sweatt. The challenge faced by the
contributors to these volumes was to combine the molecular mechanisms with the other three levels in order to
provide a coherent, systematically and intellectually satisfying understanding of learning and memory. This is
central to the new science of mind. Since memory is the glue that holds our mental life together, the topics
covered by these four volumes are central to and paradigmatic for all aspects of the neurobiology of mental life,
which has as its goal the understanding of all mental processes in neurobiological terms. Indeed, it is the
plasticity of the brain that is the key to understanding the continuity of all mental function. The goal for each of
these four volumes was to bridge the subdisciplines concerned with the various forms of memory into a
coherent science. The chapters of each of these volumes succeed admirably in doing just that. As a result, this
rich and rewarding reference work will serve as a superb framework for the decades ahead, a reference that will
provide both the student and the working scientist with the intellectual background necessary to understand
and function effectively in the study of learning and memory.

Eric R. Kandel, M.D.
University Professor, Fred Kavli Professor and Director, Kavli Institute for Brain Sciences
Senior Investigator, Howard Hughes Medical Institute, Center for Neurobiology and Behavior
Columbia University, New York, NY, USA

xvii

PREFACE

L earning and Memory: A Comprehensive Reference is the most authoritative set of volumes ever produced on
learning and memory and represents the state of the science in the early 21st century. The study of
learning (the process of acquiring new information) and memory (retention of that information for future use)
has intrigued philosophers and writers for centuries because our memories and plans for the future consolidate
who we are, and disruption of these processes dramatically interferes with our daily lives. The fascination with
learning and memory is not limited to the humanities, but has been the subject of intense scientific research.
Psychologists are concerned with elucidating the features of learning and memory processes and systems,
neurobiologists seek to determine the neuronal mechanisms of learning and memory, and neurologists and
psychiatrists focus on research and treatment of failures or disruptions in learning and memory.
The study of learning and memory represents a scientific field that has matured at all levels – from the
discovery of the protein chemistry and molecular biology of the cellular events underlying learning and
memory, through the delineations of the properties and functions of neuronal networks, to formulating and
testing the psychological and behavioral neuroscientific theories of learning and memory. In addition, many
basic research findings have applied implications on such diverse fronts as education, legal issues hinging on
eyewitness testimony, learning disorders in children, memory disorders following brain damage, and declines in
memory in older adults.
The volumes in this Comprehensive Reference are the result of a meeting in London in July of 2005 where the
editors planned the massive work of consolidating all facets of the study of learning and memory. We collected
nearly all the topics (albeit from many different disciplines and directions) that we considered constituted
scientific approaches to learning and memory and proceeded to parcel the topics into four volumes, resulting in
Learning Theory and Behavior edited by Randolf Menzel; Cognitive Psychology of Memory edited by Henry Roediger
III; Memory Systems edited by Howard Eichenbaum; and Molecular Mechanisms of Memory edited by David Sweatt.
This was a formidable task, not only because of the richness and diversity of the subject matter, but also because
we needed to logically place topics in the appropriate volume. Although some of the decisions may seem
arbitrary, and indeed there is overlap both within and between volumes, each editor ended up with a set of
coherent topics that they could organize and introduce in a logical manner.
With approximately 40 chapters per volume, it is no surprise that the editors cover an unusually wide range
of intellectual territory or that there is a difference in interpretation by some authors. The organization is a
significant editorial challenge and investment in and of itself. However, it is the editor’s selection of authors, and
the ensuing scholarship on learning and memory from different perspectives, that make this series unique.
Authors were identified and invited based on their expertise on a particular topic, and their contributions
represent a marvelous compendium of research in learning and memory. The chapters in this series not only
represent scientific strength and breadth, but also range from learning at the synaptic level to a systems level
approach, and include studies of remarkable learning capabilities in a variety of invertebrates and vertebrates,
including human beings.
The first volume in the series, Learning Theory and Behavior edited by Randolf Menzel, consists of 38 chapters
and sets the tone for the interdisciplinary and comparative approach to the study of learning and memory. He
introduces the volume by emphasizing both the value and the limitation of the comparative approach in natural
and laboratory settings, stressing that we need information from the behaving animal as well as the neuronal

xix

xx Preface

structures in order to understand the processes involved in information storage and retrieval. Several chapters
review progress from using animal models, including worms, molluscs, insects, rodents, birds, and nonhuman
and human primates. In addition, concepts such as planning, decision-making, self-awareness and episodic-like
memory, usually reserved for human beings, are discussed at several taxonomic levels. The final chapters take
an engineering perspective and describe synthetic approaches, including modeling neuronal function and
developing a concise theory of the brain.
The second volume, Cognitive Psychology of Learning edited by H. Roediger, is comprised of 48 chapters on
various aspects of cognitive ability and the underlying neuroscience. The basics of attention, working memory,
forgetting, false memories, remembering vs. knowing, the process of recognition, and episodic memory are
covered. In addition, topics that are often not included in ‘‘memory’’ volumes deservedly receive attention here,
e.g., learning of concepts and categories, learning of perceptual and motor skills, language learning, and implicit
learning. This volume also covers memory processes throughout the human lifespan and includes chapters on
individual differences in memory ability, both subnormal (learning disabilities) and supranormal (performance
of mnemonists and experts in particular domains). Finally, chapters on applied aspects of memory research,
dealing with such topics as eyewitness identification in the legal system and applications of research to
educational issues, are included.
Volume 3, edited by H. Eichenbaum, consists of 29 chapters which represent a ‘‘progress report’’ on what we
know about memory systems and their relationship to different parts of the brain. Memory Systems returns to a
comparative approach of learning and memory. This volume introduces the concepts of multiple memory
systems, and many chapters discuss in extensive detail the different features of declarative memory and their
underlying brain structures. Procedural learning in humans and other animals is addressed, and a short section
details the involvement of hormones and emotions on memory retention or loss. Finally, changes in memory
systems associated with aging, disease processes, and drug use are addressed.
The final 42 chapters in Volume 4, Molecular and Cellular Mechanisms of Memory edited by J.D. Sweatt,
represent a review of the state of the science of what we know at the systems, cell, and molecular levels on
learning and memory formation, as well as providing a look at the emerging and future areas of investigation.
Once again, this volume covers an impressive amount of information derived from studies at many taxonomic
levels, from molecular associative learning mechanisms, through an array of studies on synaptic plasticity, to the
cell level of fear conditioning.
The centrality of learning and memory to our daily lives has led to intense analysis by psychologists and
neurobiologists for the past century, and it will undoubtedly remain at the forefront of research throughout this
new century as well. It is our intention that this set of volumes will contribute significantly to the consolidation
of this field, and it is meant as a resource for scientists and students interested in all facets of learning and
memory. No other reference work covers so wide a territory and in so much depth.
Learning and Memory: A Comprehensive Reference would not have been possible without the tremendous work of
the Editorial Board, who identified the topics and their authors, and reviewed each contribution. Special thanks
also go to Johannes Menzel, Senior Acquisitions Editor at Elsevier, for supporting the project and Andrew Lowe
and Laura Jackson, Production Project Managers, and Joanna De Souza, Developmental Editor, for ensuring
that the production schedule was maintained.

John H. Byrne

Permission
Acknowledgement
The following material is reproduced with kind permission of Nature Publishing Group
Figure 1 of Neurofibromatosis Type I Learning Disabilities
Figures 2 & 5 of Second Messengers: Calcium and cAMP Signaling
Figure 1b of Action Potentials in Dendrites and Spike-Timing-Dependent Plasticity
Figure 4 of Neurogenesis
Figure 12a-c of Neural and Molecular Mechanisms of Fear Memory
Figures 3 & 4 of Transmission of Acquired Information in Nonhuman Primates
Figure 4a-b of Behavioral Analysis of Learning and Memory in: C. elegans
Figures 2a, 6a-c, 7, 8a-b, 10a-b & 12a-b of Navigation and Episodic-like memory in Mammals
Figures 1, 4 & 6 of Animal models of amnesia
Figure 4a-e of Cortical Plasticity in Associative Learning and Memory
Figures 7a-b & 9a-b of Neurophysiology of Birdsong learning
Figure 6a-b of Visual Priming
Figures 2a & 4 of The Role of Sleep in Memory Consolidation

The following material is reproduced with kind permission of American Association for the
Advancement of Science
Figures 13 & 14 of Cognitive dimension of operant learning

The following material is reproduced with kind permission of Taylor & Francis Ltd
Figure 10 of Learning to Time Intervals

xxi

3.01 Introduction and Overview
H. Eichenbaum, Boston University, Boston, MA, USA
ª 2008 Elsevier Ltd. All rights reserved.

3.01.1 Early Ideas About Multiple Forms of Memory 1
3.01.2 The Cognitive Neuroscience Revolution 2
3.01.3 The Organization of This Volume 5
3.01.4 The Declarative Memory System 5
3.01.5 The Cerebral Cortex and Memory 6
3.01.6 Procedural Learning and Memory 7
3.01.7 Emotional Memory and Modulation of Memory 7
3.01.8 Aging and Memory 8
References 8

In 2001, Neal Cohen and I reviewed the existing 3.01.1 Early Ideas About Multiple
literature on the analysis of brain systems that support Forms of Memory
memory (Eichenbaum and Cohen, 2001) and reached
two central conclusions. First, memory is a conse- Modern ideas about multiple forms of memory
quence of the fundamental plasticity of the brain. foreshadowed current conceptions that have come
Accordingly, memory storage is intimately tied to from technologically driven scientific investigations.
ongoing information processing in the brain. Second, Indeed, the fundamental insights of the current era
because the brain is organized into several functional can be viewed as resurrecting basic theories proposed
systems, there are multiple forms of memory that have many years ago by thinkers who had profound
distinct psychological and information processing insights that remain key to our modern understanding
characteristics, composing multiple, functionally and of learning and memory. The first elaborated proposal
anatomically distinct memory systems. According to about multiple forms of memory may have come from
this view, memory is not an entity stored in a localized Aristotle (c. 350 BC), who distinguished memory and
warehouse, nor is it fully distributed in an intercon- reminiscence. He proposed that memory is an exten-
nected network of the brain’s neurons. Instead, sion of the senses that incorporates the passage of time
memory should be conceived of as a fundamental since the actual sensory experience. He envisioned a
property of brain systems, and the consequent changes memory as a replicate of a perceived object on some
in behavior and remembered information as a natural sort of recording device and compared it to an
outcome of the brain’s various processing activities. impression on a plastic surface. Memory, then, is a
Thus, memory is an integral part of those ongoing passive reperception of the recorded object, indepen-
information processing activities and is tied to the dent of its original context. Aristotle also argued that
brain systems and structures that support those activ- the likeness of the memory to the original percept was
ities. Experience is reflected in changes in the quantitative and involved the knowledge of having
operation of these systems in one manner or another. perceived the item before. Aristotle contrasted this
Indeed, the manner in which memory is expressed perceptually based memory with reminiscence,
varies importantly and is entirely dependent on the which, he asserted, involved the active replay of an
functions of the brain system involved. While there is entire experience, triggered by an initial memory of
enormous consensus on this view, the number and one event in that experience. This first memory
nature of memory systems and their anatomical cir- would lead to a recovered sense of the situation and
cuits are still only partially understood. Investigations flow of events that compose the experience. Thus
aimed to characterize the functional organization of the Aristotle viewed reminiscence as a process of recover-
brain’s memory systems are still at an early stage. This ing single items in succession. The difference
volume should be considered a progress report on our Aristotle conceived between memory and reminis-
understanding of the major memory systems of the cence is recognized today in the distinction we
brain. make between familiarity and recollection. Chapters

1

2 Introduction and Overview

by Voss and Paller (See Chapter 3.05), Nyberg (See mechanism he called habit. James was guided by
Chapter 3.06), and Stern and Hasselmo (See Chapter observations from biology that identified reflex path-
3.08) consider modern psychological and biological ways in the nervous system through which a stimulus
distinctions between familiarity and recollection. (such as a pinch) automatically generates a specific
Two thousand years after Aristotle, a less familiar motor response (withdrawal). Influenced by the
philosopher using the pen name Maine de Biran descriptions of these reflex arcs, James suggested that
(1804/1929) offered a prescient view of different nerve impulses in reflex paths more readily traverse
forms of long-term memory. He began with the paths previously taken. Thus, he wrote, habits form
assumption that there is a simple and automatic when reflex paths become well worn. Like Maine de
mechanism for associating events that underlies Biran, James attributed great importance to habits as
three different forms of memory. The most complex the building blocks of more complicated behaviors.
form was called representative memory, which James said that practiced behaviors and skills are
involves the ability to consciously relive and think mediated by sequentially linked discharges that ‘‘awa-
about a prior experience. The other two forms ken each other in succession’’ through connected reflex
of memory Maine de Biran identified, mechanical paths. He held that this kind of chain reaction mediated
memory and sensitive memory, differed from repre- the production of learned movement sequences. James
sentative memory in two important ways. First, distinguished active memory as a complex cognitive
neither of these forms of memory could access con- phenomenon from the acquisition of skills and from a
sciousness. Second, both of these unconscious forms of sense of familiarity. He also emphasized the warmth
memory had rigid limits on their expression. That is, and intimacy of recollection from the cold repetition of
they could be expressed only under circumstances practiced movements and from the passive feeling of
closely resembling the learning experience, and familiarity for a reexperienced stimulus or event. In
they could not be used to solve new problems. addition, James characterized memory in terms of its
Specifically, mechanical memory involves learning a structure as an elaborate network of associations, vastly
movement through repetition and improves the speed richer and more complicated than a connected series of
or coordination of an action through practice. habits or a general sense of familiarity. Thus the under-
Sensitive memory occurs during emotional experi- lying foundation of recall involves a complex set of
ences and involves recalling a feeling without systematic associations between particular memories.
memory for the circumstances of the emotional situa- James’ characterizations of memory and habit carry
tion. In a chapter here (See Chapter 3.02), White strong similarities with the current views on declara-
presents a modern formulation of these three kinds tive and nondeclarative memory that are strongly
of memory and reviews the evidence of distinct neural dissociated in the phenomenology of amnesia dis-
substrates for each. These ideas are further elaborated cussed by Squire and Shrager (See Chapter 3.04).
in many of the chapters in this volume, with aspects of The fundamental distinctions between familiarity
Maine de Biran’s representative memory discussed in and recollection, and between habits, emotional
chapters on declarative memory, aspects of his memories, and conscious forms of memory, were
mechanical memory discussed in the chapter on pro- initially lost in early experimental psychologists’
cedural memory, and aspects of his sensitive memory efforts to characterize memory as a single system.
discussed in chapters on emotional memory. However, the distinctions were rediscovered and
At the beginning of experimental psychology, confirmed by modern cognitive and neuroscience
William James (1890/1918) made two major distinc- research. This volume provides a survey of our cur-
tions about types of memory. James differentiated rent understanding of the different forms of memory
primary memory, the ability to hold and manipulate and the brain systems that support them.
information in mind for a short period, from the per-
manent memory store he called secondary
memory. The notion of a distinct mechanism and 3.01.2 The Cognitive Neuroscience
brain system for primary memory is maintained Revolution
today in our characterizations of working memory,
discussed in chapters by Buchsbaum and D’Esposito The initial breakthrough in the rediscovery of multi-
(See Chapter 3.13) and Ranganath and Blumenfeld (See ple memory systems came from the study of patients
Chapter 3.14). James also distinguished between with pervasive, global amnesia. Scoville and Milner
secondary memory and a simple and automatic (1957) described a case study that involves probably

Introduction and Overview 3

the most famous neurological patient in the litera- midline diencephalic structures damaged in various
ture, H.M. This patient had the medial temporal lobe amnesias. Procedural memory was seen as being
area removed to alleviate his severe epileptic attacks. mediated by various brain systems specialized for
H.M. consequently suffered what appeared to be a particular types of skilled performance.
nearly complete loss of the ability to form new long- The initial reports on H.M. spurred several efforts
term memories. His impairment, tested over the last to model global amnesia in animals. As a part of those
50þ years, has been shown to extend to verbal investigations, evidence for multiple forms of memory
and nonverbal memory and spatial and nonspatial and theories of the nature of these forms of memory
memory and, indeed, seems to cut across all cate- and the brain structures that support them emerged. In
gories of learning materials. Yet a second line of the mid- and late 1970s three separate theoretical
discovery about global amnesia revealed a spared themes developed, each espousing a multiple memory
domain of learning capacity. Even from the outset, systems approach and each suggesting a selective role
a few exceptions to the otherwise pervasive deficit for the hippocampus in a distinct higher-order form of
were apparent. H.M. was able to learn new motor memory versus hippocampal-independent mecha-
skills, and he showed a facilitation of perceptual nisms for simpler forms of learning (Hirsh, 1974;
identification resulting from prior exposure to O’Keefe and Nadel, 1978; Olton et al., 1979). One
objects or words (an effect that later came to be proposal was that the hippocampal system mediates
understood as reflective of a preserved priming) cognitive mapping, whereas nonhippocampal systems
(See Chapter 3.12). support nonspatial learning and memory for responses
A further breakthrough came in 1980 when Cohen and reward values of stimuli (O’Keefe and Nadel,
and Squire (Cohen and Squire, 1980) proposed that 1978). O’Keefe and Nadel’s analysis of a large litera-
these exceptions to amnesia were indicative of a large ture on the effects of hippocampal lesions and the
domain of preserved learning capacities in amnesia. anatomy and physiology of the hippocampus sup-
Their conclusion was based on the observation of ported their conclusion that the hippocampus plays a
complete preservation of the acquisition and retention critical role in many forms of spatial learning and
of a perceptual skill (reading mirror-reversed words) memory, and conversely, many forms of nonspatial
in amnesic patients. These patients showed fully intact learning and memory did not require hippocampal
skilled performance yet were markedly impaired both function. However, their view also went well beyond
in recognizing the particular words on which they making a simple distinction between spatial and non-
trained and in recollecting their training experiences. spatial learning modalities. Their proposal about
These investigators were struck by the dissociation spatial learning involved the acquisition of cognitive
between the ability to benefit or otherwise have per- maps that corresponded roughly, if not topographi-
formance shaped by a series of training experiences, cally, to the salient features of the physical
an ability that appeared fully normal in the amnesic environment. They referred to the domain of memory
patients, and the capacity to explicitly remember or supported by the hippocampus as a locale system
consciously recollect those training experiences or characterized as maintaining a molar model of space
their contents, which was markedly impaired in the in terms of relations among objects in the environment,
patients. Cohen and Squire attributed the observed as driven by curiosity rather than reinforcement of
dissociation, together with the earlier findings of specific behaviors, and as capable of very rapid learn-
spared memory in amnesia, to the operation of distinct ing. In contrast they proposed that hippocampal-
forms of memory, which they called procedural mem- independent learning is supported by a taxon system
ory and declarative memory, respectively. These characterized as mediating dispositions of specific
forms of memory were viewed as functionally distinct stimuli into categories, as driven by reinforcement of
memory systems, one dedicated to the tuning and approach and avoidance behaviors, and as involving
modification of networks that is expressed by implicit slow and incremental behavioral adaptations.
changes in perceptual, cognitive, or motor perfor- An alternative line of study led Hirsh (1974) to
mance, and the other to the encoding, storage, and propose that the hippocampus supports a capacity for
retrieval that supports explicit expression of memories contextual retrieval, the ability to utilize the context
for specific facts and events. Furthermore, these func- in which conditioned cues occur to retrieve the
tionally distinct memory systems were tied to separate appropriate association. Hirsh’s experiments showed
brain systems, with declarative memory seen as criti- that hippocampal damage affected learning to turn
cally dependent on the medial temporal-lobe and one direction or the other in a T-maze depending on

4 Introduction and Overview

an imposed motivational context (hunger or thirst). 3.07). A general, anatomically based framework for the
The ambiguity in turn direction was, according to major memory systems has emerged from many experi-
this account, resolved by a hippocampal-dependent ments that provide dissociations among the role of
mechanism that employed motivational state as a con- specific brain structures in different forms of memory,
textual cue for retrieving one of the possible responses. combined with the known anatomical pathways of the
Conversely, Hirsh characterized the behavior of ani- key structures. A taxonomy of some of the prominent
mals with hippocampal damage as habit prone, memory pathways currently under investigation is pro-
reflecting simple stimulus–response conditioning. vided in Figure 1 (for a similar outline see Suzuki,
Many experiments have subsequently shown that ani- 1996). In this scheme, the origin of each of the memory
mals with hippocampal lesions are less influenced by systems is the vast expanse of the cerebral cortex,
contextual cues than are normal animals. focusing in particular on the highest stages of the sev-
In 1979, Olton and colleagues proposed that the eral distinct sensory and motor processing hierarchies,
hippocampus is essential for working memory, and the cortical association areas.
nonhippocampal systems support reference memory. Some forms of learning and memory are accom-
Notably, the term working memory as used in this plished mainly within the cerebral cortex. Perceptual
context differs in meaning from the same term used in learning (See Chapter 3.11), involving our capacity
characterizations of human memory (See Chapter 3.13). for familiarity and for categorizing and identifying
As characterized by Olton and colleagues, working stimuli, depends on modifications of the cortical
memory involves memory for specific behavioral epi- areas mediating specific types of stimuli (visual,
sodes and not information accrued over many tactile, and the like) as well as semantic memory
experiences, very similar to the characterization of ep- (See Chapter 3.07). The prefrontal cortex and its
isodic memory (Olton, 1984). Olton and his colleagues interactions with other cortical areas support
provide substantial evidence that animals with hippo- conscious manipulation of information and working
campus damage are severely impaired in radial maze memory (See Chapters 3.13 and 3.14). In addition, the
tasks that involve memory for specific experiences but cerebral cortex provides major inputs to each of three
not memory for consistently rewarded maze locations main pathways of processing related to distinct mem-
or nonspatial stimuli. Thus Olton’s distinction between ory functions. One major pathway is from virtually all
working and reference memory shares much in com- cortical association areas to the hippocampus via
mon with the current distinction between episodic and the parahippocampal region (See Chapter 3.03). As
semantic memory in humans, outlined in chapters introduced above, this pathway supports recognition
by Nyberg (See Chapter 3.06) and Martin and memory (See Chapters 3.08 and 3.10) and declarative
Simmons (See Chapter 3.07). memory in humans (See Chapter 3.04) and their analogs
These efforts focused on the hippocampus and on in animals (See Chapter 3.09). The main output of
distinguishing the kind of memory in which that hippocampal and parahippocampal processing is back
structure plays a critical role versus any other form to the same cortical areas that provided inputs to the
of memory. Many other lines of study now have hippocampus, and these cortical areas are viewed as the
clarified the nature and brain pathways involved in long-term repository of declarative memories.
other forms of memory. Our understanding of these In addition, motor learning is supported by areas
pathways has grown from a diverse range of studies, of the cortex itself (See Chapters 3.21 and 3.22), and
some seeking to anatomically dissociate hippocampal- these areas send inputs to specific subcortical targets
dependent versus hippocampal-independent forms of that are critical nodes in habit and skill learning. One
memory, and others arising from studies on plasticity of these pathways involves the striatum as a nodal
in various functional systems of the brain. Several stage in the association of sensory and motor cortical
distinct lines of study have demonstrated key roles information with voluntary responses via the brain-
for the basal ganglia (See Chapters 3.17 and 3.18), stem motor system (See Chapters 3.17 and 3.18). The
cerebellum (See Chapters 3.19 and 3.20), and motor putative involvement of this pathway in associating
cortex (See Chapters 3.21 and 3.22) in different aspects cortical representations to specific behavioral
of procedural learning, for the amygdala in emotional responses has led many to consider this system as
memory and modulation of memory (See Chapters specialized for habit or skill learning, two forms of
3.24 and 3.26), and for the cerebral cortex in priming procedural memory. An additional, parallel pathway
(See Chapter 3.12), working memory (See that mediates different aspects of sensorimotor adap-
Chapter 3.13), and semantic memory (See Chapter tations involves sensory and motor systems pathways

Introduction and Overview 5

Perceptual Working
memory Cortical association areas memory

Striatum
cerebellum Amygdala Hippocampus

Brainstem and Hypothalamus autonomic and
spinal motor endocrine
outputs outputs

Procedural Emotional Declarative
memory memory memory

Reinforced Conditioned Episodic and
response preferences and semantic
habits aversions recollection

Figure 1 Anatomic pathways of memory systems. Systems for perceptual and working memory involve networks within the
a cortical association areas also send major inputs into and are strongly influenced by other memory systems. The procedural
memory system involves pathways through the striatum and cerebellum that connect directly with motor outputs. The
emotional memory system involves pathways through the amygdala that connect directly with the hypothalamus to direct
autonomic, motor, and endocrine outputs, and the amygdala influences the strength of memories in other systems. The
declarative memory system involves bidirectional interactions between the hippocampus and cortical association areas.

through the cerebellum (See Chapters 3.19 and 3.20). form of memory that supports stimulus–response asso-
Another major pathway involves the amygdala as a ciations, and an amygdala-dependent form of memory
nodal stage in the association of exteroceptive sen- that supports stimulus–reinforcer associations.
sory inputs to emotional outputs effected via the Succeeding chapters consider these and other memory
hypothalamic–pituitary axis and autonomic nervous systems and brain areas that influence memory
system, as well as emotional influences over wide- separately.
spread brain areas (See Chapter 3.24). The putative
involvement of this pathway in such processing func-
tions has led many to consider this system specialized
for emotional memory. This anatomical scheme can 3.01.4 The Declarative Memory
be a useful framework for understanding how the System
brain mediates the different memory systems dis-
cussed in succeeding chapters. Several chapters are devoted to different aspects of
the cortical–hippocampal system that supports
declarative memory. We begin with Burwell and
Agster’s overview of the anatomy of this system (See
3.01.3 The Organization of This Chapter 3.03). Their survey indicates that the circui-
Volume try of this system is largely conserved across
mammalian species, with the most prominent species
The chapters in this volume are arranged to reflect the differences derived from variations in cortical areas.
organization of the major memory systems of the brain Squire and Shrager then review the literature on
outlined above. We begin with an overview of mem- amnesia associated with damage to this system in
ory systems by White (See Chapter 3.02), a pioneer in humans (See Chapter 3.04). This review reinforces
the anatomical dissociations of the forms of memory and extends the original findings on the patient
first characterized by Maine de Biran. According to H.M., showing that damage to the hippocampus,
White, these systems are best distinguished as a hip- and more so following additional damage to the
pocampal-dependent form of memory that supports surrounding cortical areas, results in a selective def-
stimulus–stimulus associations, a striatum-dependent icit in the ability to remember everyday facts and

The weight of the evidence ticity and function is essential to understanding indicates that semantic knowledge lies in a distrib. recollection. as well as associations past. Nyberg (See Chapter 3.6 Introduction and Overview events. all of which contribute to memory in a broad on the nature of semantic organization and consider variety of ways. Voss and (See Chapter 3. these studies are immen- grade component that involves the loss of memories sely valuable because they offer the ability to for information acquired for some time prior to the distinguish the functional contributions of specific damage. the the evidence on whether semantic information is cerebral cortex is both the source and recipient of stored as distinct categories of knowledge or as a information within virtually all the brain’s memory distributed network of perceptual and motor attri. the units of information processing that encode essential role in the consolidation of memories from a and retrieve memories.05) review the literature from the development of animal models of amnesia and the approach of event-related potentials (ERPs) that the eventual successes in models of recognition sheds light on distinctions among different phases of memory. a part of the cortex sur- acterizes episodic memory as composed of several rounding the hippocampus that is essential to distinct processes that involve conscious control of recognition memory. They also describe the critical role of conceptual priming). The combination of anterograde and tem. Indeed. perceptual and hippocampus. called evidence on the nature of memory representations semantic memory. a consideration of cortical plas- butes of concepts. declarative memory. Neuronal networks in the peri- information and associative processes that encode rhinal cortex encode individual stimuli that we have and retrieve unique experiences as our personal previously experienced. The impairment involves both an antero. as well as in the temporal lobe. distinct experiences. humans. which have identified the distinct roles of memory (encoding and retrieval) and different types the hippocampus and cortical areas surrounding the of memory (familiarity. the ability to associate different brain network activities. relational memory. Weinberger argues that we have learned that . as Figure 1 illustrates. episodic memory is supported among items that support our ability for recognition by large network of areas in the cerebral cortex. particularly in the frontal and parietal areas. This capacity relies on the same Memory system as other capacities of episodic memory and emphasizes the involvement of the cortical areas just The cerebral cortex receives and processes the most outside the hippocampus.10) review the for storing and retrieving factual knowledge. in the perirhinal cortex. one for remembering everyday personal to remember where important events occurred. and distinguish combinations of stimuli that compose Declarative memory is composed of two subsys. At the same time. experiences. systems. Martin and Simmons (See complex aspects of perceptual and motor information Chapter 3. Studies on animals have the grade component that involves loss of the ability to inherent challenge of identifying the psychological remember information acquired following medial processes that correspond to declarative memory in temporal damage and a temporally limited retro. brain structures and the ability to examine the nature porally graded retrograde amnesia has led to the of memory representation at the level of single neu- general view that the medial temporal lobe plays an rons. Correspondingly.11) begins this uted fashion in which particular categories of survey with an overview of studies on the plasticity knowledge are represented by distinct organizations in sensory cortical areas. Stern and Hasselmo (See Chapter 3. These studies provide strong the hippocampus in two aspects of episodic memory evidence of distinct memory processes generated by in humans.07) then review our understanding of the and supports critical attentional and cognitive func- brain’s organization of semantic memory. a relatively accessible model system in which one can The development of animal models of amnesia and study how the cortex reorganizes well-controlled studies on the physiology of memory in animals have perceptual representations as they become signifi- also contributed significantly to our understanding of cant. called episodic memory. Therefore. memory. They focus tions.09) describe the initial difficulties in Paller (See Chapter 3. and the other Brown and Eldrige (See Chapter 3. Alvarado and Bachevalier labile state to a stable permanent form. memory.5 The Cerebral Cortex and experienced stimuli. While these studies began as within functionally specific areas of the cortex. Weinberger (See Chapter 3.06) char.08) focus on a fundamental function of episodic memory. our ability to recognize previously 3. the ability tems.01. and spatial memory.

19). These characteristics of emo- Procedural learning is generally conceived as the tional memory are discussed by Maren (See Chapter process by which we learn how to do a behavior 3. Emotion and memory intersect in two ways. Nudo and Sanes (See Chapter 3. along with pathways through the frontal to short-term memory processing is by supporting an cortex and other brain areas involved in the execu- implicit form of memory called priming. Packard (See Chapter 3. one way in which the cortex contributes learning.6 Procedural Learning and independent of explicit memory for the circumstances Memory in which the emotional significance of a particular stimulus was acquired. a model for understanding how that system adapted for tual and cognitive demands. who focuses on studies in The cerebral cortex is also critical to short-term animals.01. including classical condition- cessing. in the speed or fluency of perceptual processing as a Particular forms of procedural learning are studied result of prior experience with particular stimuli. studies in animal model paradigms. These studies acquired bias or increased fluency of actions. procedural memories. within different brain systems.12) tells us. very highlight the central role of the amygdala as a nodal much the way Maine de Biran envisioned mechanical area where neural responses to emotional stimuli are memory and William James described habit learning. an increase tion of behavioral actions. Ranganath and Blumenfield (See Chapter 3. Buchsbaum and D’Esposito by Cullen (See Chapter 3. McGaugh and Roozendaal (See terizations of procedural learning and memory are Chapter 3. Both lines of study reveal the basal ganglia as memory processes. and Roberts (See Chapter 3.18). supports associative and cognitive functions as well who focus on this memory system in humans. Introduction and Overview 7 plasticity at even early stages of cortical processing reviewed by Knowlton and Moody (See Chapter 3. depending on the nature of current percep. works within the prefrontal region. acquired and a main nucleus that generates the expres- The best known examples of procedural memory sion of learned emotional responses.7 Emotional Memory and review the extensive literature showing that the func. and as perceptual memories. role of the motor cortex in habit learning. like procedural memory. Finally. memory functions of the cortex. Prather. Also.23) pro- in short-term memory. Notably. and is also the place where memories are held in mind acquisition of the vestibulo-ocular reflex as reviewed during working memory. The roles of sensory processing areas versus those of This system shares with mammalian procedural learn- prefrontal areas involved in cognitive processes are ing systems a critical role of basal ganglia and provides varied. as reviewed by Poulos.20). and Thompson (See Chapter 3.14) 3. including declarative and complex cognitive procedures as well.01.22) survey the literature on the basis of working memory. As Grill-Spector (See Chapter a critical brain structure involved in procedural 3. forms of memory emerge from many types of processes and interactions among cortical areas. the acquisition of song in birds. or indefinitely during rehear. particularly with Procedural memories are therefore expressed by an regard to forms of fear conditioning. The other major way that emotion and memory These are often embodied in choice behavior intersect is that emotionally arousing events modulate or complex motor sequences but can involve quite the strength of memories. involves a specialized memory system in which other- and that these roles may depend on different net. One way al processes in long-term episodic memory as well.21 and 3. such that short-term a specialized form of procedural learning. revealing that a large net. involve acquired habitual responses and skills. work of cortical areas maintains information briefly Mooney. procedural learning.13) review the literature on the neural (See Chapters 3. in motor reflex learning.26) review the literature on the modulation . wise neutral stimuli acquire emotional significance. The cerebral cortex Christian. focusing on behavioral and neurophysiological independent of explicit memory for what we learned. Prominent examples The study of priming offers insights into how include studies that focus on the role of the cerebellum perceptual memory is intertwined with sensory pro.17). vide a synthesis of findings on an example of expert sal and cognitive manipulation in working memory. revealing the intimacy of sensory and ing of the eyelid blink response. the acqui- sition and expression of emotional memories can occur 3. very much the way Maine de Biran described sensitive memory.24). These charac. Modulation of Memory tions of the prefrontal cortex are not limited to working memory but extend to encoding and retriev.

). specifically the projections of the me. Neurosciences 8: 3–12. Baltimore: Williams and Wilkins. Finally. Olton DS (1984) Comparative analyses of episodic memory. Behav. they rather must now also be can have different effects depending on the age of the coming from a thoughtful analysis of the ways experi- subject and duration of hormone exposure in humans ences modify our cognition and behavior. 2: 313–365.16) review the literature on noradrenergic and cho. Brain Behav. neurochemicals. and that projections from largely as the rediscovery of the notion of multiple the amygdala influence the strength of memory con. In: Ross WD (ed. The existing examples include hormones and neurotransmitters. ory. arguing that the effect of I find it just amazing that recent advances in the emotional arousal is mediated via hormonal signals cognitive neuroscience of memory can be conceived that influence the amygdala. tures in memory performance. Gold (See Chapter 3. information from memory: A theory. the brain. to truly understand the distinct opera- brain areas. (1918 well-identified pathology and in Alzheimer’s disease. and parahioppocampal cortices: Organization of ison to mechanisms of normal. modulate memory. Becker JT. 3. and that both technologies of our time.27) reviews progress in the discovery of chemicals that can enhance memory.8 Aging and Memory Conscious Recollection: Memory Systems of the Brain.25) show that other hormone legitimacy of this rediscovery: It is unlikely that the systems. 7: 250–251. perirhinal. edn. and linergic ascending systems from the basal forebrain to aging. Neurol. Olton DS. including gonadal steroids as well as adrenal current taxonomies of memory are just driven by the steroid hormones. Brickman and Buchsbaum (See Chapter 3.8 Introduction and Overview of memory by arousal. Eichenbaum H and Cohen NJ (2001) From Conditioning to 3. . Map. Progress so far should be dial septum and diagonal band to hippocampal regions viewed as only an introduction to our understanding and from the nucleus basalis/substantia innominata to of the functional circuitries of the memory systems of the cerebral cortex. striatum. Scoville WB and Milner B (1957) Loss of recent memory after the pathology of Alzheimer’s disease and provide an bilateral hippocampal lesions. New York: Oxford University Press. Neurosurg. this prominent memory disorder of aging in compar. age-associated cortical inputs and interconnections with amygdala and memory loss. Vol. have a long way to go. Juraska and by the early thinkers. Sarter and Demeter (See Chapter the various ways in which their processing is modu- 3. the current findings from neuroscience have served to An additional major modulation of memory proces. and memory. as References well as other signaling molecules that are involved in cellular plasticity mechanisms that influence memory Aristotle (350BC/1931) De memoria et reminiscentia. update on the theories and mechanisms that underlie Psychiatry 20: 11–12. 449b–453b. lated by other brain systems.) The Works of Aristotle. In addition. egies that underlie the characterizations of memory space. we still basal forebrain. consolidation or the arousal-mediated modulation of pp. Our aim that are critical to detection. We should take heart in the Rubinow (See Chapter 3. 12: 421–444.29) address Brain Behav. Oxford: Oxford University Press.15) minary understanding into the details of the focus specifically on the role of basal forebrain struc- functional organization of these memory systems. Sci. There have been quite significant strides ence distinct aspects of hippocampal and cortical made in identifying the major memory systems and memory processing. showing O’Keefe J and Nadel L (1978) The Hippocampus as a Cognitive that aging is associated with a shift in cognitive strat. and outcome as systems neuroscientists is to flesh out that preli- processing of cues. New York: Holt. memory systems and the specific systems envisioned solidation in other memory systems. J. Chiba and Quinn (See Chapter 3. Yet. Maine de Biran (1804/1929) The Influence of Habit on the Daselaar and Cabeza (See Chapter 3. memory. One of these influences arises from the tions that characterize each memory system.). Cohen NJ and Squire LR (1980) Preserved learning and retention of a pattern-analyzing skill in amnesia: Dissociation of knowing how and knowing that. Biol. Beare JI (trans. Suzuki WA (1996) Neuroanatomy of the monkey entorhinal. validate the early conceptions of how memory is sing is generated by the ascending influences of lower organized. Hirsh R (1974) The hippocampus and contextual retrieval of Aging has a particularly devastating effect on mem. and Handlemann GE (1979) Hippocampus. Oxford: Clarendon Press. Science 210: 207–210. and animals. Yet our understanding of these systems and the cortex that interact to gate attentional resources modulatory influences is still in its infancy. These projections interact to influ. abilities lost and preserved in old individuals.28) review the Faculty of Thinking.01. literature on aging and memory in humans. selection. Sci. both in its benign form of memory loss without James W (1890/1918) The Principles of Psychology.

3.02 Multiple Memory Systems in the Brain: Cooperation
and Competition
N. M. White, McGill University, Montreal, Canada
ª 2008 Elsevier Ltd. All rights reserved.

3.02.1 Introduction 10
3.02.2 Inferring Information Type from Learned Behavior 11
3.02.2.1 The Rigorous Study of Learning 11
3.02.2.2 Theories of Learning 11
3.02.2.2.1 Stimulus-response (S-R) associations 11
3.02.2.2.2 Stimulus-stimulus (S-S) associations 11
3.02.2.2.3 Stimulus-reinforcer (S-Rf) associations 12
3.02.2.3 Reinforcers 12
3.02.2.4 Information Types: Relationships among Elements 13
3.02.3 Localization of Information Processing 13
3.02.3.1 Early Localization Attempts 13
3.02.3.2 HM and the Function of the Hippocampus 13
3.02.3.2.1 Attempts to replicate HM’s syndrome with animals 14
3.02.3.3 Contextual Retrieval 14
3.02.3.4 Spatial Learning 15
3.02.3.5 Memory and Habit 16
3.02.3.6 Declarative versus Procedural Memory 16
3.02.3.7 Double Dissociation of S-S and S-R Learning in Humans 17
3.02.3.8 Dissociation of Three Information Types in Rats 17
3.02.3.8.1 Win-shift task – hippocampus-based S-S memory 17
3.02.3.8.2 Win-stay task – caudate-based S-R memory 18
3.02.3.8.3 Conditioned cue preference task – amygdala-based S-Rf memory 19
3.02.3.8.4 Dissociation by damaging brain structures 19
3.02.3.8.5 Dissociation by reinforcer devaluation 19
3.02.4 Information Processing Systems 21
3.02.4.1 Systems Concept 21
3.02.4.1.1 Systems process incompatible information 22
3.02.4.1.2 Systems are internally specialized 22
3.02.4.1.3 Coherence: Some representations are better than others 23
3.02.4.1.4 The learning-rate parameter 23
3.02.4.1.5 Cooperation and competition among systems 23
3.02.4.2 Information Processing and Memory 23
3.02.4.3 Dissociations of Memory Systems 24
3.02.5 S-S versus S-R Information Processing 24
3.02.5.1 Studies with Rats 24
3.02.5.1.1 Competition on the radial maze 24
3.02.5.1.2 Cross maze 24
3.02.5.1.3 Water maze 27
3.02.5.1.4 Medial versus lateral caudate nucleus 30
3.02.5.2 Studies with Humans 30
3.02.5.2.1 Spatial learning 30
3.02.5.2.2 Probabilistic classification 31
3.02.5.3 Summary: Competition and Coherence 32

9

10 Multiple Memory Systems in the Brain: Cooperation and Competition

3.02.6 S-S versus S-Rf Information Processing 32
3.02.6.1 Studies with Rats 33
3.02.6.1.1 CCP with spatial cues 33
3.02.6.1.2 Cooperation and competition in adjacent arms CCP learning 34
3.02.6.1.3 Path integration versus visual cue conditioning 34
3.02.6.1.4 Fear conditioning 35
3.02.6.1.5 Skeletal conditioning 37
3.02.6.2 Experiments with Humans 37
3.02.6.2.1 Conditioned preference 37
3.02.6.2.2 Conditioned fear 38
3.02.6.2.3 Skeletal responses 38
3.02.6.3 Summary 38
3.02.7 S-Rf versus S-R Information Processing 39
3.02.7.1 Win-Stay and CCP Learning 39
3.02.8 Summary and Some Outstanding Issues 39
3.02.8.1 Some Outstanding Issues 41
References 42

3.02.1 Introduction conscious memories and temporal information but
retained normal memory for habits and skills such
Both casual and scientific observation inform us that as handicrafts (Ribot, 1883). He suggested that per-
behavior changes with experience. This phenomenon sonal and conscious memories are stored in the
is called learning. Its existence implies that informa- cortex, and that memories for skills and habits must
tion about experience is retained or stored in some be stored elsewhere in the brain. Together, these two
way so it can influence future behavior. This is called early French authors described a basic version of the
memory. Learning can be observed. Memory cannot multiple memory systems idea: that different types of
be observed but is inferred from learned behavior. information are stored in different parts of the brain.
Such inferences are quite common in daily life. One Other early evidence for localization of memory
of their best-known applications is the inference that types came from observations of pathological states
students have acquired certain specific memories such as the amnesic syndromes named for Alzheimer
from their performance on an examination. (1987) and Korsakoff (Ljungberg, 1992), who sug-
Identification and description of the information gested that damage to the cortex (including the
stored as memory is a major subject of this chapter. hippocampus) and medial thalamus were involved
Perhaps the earliest recorded example of such a in storing the memories their patients had lost.
description was associationism, the notion that neural This early evidence for the localization of mem-
representations of certain ideas and events are con- ories for different kinds of information contributed to a
nected in some way. This inference was made from controversial science called phrenology, the idea that
the observation that temporally contiguous ideas and the human personality could be studied by observing
events tend to evoke memories of each other, the shape of the head (Gall and Spurzheim, 1819). The
expressed as thoughts and behavior. basis of this claim was the hypothesis that individual
As early as 1804, Maine de Biran described how differences in the development (i.e., size) of brain areas
language reveals memory for several different kinds with different functions determines both personality
of information, including motor functions (speech), and the shape of the skull that encloses them.
emotions and the situations that evoke them, and Although the idea that skull shape is related to brain
abstract associations among ideas and concepts function was discredited, the concept of localization of
(Maine de Biran, 1804). In his 1883 book, Les function in the brain has become a basic principle of
Maladies de la Me´moire (Diseases of Memory), Ribot neuroscience. The multiple memory systems idea
described a series of patients with apparent damage applies this principle to the processing and storage of
to the cerebral cortex who had lost all personal, different types of information.

Multiple Memory Systems in the Brain: Cooperation and Competition 11

3.02.2 Inferring Information Type Information types
from Learned Behavior Stimulus-Stimulus Stimulus-Response Stimulus-Reinforcer
S-S S-R S-Rf
3.02.2.1 The Rigorous Study of Learning S S S R S Rfa
Rfm
The scientific study of memory requires a rigorous
S Rfm
methodology for observing and recording learned be- cRfa
cRfm
havior. The first such investigations are usually • cognitive learning • associative learning • Pavlovian conditioning
attributed to Ebbinghaus (1885). Ebbinghaus memor- • declarative memory • procedural memory • classical conditioning
• spatial learning • response learning • affective memory
ized lists of nonsense syllables (used because differences hippocampus caudate nucleus amygdala
in the familiarity of words would have influenced
learning and recall rates). He recorded the number of S Stimulus Rfa Reinforcer - affective value
trials required to reach specific levels of performance R Response Rfm Reinforcer - memory modulation
and observed how this number was reduced each
Association
time the same list was rememorized (savings). By
Figure 1 Information types. The headings of each column
this experiment and many others, he showed that
show the names of the three types of information, derived as
memory could be brought under experimental control explained in the text. Each type consists of associations in
and studied with precision, even though it could not which three elements (stimuli, responses, reinforcers) are
be directly observed. Other researchers extended associated in different ways. The diagrams show how the
Ebbinghaus’ original work (Anderson and Bower, elements are associated in each type of information. Below
each diagram are common synonyms for each type of
1979), and some began applying the principles he
information used in the learning and memory literature. At
developed to the study of learning and memory in the bottom of each column is the name of the brain structure
animals. central to the anatomical system thought to process each
type of information.

3.02.2.2 Theories of Learning
whose theory predicted the probability of a response
During the first part of the twentieth century the based on numerous factors, including the number of
behavioral investigation of learning in rats formed times the reinforced S-R pairing was experienced, the
the basis of several major research programs which number of hours of food deprivation, the amount of
differed, sometimes radically, in the inferences they reinforcer given, the properties of the reinforcer, how
made from behavior about the kinds of information often the response had been made recently, and many
stored in memory (Hilgard and Bower, 1966). The others. One purpose of this complex specification was
concepts that emerged from this research (Figure 1) to explain behavior without considering any unobser-
are useful starting points for describing the kinds of vable variables such as conscious knowledge of a
information stored as memory. situation or awareness of its emotional (or affective)
content, which were considered inadmissible by the
3.02.2.2.1 Stimulus-response (S-R) behaviorist approach to psychological investigation
associations (Watson, 1912; Bergmann and Spence, 1964).
The simplest theory, initially proposed by Thorndike
(1898, 1911), was based on the observation that the
probability of a stimulus evoking a response depends 3.02.2.2.2 Stimulus-stimulus (S-S)
on the number of times the response has been made in associations
the presence of the stimulus and followed by a rein- In direct contrast, Tolman (1932, 1948, 1949) argued
forcer (e.g., food). Thorndike postulated that all that behavior is based on cognitive information rather
learned behavior is the result of a series of associations than being controlled by specific response tenden-
or bonds between neural representations of stimuli cies. One example of an observation leading to this
and responses that have been strengthened, or inference was the partial reinforcement extinction
‘stamped in’ by a reinforcer. This makes the stimulus effect (Skinner, 1938; Humphreys, 1939). One group
more likely to elicit the response. of rats received reinforcement after every correct
The idea that behavior is based on stored stimulus- response they made in a learning task, another
response, or S-R, associations became the foundation group received reinforcement after only half of
of an elaborate system developed by Hull (1943), their correct responses. When tested with no

12 Multiple Memory Systems in the Brain: Cooperation and Competition

reinforcement (extinction), the group that had been Vegas, 2004). Reflecting this controversy, the process
reinforced for half of its responses kept responding is often described as a CS-US/UR learning. In the
longer than the group that had been reinforced for all multiple memory systems context, this form of learn-
responses. This result is the opposite of what was ing is called stimulus-reinforcer (S-Rf) learning
predicted by Hull’s S-R theory, which held that the (Figure 1), because it is restricted to responses that
strength of a response tendency is directly related to are elicited by reinforcers.
the number of times it has been reinforced. Tolman
concluded that the rats’ behavior was based on their
3.02.2.3 Reinforcers
knowledge of the situation and what events they
expected. The rats in the 100% group expected The events called USs in Pavlovian theory are the
reinforcement for every response, so a few unrein- same ones that Thorndike labeled reinforcers. The
forced responses were enough for them to detect the three theories include three different functions of the
change in conditions and stop responding. The rats in responses elicited by reinforcers:
the 50% group did not expect reinforcement for
1. Reinforcers elicit internal, unobservable
every response, so more unreinforced responses
responses that strengthen, or modulate, S-R and S-S
were required before they stopped.
(Packard and Teather, 1998; Packard and Cahill,
Tolman (1932) postulated that learning consists of
2001) and S-Rf (White and Carr, 1985; Holahan
acquiring information about the relationships among
et al., 2006) associations when their occurrence is
stimuli and events. This information constitutes
temporally contiguous with the acquisition of the
‘knowledge’ and is represented as a series of inter-
association (Thorndike, 1933; Landauer, 1969;
locking relationships among stimuli known as
McGaugh and Herz, 1972; White, 1989b; White and
stimulus-stimulus (S-S) associations (Figure 1). S-S
Milner, 1992) (shown as Rfm in Figure 1).
associations can represent spatial relationships, lead-
2. Reinforcers elicit internal, unobservable
ing to the concept of the spatial map (see section
responses that are perceived as positive or negative
3.02.3.4), and temporal relationships, leading to
affect (Young and Christensen, 1962; Cabanac, 1971;
expectancy (knowing what comes next in a sequence
White, 1989b; White and Milner, 1992; Burgdorf and
of events).
Panksepp, 2006) (shown as Rfa in Figure 1). Humans
and animals can learn about these affective states and
3.02.2.2.3 Stimulus-reinforcer (S-Rf)
how to obtain or avoid them in, a process called
associations
instrumental learning.
The third type of information was described by
3. Reinforcers elicit observable orienting and
Pavlov (1927). Pavlov was studying the internal
approach or withdrawal responses. Normally, stimuli
secretions produced by food in hungry dogs. When
that elicit approach also elicit positive affect and are
the dogs were repeatedly exposed to the experimen-
sometimes called rewards. Stimuli that elicit with-
tal situation, the secretions began to occur in the
drawal and also elicit negative affect are described as
absence of food. Accordingly, Pavlov postulated that
aversive, or as punishments. Both rewarding and
an association was formed between the food (the
aversive events strengthen or modulate memory
unconditioned stimulus, or US) and stimuli in the
(Huston et al., 1977; White, 1989b; Holahan et al.,
experimental situation (the conditioned stimulus, or
2006).
CS). This made the CS capable of eliciting condi-
tioned responses (CR) similar to those produced by Because the learning theorists used rats in their
the US. experiments, they also had to use biologically rele-
USs are events that elicit responses without pre- vant reinforcers to control behavior and the
vious experience. These include food, water, sexual information being processed and stored during the
partners, and certain aversive events. The elicited experimental trials. Although it is not necessary to
unconditioned responses (URs) include approach or use such reinforcers with humans, who can follow
withdrawal and responses of the autonomic nervous instructions, feedback about correct and incorrect
system and certain brain structures and neurotrans- responses is also used and has many of the same
mitters. Neutral stimuli acquire CS properties when functions as biological reinforcers in rats (see
a US occurs in their presence. The issue of whether Thorndike, 1933). Both instructions and feedback
Pavlovian learning is the result of a CS-US or a CS- control human behavior and information processing
UR association is controversial (e.g., Donahoe and in experimental trials.

Multiple Memory Systems in the Brain: Cooperation and Competition 13

3.02.2.4 Information Types: Relationships Notwithstanding their methods, Franz’s and
among Elements Lashley’s experiments failed to localize memory
functions because they assumed that a single brain
Each learning theory made different inferences from
structure (the cerebral cortex) processed and stored
observed behavior about the kinds of information
all memories. It followed from this assumption that
processed and stored during learning, but, as illus-
damaging the critical structure should eliminate all
trated in Figure 1, in each case the information
forms of memory. Although the idea that different
consists of different combinations of the same three
parts of the brain process different kinds of memory
elements: stimuli, responses, and reinforcers. The had been suggested, Franz and Lashley did not make
information types differ in the relationships among use of this idea, possibly because of behaviorist stric-
these elements. S-S, S-R, and S-Rf associations can tures on the use of such unobservable entities as
all be thought of as different types of information information types.
created by experience, stored in the brain, and
recalled to influence behavior. Large parts of this
chapter describe evidence that these types of infor-
mation are processed and stored in different parts of 3.02.3.2 HM and the Function of the
the brain. Hippocampus
In the early 1950s new clinical evidence for the
localization of different types of memory in the
brain emerged. A patient with intractable epilepsy,
3.02.3 Localization of Information known by his initials (HM), underwent a bilateral
Processing excision of a major portion of his temporal lobes
3.02.3.1 Early Localization Attempts (including much of the hippocampus) as a last-
chance treatment, making him perhaps the most
S. I. Franz (1902) was among the first to apply sys- famous patient in modern neurology (Scoville and
tematic behavioral methodology to the study of brain Milner, 1957). He was given a full psychological assess-
areas involved in memory. He reported that cats ment by Brenda Milner, who reported (Milner and
could learn a series of complex responses to escape Penfield, 1955; Milner, 1958, 1959) that he was unable
from a box and that large lesions of various parts to recall personal experiences or other events from the
of the cortex had only temporary effects on this previous 20 or so years (retrograde amnesia), and that
learned behavior. He concluded that there was no he was also unable to remember any current experience
evidence for localization of memory functions in the for more than a few minutes (anterograde amnesia).
brain. HM also had difficulty finding his way around, even
Franz’s student, Karl Lashley, pursued these stud- in familiar environments.
ies, testing rats on a variety of memory tasks with However, HM retained other forms of memory.
similar results (Beach et al., 1960). No part of the He was able to learn a simple maze that required him
cortex seemed more important than any other – to move a stylus through a series of left and right
temporary deficits in learned behavior were usually turns on a board (Corkin, 1968; Milner et al., 1968).
quickly reversed with additional training. Only very He performed this task accurately, while at the same
large lesions including most of the cortex produced time claiming he had never seen it before. This
permanent, although generally still partial, deficits. showed that the memory deficit produced by tem-
Lashley concluded that the memory functions of any poral lobe resection was limited to information about
part of the cortex could substitute for any other part, HM’s previous experiences. The deficit did not
a principle he called equipotentiality. Lashley’s frus- include the information that allowed him to learn
tration at his inability to localize and understand the maze, which must have been dependent on
memory is revealed by his summary statement in a some other part of the brain.
1950 article describing a lifetime of work on the Although Milner and her coauthors carefully
problem: described their observation that the effects of bilat-
eral hippocampal ablation are limited specifically to
I sometimes feel, in reviewing the evidence on the memory for new ongoing experience and spatial
memory trace, that the necessary conclusion is that orientation, and that other kinds of memory, includ-
learning is just not possible. (Lashley, 1950, p. 477) ing verbal abilities, emotion, and skills and habits

14 Multiple Memory Systems in the Brain: Cooperation and Competition

remained normal, these distinctions were largely 3.02.3.3 Contextual Retrieval
ignored at first.
In what is now considered a major conceptual break-
through, Hirsh (1974) provided an explanation for the
inconsistent effects of hippocampal lesions on memory
3.02.3.2.1 Attempts to replicate HM’s tasks in rats. Hirsh pointed out that hippocampal
syndrome with animals lesions did not affect learning tasks that could be
3.02.3.2.1.(i) Monkeys Orbach et al. (1960) performed on the basis of a single S-R association,
proposed that a strong test of the temporal lobe but they impaired performance of tasks that could
memory hypothesis would be met only if temporal not be correctly performed using this kind of informa-
lobe damage impaired performance regardless of tion. He suggested that tasks impaired by hippocampal
what behavioral task their monkeys were required lesions involved acquisition of two or more S-R asso-
to learn. These workers did find deficits on some ciations, and that the hippocampus mediated a process
memory tasks, but only with very large temporal by which one of these was selected by the context. An
lobe lesions. example of a behavioral observation that led to the
The reasons for this (and other) failures to repli- inference of this contextual retrieval process is illus-
cate HM’s syndrome were revealed by Mahut and trated in Figure 2. Rats were trained in a T-maze to
coworkers, who reported that monkeys with hippo- turn left for food when hungry and right for water
campal (Mahut, 1971), fimbria-fornix, or entorhinal when thirsty (Hsiao and Isaacson, 1971). The choice
cortex (Mahut, 1972; Zola and Mahut, 1973) lesions point in the maze became a stimulus associated with
were impaired on spatial tasks when no local stimuli two responses: the left and right turns. The deprivation
were available to guide their behavior. In the same state was the context that selected the appropriate S-R
monkeys, nonspatial tasks were unaffected, and when association. Normal rats learned to make the correct
small cue objects were added to spatial discrimina- turn depending on their deprivation state. Rats with
tion tasks, performance actually improved relative to various impairments of the hippocampal system
that of normal controls (Killiany et al., 2005). These (Hirsh and Segal, 1971; Hsiao and Isaacson, 1971;
experiments showed that memory deficits are pro- Hirsh et al., 1978a,b) made one turn or the other,
duced by hippocampal damage in monkeys, provided regardless of their deprivation state. They were unable
tasks requiring the use of the kind of information to use the motivational context to select the correct
processed in the hippocampus are used (Mahut response.
et al., 1981). Another example is reversal learning. As shown in
Figure 3, normal rats and rats with hippocampal
lesions learned to turn left for food at similar rates.
When the food was switched to the right arm of the
3.02.3.2.1.(ii) Rats A number of workers also maze, normal rats adjusted their behavior much more
tried to demonstrate the effects of hippocampal quickly than rats with hippocampus lesions. The
lesions on memory in rats. These studies had such normal rats were able to change the direction of
limited success that several theories suggesting alter- their turn on the basis of new information about the
nate functions for the rat hippocampus were location of the food obtained on the first few trials
proposed (Kimble and Kimble, 1965; Douglas and after the switch. According to Hirsh, the switch
Pribram, 1966; Kimble, 1968; Isaacson and Kimble, altered the context that selected the appropriate
1972) and refuted (Nadel et al., 1975). These hypoth- response. Hippocampectomized rats were unable to
eses did not completely reject memory-related use this type of information. They had to extinguish
functions but were primarily concerned with the old S-R association and acquire a new one when
explaining the highly inconsistent effects of hippo- the reinforcer was switched.
campal lesions on the performance of various Hirsh’s main contribution to the idea of localized
memory tasks. The inconsistency was the result of a memory functions based on the content of the mem-
failure to appreciate the specific type of information ory was to explain it in terms of the existing animal
processed in the hippocampus and the consequent learning literature, an area of research that was very
use of tasks that could be performed on the basis of well developed and had up to that time generally
some other kind of information processed in other rejected or ignored this possibility. Although it took
parts of the brain. some time, ‘‘The hippocampus and contextual

Multiple Memory Systems in the Brain: Cooperation and Competition 15

Motivational control of maze behavior
(contextual retrieval)

Food
Water

Thirsty Hungry

Normal rat Hippocampectomized rat
100 100

80 80
Percent correct

Percent correct
60 60

40 40
Food
Water
20 20

0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 9 10 11 12
Blocks of trials Blocks of trials

Figure 2 Motivational control of maze behavior (contextual retrieval). The figure illustrates an experiment in which rats
were trained on a T-maze with food in one arm and water in the other. The rats were food or water deprived on alternate days.
The graph on the left shows the behavior of a normal rat that learned to turn right for water on days when it was thirsty first (trial
blocks 1–5). Between blocks 5 and 8 the rat also learned to turn left for food when hungry, while maintaining the correct
response for water. This shows that the rat had learned to use its motivational state to perform the correct response at the
choice point in the maze. The graph on the right shows the behavior of a hippocampectomized rat that began by learning to
turn right for water. As the rat’s performance on the days when it was hungry improved, its performance on thirsty days
declined. After a few more trial blocks, the rat’s performance when thirsty improved, whereas its performance when hungry
declined. Several more such reversals were observed, suggesting that this rat was able to learn one of two simple S-R
associations but was unable to use its motivational state to select between the two. Adapted from Hirsh R (1980) The
hippocampus, conditional operations and cognition. Physiol. Psychol. 8: 175–182.

retrieval of information from memory: A theory’’ the spatial map, a neural representation of the rela-
(Hirsh, 1974) has come to be regarded as a major tionships among constant features of an environment.
turning point in research on memory. The information in these representations can be
described as consisting of a series of S-S associations.
This relational information is purely descriptive, with
3.02.3.4 Spatial Learning no specific implications for behavior. O’Keefe and
In The Hippocampus as a Cognitive Map, O’Keefe and Nadel distinguished hippocampus-based spatial
Nadel (1978) presented evidence that the hippocam- learning (also called allocentric) from taxon (or ego-
pus is the primary structure for representing and centric) learning, in which each environmental
possibly storing spatial information. The two main feature can evoke one specific response, as in the
lines of evidence for this idea were impairments in case of S-R learning. O’Keefe and Nadel did not
spatial learning produced by hippocampal lesions and speculate about the neural substrate for taxon
the observation that the activity levels of certain hip- learning.
pocampal neurons increased whenever a rat was in a The idea that the hippocampus processes informa-
specific spatial location (O’Keefe and Dostrovsky, tion consisting of relationships among features of an
1971; O’Keefe, 1976). This suggestion that the hippo- environment has been extended and incorporated into
campus contains information about the animal’s the theory that the structure processes relational infor-
position in space led O’Keefe and Nadel to postulate mation of all kinds (Hirsh, 1980; Cohen and

16 Multiple Memory Systems in the Brain: Cooperation and Competition

Left-right reversal learning

Initial learning Reversal 1 Reversal 2 Reversal 3
(reversal 0)

Left-right reversal learning
50

Trials to criterion 40

30
Control
Hippocampus
20

10

0
0 1 2+
Reversal

Figure 3 Effects of hippocampus lesions on reversal learning. Hungry rats learned to find food in a maze by turning left at
the choice point. Both normal rats and rats with hippocampal lesions learned the initial response to criterion (five consecutive
correct responses) equally well (Reversal 0). However, when the correct response was reversed, the hippocampectomized
rats were severely impaired. The impairment continued on subsequent reversals. All rats were given a total of 100 trials, during
which the control rats met the criterion and were reversed an average of 8.1 times. Hippocampectomized rats met the
criterion and were reversed an average of 2.2 times. Adapted from Kimble DP and Kimble RJ (1965) Hippocampectomy and
response preservation in the rat. J. Comp. Physiol. Psychol. 60: 474–476.

Eichenbaum, 1991; Eichenbaum and Cohen, 2001). pairs were presented repeatedly, normal and hippo-
This information, stored as S-S representations, campal monkeys learned to respond to the correct
can be used flexibly to produce behaviors appropriate member of each pair equally well. Thus, hippocam-
to different circumstances. The theory that the infor- pal lesions impaired memory for novel objects seen
mation processed in the hippocampus is purely spatial, for the first time but did not affect learning to make a
and the idea that it represents relationships of a consistent response to each pair of objects after
more general nature including spatial have been repeated reinforced trials. Mishkin et al. described
debated (Wiener et al., 1989; Cohen and Eichenbaum, the latter behavior as a habit and speculated that
1991; Nadel, 1991; Eichenbaum et al., 1992). the memory for this type of behavior might be
mediated in the caudate nucleus.

3.02.3.5 Memory and Habit
3.02.3.6 Declarative versus Procedural
Mishkin et al. (1984) described an experiment in
Memory
which monkeys were shown pairs of objects.
Displacing one of them revealed a morsel of food. Cohen and Squire (1980) trained human participants
When shown the same two objects a short time later, to read words in a mirror. Both normal participants
displacing the other object revealed food. The mon- and patients with amnesia resulting from medial
keys were given a large number of trials over several temporal lobe dysfunction learned the mirror reading
days, with different pairs of objects on every trial. skill; they improved with practice and improved
Normal monkeys learned this ‘delayed nonmatching more for words that were repeated in each session
to sample’ rule, but monkeys with hippocampal than for words that were new. However, when tested
lesions were unable to do so – a deficit that is con- shortly after a training trial, normal participants were
sistent with a general loss of memory for recent able to remember most of the words they had just
events. In contrast, when a large number of object read; the amnesic patients were severely impaired.

Multiple Memory Systems in the Brain: Cooperation and Competition 17

Cohen and Squire concluded that memory for the S-R information and dissociate it from S-S informa-
words (declarative memory) requires a functional tion processing. The studies point to the basal
hippocampus (See Chapter 3.03), but that the mirror ganglia, specifically the caudate nucleus, as the loca-
reading skill (procedural memory) must be dependent tion of S-R information processing. A number of
on some other part of the brain. experiments with rats are consistent with this con-
Declarative and procedural memory describe clusion about the caudate nucleus (Divac et al., 1967;
different kinds of remembered information, corre- Divac, 1968; Thompson et al., 1980; Chozick, 1983;
sponding to S-S and S-R learning as described by Mitchell and Hall, 1987, 1988; Cook and Kesner,
Tolman and Thorndike, respectively, as well as to 1988; Packard and Knowlton, 2002). Phillips and
Mishkin et al.’s distinction between memory and Carr (1987) specifically proposed that S-R learning
habit, to Hirsh’s descriptions of contextual memory is mediated in the basal ganglia.
versus S-R memory, and to O’Keefe and Nadel’s
concepts of the spatial map and taxon learning.
3.02.3.8 Dissociation of Three Information
These distinctions were all made by showing that
Types in Rats
hippocampal damage impairs performance on tasks
requiring S-S information but has no effect on tasks Packard et al. (1989) used a double dissociation to
that can be performed using S-R information. None compare the impairments in information processing
of these distinctions includes evidence concerning capacity produced by lesions to the fimbria-fornix (a
the neural substrate of S-R information processing. major input-output pathway of the hippocampus)
and the caudate nucleus. This was extended by
McDonald and White (1993) to include the amygdala
3.02.3.7 Double Dissociation of S-S and
in a triple dissociation of memory tasks. All three
S-R Learning in Humans
tasks were performed on an eight-arm radial maze
Butters and coworkers (Martone et al., 1984; Butters (Figure 4) consisting of a center platform with eight
et al., 1986; Heindel et al., 1989) found that patients arms radiating from it in a sunburst pattern about 1 m
with Korsakoff’s syndrome, who are amnesic as a from the floor. Each task required the use of a differ-
result of hemorrhagic lesions of medial thalamus ent type of information.
and mammillary bodies, learned the mirror reading
skill normally but were unable to recognize words 3.02.3.8.1 Win-shift task – hippocampus-
they had recently read. In contrast, patients with based S-S memory
Huntington’s disease, in which neurons in the basal The win-shift radial maze task (Olton and Samuelson,
ganglia (including the caudate nucleus) degenerate, 1976; Olton and Papas, 1979) was used to examine S-S
had the opposite pattern of disabilities: They were information processing. The maze was situated in a
unable to acquire the mirror reading skill but recog- room with various extra-maze stimuli (or cues) that
nized previously seen words normally. constituted a spatial environment. A small food pellet
This pattern of effects constitutes a double disso- was placed at the end of each arm. Hungry rats were
ciation, in which subjects with impairments of one of placed on the center platform and allowed to forage
two brain areas were compared on two memory tasks. for food. Pellets consumed were not replaced, so to
Impaired function of each brain area affected only obtain the eight available pellets most efficiently the
one of the two tasks, leaving performance on the rat had to enter each arm once only. Second or more
other task intact. This led to the conclusion that entries to arms were considered to be errors. Normal
each brain area was involved in processing the type rats attained an average of less than one error per trial
of information required for the impaired task, but not after five to seven daily trials.
the information required for the unimpaired task. As To enter each arm once only, a rat must remem-
described in the previous section, recall of recently ber which arms it has entered as the trial proceeds
seen words requires declarative, or S-S, information; (working memory). To do this, it must be able to
mirror reading is a skill that may require a complex of discriminate the arms from each other. There is
S-R associations. evidence that rats perform this discrimination by
Although there was no direct confirmation of the associating each arm with the environmental cues
brain damage in these studies, they are significant that are visible from it (O’Keefe and Nadel, 1978;
because they provide evidence for the neural sub- Suzuki et al., 1980). Since all the individual cues are
strate that processes information that may include visible from several of the arms, no individual cue can

18 Multiple Memory Systems in the Brain: Cooperation and Competition

Triple dissociation of memory systems
Win-shift Win-stay CCP
S-S S-R S-Rf
food light

200
30 10000

180

Log mean percent lesion effect
25
Mean total errors (10 d)

Mean total errors (12 d)

1000
20 160

15 100
140

10
10
120
5

0 100 1
C FF CN AM C FF CN AM C FF CN AM
Figure 4 The triple dissociation. Three different tasks on the eight-arm radial maze are illustrated at the top. The type of
information processing thought to be required to perform each task is indicated, corresponding to the information types
described in Figure 1. As explained in the text, correct performance of the win-shift task requires rats to obtain the food pellet
at the end of each arm without reentering arms. This requires spatial (or S-S) information processing. Performance of this task
was impaired by lesions of fimbria-fornix (FF), but not by caudate nucleus (CN) or amygdala (AM) lesions. The win-stay task
requires a simple association between a cue light and the arm entry response, reinforced by eating the food at the end of the
lit arms. A different set of four arms is lit and baited on each daily trial. This is an instance of S-R information processing.
Performance on this task was impaired by lesions of the caudate nucleus but not by lesions of fimbria-fornix or amygdala. The
improved performance on this task produced by fimbria-fornix lesions is explained in Figure 5. In the conditioned cue
preference (CCP) task, the rats acquire an S-Rf association between the light and the food reinforcer, causing the light to
acquire conditioned stimulus properties. This results in a preference for the food-paired arm even when no food is present on
the test trial. Performance on this task was impaired by lesions of the amygdala, but not by lesions of fimbria-fornix or caudate
nucleus. Adapted from McDonald RJ and White NM (1993) A triple dissociation of memory systems: Hippocampus, amygdala
and dorsal striatum. Behav. Neurosci. 107: 3–22.

identify any single arm. However, the arms can be 3.02.3.8.2 Win-stay task – caudate-based
identified by their unique locations within a spatial S-R memory
map of the environment. This requires relational or The win-stay task was used to study S-R information
S-S information. processing. The same maze was used, but the location
Good performance on this task requires continu- of the food was indicated by a small light mounted at
ally updated information about the status of each arm the entrance to each arm. When a hungry rat was
as a trial proceeds. At the start of a trial all arm- placed on the center platform at the start of a trial,
identifying cue arrays indicate the location of a food four of the arms were lit, and only those four arms
pellet, but once an arm has been entered and the contained a food pellet. The other arms were empty.
pellet consumed, the same array must indicate the Efficient performance required the rats to enter lit
absence of food. In this way information about the arms and avoid dark arms. The information required
spatial environment is used flexibly to guide behavior to produce this behavior is simple: When a rat
according to changing conditions. approaches a lit arm (S), enters it (R), and consumes

Multiple Memory Systems in the Brain: Cooperation and Competition 19

the reinforcer, the S-R association is strengthened, involved in processing only one of the three kinds
increasing the probability that the stimulus will elicit of information.
the response again. No spatial information or work- None of these attributions was original to the triple
ing memory is required for the win-stay task. dissociation experiments. It was already well known
that lesions of the hippocampus and related structures
3.02.3.8.3 Conditioned cue preference impair spatial learning (Hirsh and Segal, 1971; O’Keefe
task – amygdala-based S-Rf memory et al., 1975; Olton et al., 1978; Olton, 1978; Morris et al.,
The conditioned cue preference (CCP) task was used 1982), and that Pavlovian conditioning involving S-Rf
to study S-Rf information processing. The radial maze information processing is impaired by amygdala lesions
was surrounded by curtains. Hungry rats were con- (Weiskrantz, 1956; Bagshaw and Benzies, 1968; Jones
fined on an arm with a light and a supply of food and and Mishkin, 1972; Nachman and Ashe, 1974;
on alternate days were confined on a different, dark Peinado-Manzano, 1988; Everitt et al., 1989, 1991;
arm with no food. After several days of such trials the Hatfield and Gallagher, 1995; Davis, 1997; Fanselow
rats were placed on the center platform of the maze and Gale, 2003). There were also numerous reports of
and given a choice between the two arms, neither of learning impairments produced by caudate nucleus
which contained food. Normal rats spent more time on lesions, but the evidence that these involved S-R learn-
the food-paired than on the unpaired arm. ing was not clear (e.g., Gross et al., 1965; Divac et al.,
Since the rats were exposed to the food while 1967; Divac, 1968; Kirkby, 1969; Winocur and Mills,
confined on a maze arm they could not have acquired 1969; Mitchell and Hall, 1987, 1988). The contribution
any S-R associations or instrumental responses of the triple dissociation experiment was to define
leading to their preference for the food-paired arm. these various tasks in terms of the kinds of information
The curtains surrounding the maze and confinement required to perform them and to show that the proces-
in one arm at a time severely limited the rats’ sing of each type of information was dependent on a
ability to acquire a spatial map of the environment different part of the brain.
(Sutherland and Linggard, 1982; Sutherland and
Dyck, 1984; Sutherland, 1985; White and Ouellet,
1997). Therefore, the preference was probably not a 3.02.3.8.5 Dissociation by reinforcer
result of learned information about the spatial loca- devaluation
tion of food (i.e., S-S learning). The alternative is that Reinforcer devaluation is a procedure in which a
consuming the food (US) during the training trials rewarding reinforcer is paired with an aversive
elicited an internal rewarding response (UR). This event, reducing the positive affective value of the
caused the arm cues (light or dark) in the food-paired reinforcer (Young and Christensen, 1962; Dickinson
arm to acquire CS properties. On the test trial, the CS et al., 1983). The prototype is the conditioned
elicited a conditioned reward response, causing the taste aversion (Garcia and Koelling, 1966; Nachman
rat to remain in the food-paired arm longer than in and Ashe, 1974; Yamamoto and Fujimoto, 1991),
the no-food arm. This analysis explains the prefer- in which pairing consumption of a rewarding sub-
ence for the food-paired arm as the product of S-Rf stance such as a sugar solution with injections of
information processing. lithium chloride (LiCl) (which produces gastric ill-
ness) decreases or eliminates consumption of the
3.02.3.8.4 Dissociation by damaging brain solution.
structures Yin and Knowlton (2002) used the devaluation pro-
As shown in Figure 4, performance on each of the cedure in conjunction with the CCP task on the radial
three tasks was impaired by only one of the three maze. After CCP training the rats ate some of the same
lesions. (Performance of the win-stay task was actu- food pellets in their home cages, followed by LiCl or
ally improved by fimbria-fornix lesions. This saline (control) injections. When tested in their home
phenomenon is explained in Figure 5.) Since the cages the rats that received LiCl ate much less than the
three tasks had identical sensory and motor require- rats that received saline, demonstrating the reduced net
ments, and the same reinforcer was used for all three, reward value of the food. When given a preference test
the differences in the effects of the lesions were on the radial maze with no food present the rats that
attributed to differences in the kinds of information received saline preferred their food-paired arms, but
required to perform the tasks. The effects of the the rats that received LiCl exhibited an aversion to
lesions imply that each lesioned structure was their food-paired arms (see Figure 6). This finding

20 Multiple Memory Systems in the Brain: Cooperation and Competition

Fimbria-fornix lesions Improve
performance on the win-stay task

Effect of lesions on win-stay performance Trial 1 Trial 2
100
Caudate
Control
Percent correct

Fimbria-fornix
80

60

40
1 2 3 4 5
Blocks of 3 trials
light
current location of food (correct response)
former location of food (error)
Figure 5 Fimbria-fornix lesions improve performance on the win-stay task. The graph shows the effects of lesions on
performance of the win-stay task (see Figure 4). Compared with normal control rats, rats with caudate lesions performed
poorly, suggesting that this structure is involved in processing the S-R information required for performing the task. In
contrast, lesions of fimbria-fornix (part of the hippocampal system) improved win-stay performance compared with controls.
The maze diagrams on the right explain this improved performance. They show two consecutive trials on the win-stay task. In
both cases the arms containing food are indicated by lights at their entrances, but the food is located in different arms on the
two trials. Caudate-based processing of information comprising the consistently reinforced S-R association between the
lights and the arm entry response results in correct behavior on both trials. However, on Trial 1 the hippocampus system
acquires information about the fixed cues in the spatial environment and about the location of food in relation to those cues.
As shown in the figure, this information is incorrect on Trial 2, because a different set of four arms contains food. Therefore,
any tendency to enter arms that previously contained food promoted by the hippocampal system would result in errors.
Fimbria-fornix lesions eliminate this S-S information and the erroneous responses it promotes, resulting in improved
performance of the S-R task. Because the information processing capacities of the two systems cause them to promote
different behaviors in the same situation, the effect of fimbria-fornix lesions on win-stay performance reveals a competitive
interaction between the behavioral effects of the two kinds of information processing. Adapted from Packard MG, Hirsh R,
and White NM (1989) Differential effects of fornix and caudate nucleus lesions on two radial maze tasks: Evidence for multiple
memory systems. J. Neurosci. 9: 1465–1472.

suggests that the preference for the food-paired arm after devaluation (Figure 6). These rats did not com-
was the result of a memory that involved the affective pletely reject the food while running the maze. Most
property of the reinforcer. This memory was positive in of them ate the pellets in the first few arms entered and
the saline-injected rats, leading to a preference for the then stopped eating while continuing to perform cor-
food-paired arm. It was much less positive, even nega- rectly. This suggests that they did not spontaneously
tive, in the LiCl-injected rats, leading to avoidance of transfer the aversion from the home cage to the new
the food-paired arm. This is consistent with the idea situation but recalled it only after tasting the food in
that the CCP is based on information comprising an the new context. That this rejection of the food
S-Rf association. had little effect on the accuracy of their win-shift
In contrast to these findings, Sage and Knowlton performance is consistent with the hypothesis that
(2000) found that the same devaluation procedure information involving the affective value of the food
did not affect the performance of well-trained rats is not required for win-shift performance, which
on the win-stay task, even though they stopped eat- depends on information about the presence and
ing the food at the ends of the lit arms (Figure 6). absence of food in each arm as the trial progresses.
This observation is consistent with the idea that Although the devaluation data do not dissociate the
performance on this task is unrelated to the affective neural substrates of information processing in the three
value of the reinforcer, which acts simply to tasks, they constitute evidence that S-S and S-R infor-
strengthen the caudate-based S-R association that mation can control behavior independently of the
produces the win-stay behavior. affective value of reinforcers in the situation. This con-
Sage and Knowlton (2000) also found that rats trasts with the evidence for the representation of these
trained on the win-shift task performed normally affective properties in S-Rf information processing.

02.02. Behav. Neurosci. 114: 295–306. Before devaluation the rats ate all pellets available. this did not affect the accuracy of their win-stay performance. However. 116: 174–177.3). The pellets were not completely rejected.4 Information Processing dissociations among brain structures and processing Systems of information types. The effect of devaluation on the preference is evidence that the rats’ behavior is controlled by information that includes a representation of the affective value of the reinforcer.4. 3. Neurosci. This constitutes evidence that the affective value of the reinforcer is irrelevant to win-stay behavior and is consistent with the idea that the reinforcer acts to strengthen the S-R association that controls behavior in this situation. Rats that experienced reinforcer devaluation (explained in the text) exhibited an aversion to their food-paired arms. which can be produced by information about the presence and absence of food in the arms.02.8. Multiple Memory Systems in the Brain: Cooperation and Competition 21 Effects of reinforcer devaluation on three radial maze tasks Conditioned cue preference Win-stay 100 100 Food-paired arm 500 90 90 Correct arm entries (%) Unpaired arm Pellet consumption (%) 80 80 400 Time in arm (sec) 70 70 60 60 300 50 50 40 40 200 30 30 100 20 20 10 10 0 0 0 Not devalued Devalued Pre-devaluation Post-devaluation Win-Shift 100 100 90 90 Correct arm entries (%) Pellets consumed (%) 80 80 70 70 60 60 50 50 40 40 30 Devalued-correct entries 30 20 Not devalued-correct entries 20 Devalued-pellets 10 10 Not devalued-pellets 0 0 Pre-devaluation Post-devaluation Figure 6 Effects of reinforcer devaluation on three radial maze tasks. However. Performance was maintained after devaluation. Rats that did not experience devaluation exhibited a normal conditioned preference for their food-paired arms. The ideas are suggested by the Although the idea that different kinds of informa- findings already described and will be useful for tion are processed in different parts of the brain is organizing and interpreting the literature describing suggested by the effects of damage to individual . Adapted from Sage JR and Knowlton BJ (2000) Effects of US devaluation on win-stay and win-shift radial maze performance in rats. after devaluation they ate very few. This section describes a series of theoretical ideas 3. the maintenance of accuracy shows that positive affect was not required for performance of this task.3.1 Systems Concept that constitute a theory of multiple parallel memory systems in the brain. suggesting that a representation of the affective value of the reinforcer may have contributed to win-shift performance. whereas pellet consumption decreased. CCP: The graph on the top left shows the amounts of time spent by rats in their food-paired and unpaired radial maze arms in a CCP test (see section 3. Behav. Yin HH and Knowlton BJ (2002) Reinforcer devaluation abolishes conditioned cue preference: Evidence for stimulus-stimulus associations. Win-stay: The graph on the top right shows the percent correct responses and the percent of total food pellets available that were consumed before and after reinforcer devaluation. Win-shift: The graph on the bottom shows the percent correct responses and percent of total pellets consumed on the win-shift task before and after reinforcer devaluation.

There have been several attempts exclusive involvement of the hippocampus and cau. 2005. Leutgeb et al. dala (Pitkänen et al.. Although campus (Muller et al. ways they represent the relationships among the ele- patible. Although neural activity incompatible information.. 3. ible memory for facts and unique situations and their internal specializations lead to differences in the memory for stereotyped skills and habits are incom.02. tion comes to represent spatial relationships in one 1. and the hippocampus represents stimulus- through the systems as neural activity. Taube. and interoceptors. how they are represented mation into a unique representation in each system. 2004. 1996. 2001.02. and amyg- structure. each of which is specialized to extract and the external and internal sense organs is processed represent a different kind of information contained in the and represented in the thalamus and cerebral cor. Information originating in areas. Both systems originate in the striate S-R S-Rf S-S 3 Caudate nucleus Amygdala Hippocampus and prestriate visual cortex. 3. Fendt and Fanselow.1 Systems process incompatible specialized information The information processing specialization of each Sherry and Shachter (1987) suggested that indepen. This information is distributed to several reinforced stimulus-response (S-R) associations. According to this idea. Knierim et al.. These patterns converge on A similar concept can be applied to the evi. 2002.1. . 2 Thalamus and Cortex Mishkin et al. where they are processed dence for independent processing of different and combined into activity patterns that represent the current external and internal environments. 1999. to describe how the internal structures of the hippo- date nucleus in these types of memory.. 1998. 2001. the amygdala represents stimulus-reinforcer (S-Rf) different neural pathways or systems. This results in the differentiation of the infor. the fact that no individual brain area can Multiple parallel memory systems perform any function on its own leads naturally to Receptors the idea that information is processed in neural 1 systems or pathways that can. In this way the visual informa- Figure 7 The multiple memory systems concept.. suggests that similar considerations may apply to 2006). behavior and thought. flex. Gurney et al. thalamic and cortical areas.. representing all of the elements reaches all systems. Patterns of neural activity originate in exteroceptors system and object properties in the other. 2004). See text for more information about the information types. 2. damage to structures in this pathway impairs object recognition. The two visual systems described by Ungerleider and Mishkin (1982. Sherry and Shachter did not discuss the amygdala. making these structures parts of both pathways. These types of information that is the subject of this processed activity patterns are transmitted to subcortical chapter (see Figure 7).22 Multiple Memory Systems in the Brain: Cooperation and Competition brain areas.2 Systems are internally 3. A ventral path connects to inferior temporal areas. thought. in principle. 1997. 2001. undergoing stimulus (S-S) relationships. The outputs of the systems processing and modification at stages along the converge on brain areas that mediate behavior and way. Lever evidence from the triple dissociation experiment et al.. Mizumori and Leutgeb. It flows relationships. be defined anatomically. system is determined at the level of its neuronal dent neural systems evolved to process fundamentally and synaptic microcircuitry. activity patterns: the caudate nucleus represents tex.. the unique kind of information processed in that Graybiel. The dorsal path connects to inferior parietal areas.4. in the systems. 1983) are an example of anatomically defined neural systems that process different kinds of information.4. and how the outputs of the systems The outputs of the systems converge to influence interact. 4. This concept corresponds to the mutually ments (Figure 1). caudate nucleus (Centonze et al. Visual information originating in striate cortex is integrated with other information at stages along each pathway. 1998. damage to this pathway impairs learning the visual location of 4 Behavior Thought objects.1. Sargolini et al..

2 Information Processing and to a novel environment (O’Keefe and Dostrovsky. This is determined by the In some situations the outputs of the systems may correspondence between the information content of promote the same behavior or different parts of a the learning situation and the representational complex behavior required to perform a task. The development of hippocampal place cells. All of these possibilities may apply in various combi- 3.4 The learning-rate parameter nations at different stages of the learning process. The existence of a learned behavior ent representation of a spatial map (S-S information) also implies the existence of a memory for the infor- of the environment. Kim and Jung. the caudate system acquires a coherent representa- or they could be distributed throughout the systems. the more coherent the representation. Schafe et al. In other situations the spondence. Maren.5 Cooperation and competition 2005. This is suggested by the failure system.02. This tions of information processed by each system could and other evidence (see section 3. may be an example of the creation of a coher- processing. outputs of the systems may promote different beha- The coherence of the representation of a learning viors.4. Accordingly. 2003.4. 1976. converge (Figure 7). before the relationship between neuro- coherently after only a single trial. but this may require many more trials information stored in each system. is and/or thought when the representation is activated thought to reflect such an interaction. The amount of exposure of this information. Kentros et al.4.02. 2005.1.1. These specialization of the system. The definition of very slow to form a coherent representation of the a system means that the information processed by S-R information content of a learning situation such the system must be stored somewhere within the as the win-stay task. Memory 1971.02. 2006) process and store the among systems information mediated in these structures. could store and provide information to all systems. In the absence to attain coherence. As we have seen. Conversely. of rats with caudate lesions but an intact hippocam- The neuroplastic processes that store representa- pus to learn the win-stay task (see Figure 4). sents a particular situation. evidence for synaptic or required to form a coherent representation in a sys. The nature of the interactions The coherence of a representation reflects the degree among these outputs depends on the behavioral ten- to which all parts of the system are activated in a dency promoted by the representation in each system similar way when processing information that repre- and on the relative amplitudes of their outputs. Ferbinteanu and Shapiro. observation of learned behavior is 2003. The situation in a system is assumed to determine the improvement in win-stay performance produced by degree to which the system influences behavior fimbria-fornix lesions. tion of S-R information more easily (in fewer trials) A single storage location (e. This results in competitive interactions. preted. Kennedy and Shapiro. a tool for studying the neural basis of information 2004). explained in Figure 5.02. it is critical to identify the type of same situation.3 Coherence: Some determines the amplitude of its output. the evidence reviewed in this tions among the systems.4. neurophysiological changes correlated with the tem is a learning-rate parameter. O’Keefe. 3. The coherence of the representation of a situation in The idea that different kinds of information are a system usually improves with repeated exposure to processed in differently specialized parts of the brain the situation. the same system is mation processed by each system. At any given point during training in any learning situation. 2004.. the cerebral cortex) than the hippocampus system. Multiple Memory Systems in the Brain: Cooperation and Competition 23 Maren and Quirk. during recall.1. The better the corre- are cooperative interactions.5) suggest that be located in a discrete part or parts of each system. another system plastic processes and the storage of information can may simultaneously form a representation of the be studied. which occurs during the first 10–30 min of exposure 3. chapter does not pertain to memory storage . each system has acquired a representation of the situation with some degree of coherence that 3... Differences in this development of learned behavior cannot be inter- parameter are important determinants of the interac.g. One system may represent a situation means that. The fact that representations are better than others the systems all influence the behavior of the same Coherence is a hypothetical property of the repre- animal (or person) means that these outputs must sentation of a learning situation in a neural system.02.

1947).1. 1993). The food remained in the and the anatomical focus is on the hippocampus same spatial location. it was soon shown that both systems that probably process subtypes of information. Rats with caudate lesions amygdala.02. a double disso. normal controls.1. This observation can be explained In other groups of rats either the hippocampus or as the result of competition between the tendency the caudate nucleus was temporarily inactivated with to enter lit arms.02.(i) Competition on the cross 3. of information types (Mizumori et al. Although the task was originally introduced as a 1997. but either a right or a left turn and fimbria-fornix.. location of the food. 1946. Increasing the number of training trials favors behavior controlled by S-R 3.2. 2001). McDonald and White. sug- processing in the three structures. the nature To start each trial the rat was placed in the arm of the information being processed is inferred from immediately to the right or the arm immediately to the ability to perform a memory task of some kind. 2006).5 S-S versus S-R Information information. 1957). Rats were in the review of the historical development of the trained to find food in a constant arm of a radial maze. response interfered with behavior based on informa- The review is limited to studies in which parts of tion about the spatial location of the food. Evidence for independent information made fewer erroneous turns than normal rats.1 Competition on the radial maze maze Packard and McGaugh (1996) exploited the As we have already seen.2 Cross maze neurophysiological evidence for localized processing The cross maze paradigm (Tolman et al. and lesions of the caudate nucleus impair shown in Figure 9. illustrated and et al. dent on S-R information. Packard et al.24 Multiple Memory Systems in the Brain: Cooperation and Competition mechanisms. represented as S-R information in lidocaine during the test trials.1. information. the left of the food arm. spatial (S-S) and S-R information are acquired dur- ing training (Restle. caudate and hippocampal memory systems.4. is a simple and elegant obtained by manipulations of memory consolidation method for distinguishing between behavior pro- (Packard and White. when tested again after 8 more days of training the fimbria-fornix lesions did not just impair win-stay same rats made the turn that had been reinforced learning. Davis Blodgett and McCutchan.. Packard and Teather. 1991. As ing. is described. Space limitations do not permit a consideration of 3. during training. 2005. which led away from the food. indi- formance of the lesioned rats was better than that of cating S-R based learning. the caudate nucleus. with a tendency to forage for the processing and storage of different kinds of food in places where it had previously been available.5. lesions of the fimbria-fornix training trial effect on this task to dissociate the impair win-shift (S-S) but not win-stay (S-R) learn. 2004. After 8 days of . 1988). duced by processing S-S and S-R information. result- 3. It addresses the prior issue of localizing the caudate system.5.. increasing the availability of spatial Processing stimuli favors behavior controlled by S-S information 3. Packard and McGaugh. as described in Figure 5. 1992. represented as S-S information in the hippocampus system. they actually improved it so that the per. most normal rats ran to the correct spatial 1989. idea of independent memory systems. Gill and Mizumori.02. when tested after 8 days of train- win-stay but not win-shift learning (Packard et al. 1994.. Furthermore. and the was required to reach it. indicating spatial learning. brain areas (primarily imaging studies in humans).3 Dissociations of Memory Systems ing in improved win-stay performance.02.. ing. As caudate lesions (Mitchell and Hall. or evidence demonstration of learned behavior that is not depen- that each of the three systems appear to include sub. but ciation. Packard and Cahill. Disabling the hippocampus-system with fimbria-fornix lesions eliminated these errors. Caudate the brain are functionally disabled in various ways and lesions eliminated S-R learning and the erroneous to studies that measure relative levels of activation in responses it produced. Evidence for competition among local maze cues and either the left or right turn the systems is a major feature of the review.1 Studies with Rats 3.5.5. evidence explained in Figure 8.02. The remaining sections of this chapter focus on Competition between caudate-based S-R learning evidence that dissociates the information processing and spatial learning has also been demonstrated with properties of the three proposed memory systems.02. 1998. taken two at a gesting that in normal rats S-R associations between time.

but caudate nucleus inactivation had no effect required for the formation of a coherent caudate- (Figure 9). which is available on every trial. Increased nucleus. during the training trials rats are placed additional trials had strengthened the S-R represen- into the start arm and allowed to explore until they find the food. These observations are eters that produced approximately equal numbers of consistent with others showing that processing S-R rats that made the spatial and S-R responses on the information requires a functional caudate nucleus but test trial. As caudate system assumed control of the rats’ behavior. After a number of training trials a test trial is given. indicates These findings illustrate several important points. This direction of the turn the rat chooses to make on this trial shows that the behavioral outputs of the systems were indicates what kind of information was acquired during the training trials. Columbo et al. Training trials Test trials After 16 days rats that underwent caudate inacti- start vation made the spatial response. It is used ciation. start The hippocampus system apparently acquired a food former location of food coherent spatial representation of the location of the Figure 8 Use of the cross maze to distinguish between food in fewer trials than the caudate system required responding based on S-S and S-R information. but hippocampal inactivation neural function Using cross-maze training param- had no effect (Figure 9). 8 days hippocampal inactivation resulted in a random This double dissociation is consistent with the find- choice of arms. In contrast. training). CREB has been linked cessing with respect to the hippocampus and caudate to the formation of long-term memories). correct turn The spatial information was still present and able to on all trials influence behavior when the caudate was inactivated. showing that processing spatial information requires Finally. . After 16 days of training. A right turn (the response made during in competition with each other. indicates behavior conditions. This shows that no other information ings in the inactivation experiment. Multiple Memory Systems in the Brain: Cooperation and Competition 25 Cross maze paradigm capable of influencing behavior had been acquired at that time.2. (2003) measured phosphor- not a functional hippocampus. Taken together with ylated response element binding protein (pCREB) in the findings for 8 training days. the behavioral output of the turn. of the systems compete with each other for the con- date nucleus. they show that S-R memory is inflexible. When illustrated on the left. Second. This shows that the Barrier S-R Spatial (S-S) hippocampus-based spatial information was not Information Information eliminated by the additional training that resulted in acquisition of caudate-based S-R information.(ii) Dissociations in measures of nucleus inactivation. respectively. pro- response to the choice point stimulus. shown on the right. environment was acquired in fewer trials than are tion.5. behavior was controlled by the with a barrier that makes it function as a T-maze. The to acquire a coherent representation of the S-R asso- apparatus is a maze built in the shape of a cross. levels of pCREB expression were found in the hip- Perhaps the most interesting observation in this pocampus but not the caudate nucleus of the rats that study focuses on what the rats did when an inactiva. controlled by an S-S association – spatial information about the location of the food. the normal tendency to make the spatial ent hippocampus-based representation of the spatial response was eliminated by hippocampal inactiva.1. the S-S system is flexible. using sensory information at the choice point to produce an appropriate response. This requires a right tation sufficiently. they reveal a difference in the learning rate parameter: A coher- training. A left turn. As hippocampal system after 8 days of training.02. the normal tendency to make the S-R response was eliminated by caudate 3. which ducing the same response to a stimulus regardless of leads toward the food location. and in the caudate but not tion treatment impaired their normal behavior. This is consistent with other findings based representation of a reinforced S-R association. the barrier is moved and the rat is placed Inactivation of the caudate system allowed the output into the maze at the opposite end of the start arm. they constitute a the hippocampus and caudate nucleus 1 h after test- double dissociation of S-S and S-R information pro. ing (as explained elsewhere. trol of behavior. the pattern of effects shows that the outputs a functional hippocampus but not a functional cau. The of the hippocampus system to reassume control. After the hippocampus of rats that made the S-R response. Therefore. made the spatial response. behavior controlled by an S-R association – a specific First. which leads away from the food location.

indicating use of S-R information on the test trials given after 8 or 16 days of training. leading away from the food. as predicted by the hypothesis that turning toward the food was the result of hippocampus-based spatial information processing. causing them to make the response acquired during training. R (response) 8 days 16 days Caudate nucleus inactivation Cortex Cortex 12 number of rats S-S S-R S-S S-R 8 Hippocampus Caudate nucleus Hippocampus Caudate nucleus 4 Behavior Behavior 0 L (spatial) L (spatial) 8 days 16 days Figure 9 Dissociation and competition on the cross maze. or away from the food. This is con- and caudate nucleus while rats were being trained sistent with the finding in the inactivation . 1996). Top row: normal rats. 65: 66–72. After 8 days. but that it was prevented from doing so in normal rats by competition from the output of the caudate system. caudate inactivation (grey) had no effect. After 8 days most rats turned toward the food as a result of hippocampus-based processing of spatial information (light green). hippocampal inactivation (grey) produced random responding. allowing the rats to turn toward the food. ACh release in the observed in certain brain structures while learning hippocampus increased early in training and occurs (Ragozzino et al. This is predicted by the hypothesis that the caudate nucleus mediated the S-R information that produced the turn away from the food in the normal rats. Bottom row: inactivation of the caudate nucleus during the test trials.. even after the rats had (2003) measured ACh release in the hippocampus switched from spatial to S-R behavior. At this time insufficient caudate-based S-R learning had occurred to influence the response (white).26 Multiple Memory Systems in the Brain: Cooperation and Competition Normal 8 days 16 days 12 Cortex Cortex Spatial Response number of rats 8 S-S S-R S-S S-R Hippocampus Caudate nucleus Hippocampus Caudate nucleus 4 Behavior Behavior 0 8 days 16 days L (spatial) R (response) Hippocampus inactivation Cortex 12 Cortex number of rats S-S S-R S-S S-R 8 Hippocampus Caudate nucleus Hippocampus Caudate nucleus 4 Behavior Behavior 0 -. as predicted by the idea that this structure was not involved in processing the spatial information that produced the turn toward the food. After 8 days. Mem. After 16 training days. The graph shows the number of rats that turned in the direction of the food. which turned away from the food at this point (top row). consistent with the hypothesis that the hippocampus was not involved in this S-R type behavior. hippocampal inactivation (grey) did not affect the tendency to turn away from the food. Chang and Gold remained high throughout. Data from Packard MG and McGaugh JL (1996) Inactivation of hippocampus or caudate nucleus with lidocaine differentially affects expression of place and response learning. Learn. After 16 days. After 16 days. caudate inactivation (grey) resulted in turning toward the food. Neurobiol. reversing the behavior of normal rats. S-R learning had become strong enough for the output of the caudate system (dark green) to compete with the hippocampus-based response and take control over the rats’ behavior. Middle row: inactivation of the hippocampus during the test trials. indicating the use of spatial information. Eliminating the caudate output allowed the spatial information to resume its influence. Increased release of acetylcholine (ACh) is and tested on the cross maze. It also shows that the hippocampus-based spatial information was still present and able to influence behavior.

ACh release in the caudoputa. most of the its spatial location.02. but on S-S information about Figure 10). Normal rats learned to same time as they switched from spatial to S-R be. rats. Figure 11 also shows that half of the rats in the Learning to discriminate between two different stim. When they failed to trasted with learning to swim to a platform that find it. increasing with repeated in the pool. whose behavior was controlled by S-R information . fact that these shifts in behavior were produced by Using a slightly different version of this task lesions of fimbria-fornix and caudate nucleus is Packard and McGaugh (1992) found that learning consistent with the hypothesis that these structures to discriminate between two identical stimuli on the are critical for S-S and S-R information processing. The spatial information that allowed the cau- of a small platform submerged just below the surface date-lesioned rats to locate the platform when it was so that no local cues.1. This behavior suggests that it is visible to the rats. which allowed learning to swim to a hidden. all of the rats with the rats are responding to a local cue. climbing onto the platform. find a platform that remained in a constant location men remained low at first. while the other half swam to its former was impaired by caudate nucleus but not by fimbria. a final test trial was given. The rats can escape from the water by form was visible.5. It was visible on some trials and hidden trials and reaching its peak in individual rats at the on others (see Figure 10). They also learn to swim these rats were impaired at processing S-R infor- directly to this platform. location first. but that its influence on hidden because of their inability to process spatial behavior is eliminated by competitive behaviors information. This is consistent with an impairment in reported that lesions of the hippocampus impaired spatial (S-S) information processing. The rats whose could be performed on the basis of an S-R association behavior was controlled by spatial information is further evidence for the differences in the kinds of on the final trial initially learned the location of information processed in the neural systems that the hidden platform much faster than the rats include the hippocampus and the caudate nucleus. If the platform is moved the rats rats with caudate lesions swam directly to the for- have to relearn its location. including visual cues. coherent S-R representation more slowly. by fimbria-fornix but not by caudate nucleus lesions. they swam to the new location. These observations are consistent with those viously shown. no deficit in their ability 3. This task is usually con. suggesting normal spatial information processing.(i) Competition in the water to process information even when it is no longer maze McDonald and White (1994) trained rats to reflected in behavior. In contrast.1. Rats learn to swim To examine the possibility that S-R information directly to the platform location regardless of their processing was impaired in the caudate-lesioned starting position in the pool. identify hidden could serve the same purpose when the plat- its location.3 Water maze to locate the visible platform would be expected In this task (Morris. where it was visible (see sequence of responses. Morris and coworkers (1982) platform. showing that their be. The platform. swim to the platform in both conditions. normal control group swam directly to the visible uli that were moved to new locations on each trial platform. acquired S-R information to control behavior. rats with fimbria-fornix lesions from the inactivation experiment showing that swam to the platform normally when it was visible the hippocampal system acquires a coherent spatial but were severely impaired at finding it when it was representation quickly. The behavior of these two subgroups fornix lesions. As pre- havior.02.5. The platform was havior does not depend on any specific response or moved to a new location. but not to a visible.3. As shown in Figure 11. showing that to control behavior. 1981) rats are placed into a large because the platform was in the same location on all pool of opaque water and allowed to swim in search trials. but this behavior is not mation. This double dissociation between a of control rats on the hidden platform trials during task requiring spatial information and one that training is shown in Figure 10. Multiple Memory Systems in the Brain: Cooperation and Competition 27 experiment that the hippocampal system continues 3. which could be fimbria-fornix lesions swam directly to the visible a result of S-R memory. basis of their spatial locations in a pool was impaired respectively. where the protrudes just above the surface of the water so that platform was visible. Rats with caudate nucleus lesions originating from the caudate system that acquires a learned to swim to the hidden platform normally. mer location of the platform. Although S-R information processing may have been impaired in these rats. allowing the competing spatial information affected when the platform is moved.

1.5.02.5 m in diameter. The platform was always in the same location.(ii) Involvement of synaptic sibility of constitutional individual differences in functions Packard and Teather (1997) found the relative learning rates of the hippocampus and that injections of AP5. 61: 260–270. Neural Biol. Behav. five. 1998). Note impairment in fimbria-fornix lesion group on the hidden but not the visible trials.28 Multiple Memory Systems in the Brain: Cooperation and Competition Competition in the water maze: behavior on training trials Training Starting points platform visible on some trials. submerged on others Test platform moved to new location (visible) (a) Time to find visible platform (b) Time to find hidden platform 25 25 Caudate Caudate FF FF 20 Control 20 Control Seconds 15 Seconds 15 10 10 5 5 0 0 1 2 3 4 5 6 7 8 9 1 2 3 Trials Trials Control rats (c) hidden platform trials 25 Spatial location Visible platform 20 15 Seconds 10 5 0 1 2 3 Trial Figure 10 Competition in the water maze: training trials.3. (c) Two subgroups of the control group. NMDA receptors have been implicated in the synaptic . on the final trial. an N-methyl-D-aspartate caudate systems (see also Colombo and Gallagher. The significance of the difference between these two groups is explained in Figure 11 and in the text. The blue circle at the top of the figure represents a circular swimming pool about 1. 3. Hidden platform trials were given after three. (a) Average time taken to swim to the platform over nine trials when it was visible. (NMDA) receptor antagonist (as described elsewhere. it was visible on some trials and submerged (invisible) on others. and nine visible platform trials. This difference suggests the pos. Rats were placed into the pool once at each of the four starting points on each training trial. (b) Average time to locate the platform during three trials when it was hidden. which differed in the rate at which they learned to locate the hidden platform. Data from McDonald RJ and White NM (1994) Parallel information processing in the water maze: Evidence for independent memory systems involving dorsal striatum and hippocampus.

hippocampus and caudate nucleus. This suggests impaired processing of spatial information. 8/8 rats with fimbria-fornix lesions swam directly to the visible platform. 61: 260–270. Data from McDonald RJ and White NM (1994) Parallel information processing in the water maze: Evidence for independent memory systems involving dorsal striatum and hippocampus. platform. in the cau- the possibility that synaptic changes involved in the date nucleus more expressive cells were found storage of S-S and S-R information may occur in the following visible than hidden platform training. This is illustrated by the swim path (red line) of a rat with caudate nucleus lesions shown at the bottom left. dissociation has recently been made using cyto- ate early genes thought to be expressed in the chrome oxydase as a measure of metabolic . The behavior of these two subgroups of control rats during the initial hidden platform training trials is shown in Figure 10. events involved in memory storage). In visible platform. Behav. In the presence of a visible platform. whereas injections into the caudate the hippocampus more cells expressing these prod- nucleus impaired learning to swim to a visible. (2005) control rats not trained on either task. allowing spatial information to control the behavior of most rats in this group. this behavior suggests impaired processing of S-R information. On the test trial the rats were placed into the pool with the visible platform in a new location. A similar counted cells expressing c-Fos and c-Jun (immedi. Neural Biol. These findings suggest training than visible platform training. but ucts were counted following hidden platform not a hidden. they swam to the visible platform. Multiple Memory Systems in the Brain: Cooperation and Competition 29 Competition in the water maze: test trial behavior Visible platform in a new location 9 Previous location 8 New visible location 7 Numbrer of rats 6 5 4 3 2 1 0 Caudate F-F Control Lesion group Caudate nucleus lesions Fimbria-fornix lesions Former platform location New location of visible platform Figure 11 Competition in the water maze: test trial. allowing S-R information to control the behavior of these rats. when they failed to find it there. In the control group. four rats swam directly to the visible platform and four swam to its former location first. presence of neural activity and synaptic changes campus impaired learning to find a hidden but not a involved in memory) after water maze training. In contrast. These increases were also observed relative to In another demonstration Teather et al. respectively. so that spatial learning about the location of the platform competed with the S-R tendency to swim to the visible stimulus. as shown in the graph and illustrated on the bottom right. into the hippo. The graph at the top shows that 7/9 rats with caudate nucleus lesions swam to the old spatial location of the platform first.

30 Multiple Memory Systems in the Brain: Cooperation and Competition activity in the hippocampus and caudate nucleus Hartley and Burgess. Iaria et al. these observations maze on a computer screen and tested normal par- again suggest the possibility of synaptic activity ticipants on the win-shift task originally developed related to memory storage in the appropriate for rats. (2003) created a virtual eight-arm radial measures of brain function.3).02. The performance of all participants improved mation (possibly as it interacts with the dorsolateral over a series of trials. to identify and avoid et al.5. participants started by exhibiting significantly more .17. rocks. where the structure is often called stairs at the end of each arm.02. 2005a. Participants using this strategy showed 3. as few arms as possible. 2006). facilitated reinforced S-R responding the first arm they saw or from a single landmark. This errors. 2003. As the participant suggest that the dorsomedial caudoputamen in the rat returned to the platform from each arm (stimulus). such as proximity to somedial caudoputamen lesions have similar effects the sun and a tree for one arm and to the sun and a in the water maze (Devan et al.02. an elimination of the competition Analysis of this behavior showed that it involved a effect similar to that produced by inactivation or consistent series of turns to enter each of the correct lesions of the hippocampus system.1 Spatial learning havioral tasks and the brain areas with different Iaria et al. this function is actually confined instructed to obtain all of these objects by entering to its dorsolateral part (White.2. One set of experiments suggesting selves into three groups. (2003) concluded that this behavior was produced by processing S-R information. each of the eight arms and see a virtual landscape of mountains. A selection of this evi- while rats learned to locate a platform using dence is reviewed here. They could freely control their movements nucleus through the maze with the computer keyboard.b).. nucleus. The caudate nucleus has been described as part of the Goal objects were not visible from the platform but system that processes S-R information.3. A trial started with the participants at the brain areas when either S-S or S-R information is center of the maze.02. Each of these Chapters 3.18. of the caudoputamen in the rat processes S-S infor. One group reported they this possibility shows that fimbria-fornix and dor. on the cross maze. but not of the anterior or dorsolateral parts of strategy: counting arms in a constant direction from the structure. Participants using this strategy showed spatial information.1. These findings arms in order around the maze..5. 2004. Poldrack and Packard. but evidence could be obtained by descending a virtual flight of shows that in rats. 3. Aside from replicating the dissociation between the be. Another group reported using a nonspatial men. 3. spatial or local cues (Miranda et al.2 Studies with Humans significantly increased activation in the caudate Evidence from studies with humans on the dissocia. Yin et al. from where they could look down being processed. 1999). Holahan rock for the adjacent arm. (2005) found that blocking NMDA receptors in reentering arms. Reentries were scored as Featherstone and McDonald. When asked how they had caudoputamen in the selection of responses – see solved the task.. Yin and Knowlton (2004) showed significantly increased activation in the hippocam- that lesions of the posterior dorsomedial caudoputa. leading to processing hippocampal system. Accordingly. might be considered part of the S-S information the correct turn (response) was elicited. 1996.4 Medial versus lateral caudate maze. Some of the experiments were conducted portion of the caudoputamen in the rat may be with the subjects in an fMRI scanner. had used extramaze landmarks.5.. Participants were the caudoputamen. which is homologous with the caudate nucleus in primates thought to provide indirect information about the (Heimer et al. 2005). pus. relative levels of neural activity in different parts of There is also evidence that the dorsomedial part the brain. This constitutes a spatial strategy the dorsal hippocampus and dorsomedial caudoputa. they spontaneously sorted them- section 3. 2004a. the retrieval of a goal object (reinforcement). trees. and the sun surrounding the 3. tion of hippocampus-based declarative (S-S) learning A third group of participants said they started with from caudate nucleus-based procedural (or S-R) the spatial strategy but switched to a nonspatial strat- learning parallel the rat findings quite closely (See egy at some point during the session. 1985). 1989a.b. requiring the processing of relational S-S informa- men had similar effects on the long-term retention of tion.

(2003) about Parkinson’s patients performed normally. Poldrack and coworkers (1999) studied the brain 3. Although it is possible to perform the task of declarative. This used a nonspatial strategy. the set predicts rain or shine. whereas the Findings similar to those of Iaria et al. After they respond by However. classification task (see also Shohamy et al. improve beyond the 70% level. respectively. the weather forecasting task (Knowlton et al. processed were unable to learn the weather forecasting task at in the hippocampus and the caudate nucleus.02. whereas the normal based spatial (S-S) information and caudate-based participants continued to improve. participants were shown the same sets of cues . based on information processed by the hippocampal These findings show that two different kinds of system. 2004) and back is given on most trials. The feedback provided for a given Knowlton et al. This means that consistent feed. The basis of the shift may be a difference in questions about the conditions of the experiment the rates at which the two systems acquire coherent (declarative memory). situation and probabilistic classification (S-R) learn- Hartley et al. 1996). information by a hippocampus using declarative information about the feedback system. acquisition of an sition of this task appears to depend on information S-R association that produces the correct response processed by the caudate system. possibly because nonspatial (S-R) information. and participants are shown the stimulus. all (Knowlton et al. increased activation was seen in the right caudate 1996). Multiple Memory Systems in the Brain: Cooperation and Competition 31 activation in the hippocampus than in the caudate Knowlton et al. During the early trials.. the specific of procedural or S-R information by a neural system mapping between the cues and the feedback is quite that includes the caudate nucleus. 1994. but this relationship was reversed as train. to answer most of the questions. but on some trials the with the general distinction between the processing opposite feedback is given. the amnesic patients did not information are processed in this task: hippocampus. These the relationship between hippocampal and caudate findings constitute a double dissociation between nucleus activation in humans measured in more nat. Hartley and Burgess. the hippocampus became more active as pressing one of two keys on a keyboard. in which response to a set of cues is sufficiently high.2 Probabilistic classification areas involved in probabilistic classification in nor- In the probabilistic classification task.. participants are presented with a set of distinct nucleus (and in the frontal and occipital cortex). When given a series of tively. normal participants and a group of amnesic patients ing progressed and the strategy switch occurred. when the stimulus and the reinforced response have and receive feedback. These findings coincide response to each cue or combination of cues within with the neuropsychological data on the probabilistic a set is probabilistic. given most often. or S-S. 2003. also known as mal participants using fMRI. initial acqui- been repeated sufficiently often. hippocampus and caudate nucleus function with uralistic virtual environments have also been reported respect to declarative memory for the experimental by Maguire and coworkers (Maguire et al. 2001). cated in a study that also reported that Parkinson’s duces a shift from behavior controlled by S-S to patients (with impaired caudate nucleus function) behavior controlled by S-R information. As already shown. cues selected from a large group and asked whether Activity in the left hippocampus was suppressed. In a new task the can occur. 2005). ing.. make a response.. Furthermore. 1998. and the processing complex.04).. This finding was subsequently repli- rat evidence showing that increased training pro. The findings for mapping between the cues and outcomes (See the group that switched strategies also corresponds to Chapter 3. (1994) study. participants who used task at the same rate until both groups attained a the spatial strategy made more errors than those who moderate level of performance (70% correct). However. when the casting task (Poldrack et al. they are told performance continued to improve and reached that their response was either correct or incorrect levels unattainable by amnesic patients in the (feedback). This dissociation is they acquired some declarative knowledge of the similar to the one observed in rats. However.2. respec.. this requires a large number of These findings were replicated and extended by trials because of the probabilistic feedback and the comparing two different versions of the weather fore- complexity of the mapping.5. One was the probability of consistent feedback for a given standard probabilistic classification task. (1994) found that a group of nucleus. (with impaired hippocampal function) learned this Throughout the experiment. Errors for the group that suggests that within this range. performance is not switched strategies decreased as soon as they switched. the amnesiacs were unable representations of the information they process.

and that caudate nucleus function is implicated systems (Poldrack and Rodriguez. Several instances of competition for control of behavior between the two systems in which these structures reside have been described.3). these findings constitute This may be a result of the relationships between the a double dissociation of caudate-based processing of processing capacities of the structures and the infor- reinforced S-R information and hippocampus-based mation that activates them.1) suggests the possibility that. behavior. differences in the coherence or chemical inactivation.5. was observed. caudate nucleus. whereas the paired associ. These relationships temporal lobe function (Warrington and Weiskrantz. and increased activa- among the stimuli and the weather they predicted. 2004). Instead. tion in the caudate nucleus occurs when S-R This type of task is known to depend on medial information is being processed. 2004).4.6 S-S versus S-Rf Information mation processing functions in the systems. lead to the hypothesis that the level of activation 1982.02. the weather prediction performance coherence of the representations of these information of both groups was similar.32 Multiple Memory Systems in the Brain: Cooperation and Competition together with the correct response for each set.02. Applying Sherry and processing of S-S information consistent with other Shachter’s (1987) suggestion that these systems pro- animal and human data described. increased hippocampal activation occurs when S-S they acquired information about the relationships information is being processed. inhibitory task can be characterized as consisting of S-S associa. the information available in that situation 3. The triple dissociation experiment (Figure 4) included able in each learning situation – the task demands a double dissociation between performance on the (Poldrack and Rodriguez. and indices of increased of the representations formed in the systems deter- functionality. Accordingly. with depressed activation in the non- ates version activated the hippocampus more than the dominant structure (Poldrack and Rodriguez. In the fMRI studies. this sugges- when the task can be performed using information tion shows that it is not necessary to postulate such consisting of S-R associations. 3. including activation detected by mine which one wins the competition to control fMRI. which can cating these brain structures includes the effects of be understood simply in terms of independent paral- impaired function produced by trauma (including lel systems that process incompatible information. ACh release. did not make responses or receive feedback. lesions made in the laboratory). . direct influences to explain the data.4.1. studies strongly suggest that hippocampal function is While not ruling out the possible existence of implicated when the information required to solve a direct or indirect functional interactions. Marchand et al.3 Summary: Competition and is incompatible with the specialized representational Coherence capacity of the other system. This is consistent CCP (impaired by amygala lesions) tasks. This with the idea that the information processing special.02. disease processes. types in the systems. The system with the most interpretation. the representation it forms of a learning situation (see Other studies with rats and humans support this section 3. data from both animal and human activation in the fMRI. when one of the two systems develops a coherent representation of a situation. 2004) – determine which win-shift (impaired by fimbria-fornix lesions) and neural system controls behavior. cess incompatible information (see section 3. or otherwise. reflected in decreased Packard. 2004). The evidence impli. and c-Fos expression. the participants nant effect on behavior. According to this view. fMRI scanning showed that the reinforced S-R In some studies an inverse relationship between version of the task activated the caudate nucleus activation of the hippocampus and caudate nucleus more than the hippocampus. 2003). parallel infor. These findings are strong evidence for independent. In this coherent representation is the one that has the domi- paired associates version of the task.02. By the end of the revealed by the fMRI may be a reflection of the training trials. was interpreted as a double dissociation between the ization of each system determines the coherence of neural systems that process S-S and S-Rf information.. This may lead to an As reviewed here and elsewhere (Poldrack and incoherent representation. Processing The data from studies with normal animals and humans suggest that the kinds of information avail. between the hippocampus and caudate tions.

training. see Figure 12). ABC and DEF are sets of environmental cues visible from the maze arms. This preference is elimi. resulting in a preference for that arm. The three phases of the conditioned cue preference (CCP) paradigm (preexposure. and the arms were differentiated by a light two separated arms (Figure 12). with ABC as the conditioned stimulus (CS). completely different sets of cues are visible from the food-paired and no-food arms. the same cues (ABC) are visible from both the food-paired and the unpaired arms. 1993). During the CCP training trials in the triple dissocia- In the sCCP situation. The lower part of the figure shows the maze configurations for each phase in the separated and adjacent arms conditions. but not by amygdala lesions (Chai and paired cues. (White and McDonald. In the separated arms CCP situation. and test) are shown at the top.6. completely different sets tion experiment. When the two arms because no food is available.02. the radial maze was surrounded by of extramaze cues are visible from the ends of the curtains.1.1 CCP with spatial cues visible from separated and adjacent maze arms. the are adjacent to each other (the adjacent arms CCP conditioned response causes the rats to approach [aCCP]) the preference is eliminated by hippocam. On the test trial. During the training trials. No response to DEF is learned. They do not acquire a response to the cues nated by amygdala but not by fimbria-fornix lesions visible from the unpaired arm while confined there. The rats acquire a in one of them. Discrimination between the two arms using these cues requires spatial learning. in the other exploration (no preference) arm with no food (preference?) Separated arms (sCCP) impaired by amygdala lesions B C A F E D Adjacent arms (aCCP) impaired by hippocampus lesions B C A F E D Figure 12 Conditioned cue preference with separate and adjacent radial maze arms. On the test day. 2004).1 Studies with Rats White.02. Similar preferences are learned when conditioned response to the environmental cues the maze is open to the surrounding extramaze cues visible from the food-paired arm when they eat and the food-paired and no-food arms are on oppo- while confined on that arm during the training site sides of the maze (the separated arms CCP [sCCP].6. . In the adjacent arms CCP situation. trials. Multiple Memory Systems in the Brain: Cooperation and Competition 33 3. Conditioned cue preference (CCP) on the radial maze Preexposure Training Test Unreinforced Confined in one arm Unreinforced exploration with food. This double dissociation is explained by differences in the nature of the extramaze cues 3. conditioned reward elicited by the CS causes the rat to enter and spend more time in the food-paired than in the unpaired arm. a conditioned rewarding response is acquired. and spend more time in the presence of the food- pal lesions.

1929. As shown in Figure 13. is hippocampus dependent (Smith. other aspects of the two procedures are identical. 2004). sions. When the competition between the processing of S-S and S-Rf information. the aCCP (Figure 13) (Chai and White. The parallel occurrence of hippocampus-based spatial learning and amygdala-based S-Rf learning in the aCCP paradigm results in competition 3.34 Multiple Memory Systems in the Brain: Cooperation and Competition This arm discrimination is produced by amygdala.1. 1998). is eliminated by amygdala lesions. all been investigated in more detail (White and Gaskin. (Sutherland. hence. hippocampus-based S-S learning promote a tendency paired arm is acquired in the sCCP procedure. called prevented from moving around in an environment path integration. 1997.2 Cooperation and competition in expression of the aCCP. In In the aCCP situation. establishing the light as a con- sessions to express the aCCP.02. ditioned cue. adjacent arms CCP learning The competitive interaction between the amyg- Although the aCCP and sCCP paradigms differ in dala and hippocampus systems in the aCCP task has the relationship of the arms to be discriminated. 1985. 1997. but rats is revealed by the finding that rats with amygdala with hippocampal lesions required fewer trials than . White and Ouellet. purpose (Etienne et al. 1930). system to acquire a sufficiently coherent representa- Therefore. This prediction has been confirmed (Chai competitive interaction. During the first phase.1. requiring maze environment is required for the hippocampus hippocampus-based processing of S-S information. rats also use internal cues based on proprio- information about the layout of the maze cannot ceptive stimuli generated by their movements for this be acquired during these trials. rats do not learn the Ito et al. the amygdala-based tendency to enter cues are visible from the ends of both arms (Figure 12). The cue was removed before some time that rats acquire spatial information dur. During these ses. At the same time. Prior to each trial. However. a cooperative interac- same response should also be acquired in the aCCP tion. because spatial learn. The rats acquired a tendency to Interference with expression of this spatial infor. information it provided was a result of path integra- a phenomenon called latent learning (Blodgett. but this tendency visible from both arms in the aCCP procedure. a flashing light together Tolman and Honzik. 2004). 1996. 2001. Terrazas 1997. a less coherent respectively. Whishaw et al. the S-Rf learning promotes a procedure. a both arms. since most of the same cues are tendency to enter the no-food arm. Whishaw.. cue conditioning Although a conditioned response to the food-paired In addition to using visible spatial cues to navigate in cues is acquired during the training trials. rats with amygdala mation by the amygdala-based conditioned response lesions required more trials than controls. most of the same extramaze normal rats.. this is blocked by an interfering tendency resulting from response should produce an equal tendency to enter S-S information about the lack of food in the arm. A mini- about differences in the relationships of their locations mum of three freely moving preexposures to the to those cues. the rats are allowed to move around freely on the rats were exposed to a polarizing cue in the the maze with no food present. the trial started. the to enter the food-paired arm. hippocampal representation (with. For this reason. with a sucrose solution were presented in any of the normal rats require a minimum of three preexposure three compartments. lesions require only one preexposure session to learn based processing of S-Rf information. McNaughton et al. et al. spatial space. If a 2006). This form of information processing.6..6. both arms competes with the hippocampus-based Discriminating between them requires learning tendency to discriminate between the arms. (2006) trained rats in an apparatus consisting aCCP unless they are preexposed to the maze before of a triangular central platform with compartments the training trials (see Figure 13).02. attached to each of the three sides. This is spatial learning.. ing in rats is severely attenuated when they are 1996). It has been known for experimental room. These cooperative and and White. Both amygdala-based S-Rf learning and conditioned response to cues visible from the food. competing tendencies are illustrated in Figure 13.3 Path integration versus visual between the behavioral outputs of the two systems. the double dissociation between the effects tion of the spatial environment for its output to win of amygdala and hippocampus system lesions on the the competition with the output from the amygdala aCCP and sCCP tasks constitutes a dissociation system for control of behavior. approach the flashing light. less preex- posure) is required to produce output that results in 3. tion. so subsequent use of the orienting ing unreinforced exploration of a novel environment. 2005).

However. the context (Fanselow. 1993. Ito et al. but the rats with hippocampus In multiple memory systems terms. Because of the amygdala but not the hippocampus when the neither the conditioned cues nor visual spatial cues CS is a discrete cue suggests that this form of fear were available during this test. Neutral stimuli become CSs with the issue requires experiments that directly compare appe- capacity to elicit the internal state of fear without titive and aversive learning in the same experimental the skeletal responses. . as in the case of flashing lights appeared in any of the three compart. fear conditioning. This appears to be The rats were then given a series of trials in which another instance of latent learning. The sham-operated impair contextual fear conditioning. 1992). context The results of the two tests taken together double dissociate the use of path integration based on the conditioning also requires an intact amygdala. The context conditioning. its effect on precedes the shock. In cue conditioning. (2006) attrib- conditioning is a result of amygdala-based proces- uted the preference to information about the spatial sing of S-Rf information. but sucrose delivery accompanied this cue situation. This informa- the involvement of the hippocampus suggests tion would have been incidentally acquired during that conditioning requires hippocampus-based pro- the cue-training procedure. but only amyg- rats and the rats with amygdala lesions showed a dala lesions impair conditioning to a discrete cue preference for the compartment in which they had CS (Phillips and LeDoux. sug- processing of spatial information derived from move. minimum amount of unreinforced preexposure to formance. Both lesions of the hippocampus (Kim and The compartment in which sucrose was available Fanselow. The rats were then Frankland et al. The demonstration of partial dissociation of S-S and S-Rf information interference with the expression of the amygdala. 1992. However. remained constant on all trials. the fear 3. cessing of S-S information..1. as described in Figure 5. Fanselow and processing of S-Rf information and that information Gale. ences in the experimental apparatus and procedures The shock evokes a constellation of URs including used in the two sets of studies make it difficult to an aversive internal state (usually called fear) and conclude that the systems have different functions in skeletal responses. When the CS is a context location of the sucrose based on the polarizing cue. received sucrose. 1990). unreinforced preexposure in the appetitive aCCP ments.6. the involvement lesions did not exhibit this preference. processed in the hippocampus of the normal rats Two kinds of stimuli can serve as the CS. In interfered with the expression of this behavior.. 1992. 2004). Gale et al. 2003). primarily consisting of jumping appetitive and aversive learning. a with- the cue preference was a result of amygdala-based drawal response (LeDoux. processed in the hippocampus system. 1998) and lesions of the amygdala given a compartment preference test in the absence (Phillips and LeDoux. Context conditioning requires a spatial information can interfere with win-stay per. The cue becomes a CS. This suggests that state is thought to be reflected by freezing. processing by the hippocampus and amygdala in based information by hippocampus-based path inte. (as well as the room in which it is located and every neous acquisition of information about the location of other feature of the situation). 2004) of flashing lights and sucrose. 1992. 1992. the CS is the test cage itself interference could have been caused by the simulta. relationship between hippocampus-based spatial learn- ing with appetitive reinforcers and amygdala-based appetitive conditioned responding. Gale et al.02. discrete cue such as a tone or light immediately Because this information was irrelevant. rats are placed into a test memory systems hypotheses. This conditioned internal conditions. Examination of this and running. gesting that S-Rf information is also required for ment by the hippocampus and the processing of S-Rf context conditioning..4 Fear conditioning conditioning studies do not explicitly test multiple In these experiments. Phillips and LeDoux. requiring preexposure for conditioning to occur. but the behavior may have interfered with expression of context also becomes a CS in this situation (Phillips the amygdala-based cue preference in the same way and LeDoux. and the numerous differ- cage and given one or more foot shocks as the US. a the sucrose solution based on the polarizing cue. Multiple Memory Systems in the Brain: Cooperation and Competition 35 controls to acquire this behavior. gration is evidence for independent functioning of This pattern of effects differs from that for the the two systems. only when it appeared in one of the compartments. These findings constitute a information by the amygdala.

36 Multiple Memory Systems in the Brain: Cooperation and Competition Competition in the adjacent arms CCP (a) 300 Food-Paired arm Unpaired arm Time in arms (sec) 200 100 0 0 Preexposures 1 Preexposure 3 Preexposures 1 Preexposure 1 Preexposure Amygdala lesion Amygdala+Fimbria- fornix lesion (b) Pre-exposure Training Test Competition Cooperation Unreinforced Forage for food exploration Confined in one arm with food. in the other arm with no food Acquire spatial Acquire information Acquire conditioned (S-S) information Latent about spatial response to food- about maze locations of food paired stimuli visible environment learning and no food from both arms (fimbria-fornix (hippocampus) (amygdala) dependent) Tendency to discriminate Tendency to enter between arms both arms Cooperation on food-paired arm competition on no food arm .

2. each opened box also revealed one of six of the CS is maintained during the CS-US interval. eye-blink conditioning in rabbits and 3. the (Thompson and Krupa. Adapted from Chai S-C and White NM (2004) Effects of fimbria-fornix. but not to appetitive CCP response is similar to other URs because it is elicited learning. These hippocampus-based behaviors are shown as green arrows on the maze illustrating the test trial in (b). (a) Learning the aCCP. Trace conditioning is were paired with reward on 90% of the trials on impaired by lesions of the nucleus interpositus of the which they appeared. (b) Amygdala-based competition. the CS is presented first. hippocampus and amygdala lesions on discrimination between proximal locations. Thompson. two were paired on 50% of Figure 13 Competition in the adjacent arms conditioned cue preference (aCCP) task. Using this procedure. because the hippocampus produces the same tendency. normal participants were (Skelton. 2005). by a reinforcer without previous experience. Behav. 2002).02.2 Experiments with Humans and the two stimuli terminate simultaneously. 1988. 1994) – but told that one of three black boxes on a computer not by lesions of the hippocampus (Christian and screen contained a red ball. This results in cooperation between the two systems on the food-paired arm. This is called trace conditioning appeared in each of the three black boxes. 1999). and the hippocampus processes both S-Rf and S-S 2003. the suggest that S-Rf information involving a skeletal participant received a reward (a raisin or small response (eye blink) is processed in the cerebellum. If the box contained the red ball. 1996.0 s. The amygdala-based tendency to enter the no-food arm competes with the hippocampus-based tendency to enter the food-paired arm instead of the no-food arm. Multiple Memory Systems in the Brain: Cooperation and Competition 37 3. Neurosci. 1994. Amygdala lesions eliminate the competition. the CS is pre. three preexposure sessions permit the acquisition of sufficient spatial information to express an aCCP when combined with information about the locations of food and the absence of food acquired during the training trials. 118: 770–784. As shown in (a). 2004). leg flexion (Maschke et al. The top part of the figure shows the effects of preexposure on the separated arms CCP (sCCP) learning. 2002. If the box contained a black ball.02. Gaskin S and White NM (2006) Cooperation and competition between the dorsal hippocampus and lateral amygdala in spatial discrimination learning. Hippocampus 16: 577–585.. All designs were resulting in the formation of an S-S association presented an equal number of times. information. responses directly elicited by a US: eye blink These findings suggest that. 3. This pattern cor- In eye-blink conditioning.5–1.1 Conditioned preference cats is impaired by lesions of specific parts of In an adaptation of the rat CCP task for humans the cerebellum – cortex or nucleus interpositus ( Johnsrude et al. This suggests that in normal rats. suggesting that amygdala-processed information competed with the hippocampus-based arm discrimination. acquired during the training trials.1. Amygdala lesions eliminate the amygala-based tendency to enter the food-paired arm. The participants were in- sented and terminates. These findings it to open.. candy). results in an equal tendency to enter both arms (red arrows). in this paradigm. a conditioned cue preference.5 Skeletal conditioning cerebellum in rabbits (Woodruff-Pak et al. After an interval of 0. 2005) and cerebellum processes S-Rf information only.6.. The bars show the mean amounts of time the rats chose to spend in the maze arms during the 20-min test trial with no food present. the US follows after a brief delay.6. As explained in the text.6. Rats that received fewer than three sessions of unreinforced preexposure to the maze (see text for explanation) failed to learn the aCCP. an amygdala-based conditioned response to ambiguous cues visible from both arms. This is a case of amygdala-based processing of S-Rf information competing with a spatial discrimination resulting from hippocampus-based processing of S-S (spatial) information. aversive tone sounded. Thompson and Krupa. allowing hippocampus-based spatial information to produce the CCP after only a single preexposure session. In the delay conditioning procedure.02. Only the former will be discussed here. a mildly In a slightly different procedure. abstract designs in the background. structed to count the number of times the red ball the US is presented. Touching a box caused Thompson. used to study at least two overt discrete skeletal 1990. a puff of air to the eye or a responds to that for aversive fear conditioning mild orbital shock (US) elicits a blink (UR). but rats that received three sessions of unreinforced preexposure to the maze spent more time in their food-paired than in their unpaired arm. Dimitrova et al. a partial dissociation. but this does not eliminate the preference. This with a contextual CS. Clark et al. Two of them between the trace and the US. rats with amygdala lesions expressed a CCP after only one session of unreinforced preexposure to the maze (compared to a minimum of three for a normal rat). 2003... . Thompson. Thompson and Kim. In addition because it is thought that a trace (or representation) to a ball. 1985) and The Pavlovian conditioning paradigm has also been by lesions of the hippocampus in rats (Moyer et al..

ball counting. and that trace conditioning evoked involving amygdala and the hippocampal systems in only activation in the hippocampus. . contextual fear conditioning or for trace conditioning making this discrete cue the CS. The patient with damage to which one they preferred. task ( Johnsrude et al. They were function acquired the conditioned electrodermal then shown the abstract designs in randomly selected response but had poor memory for details of the pairs with no balls present and asked to indicate experimental situation.. impaired by lesions of the cerebellum (Gerwig et al.02. In the final phase of S-Rf information from hippocampus-mediated pro- the experiment the participants were shown their cessing of S-S information in humans. All did very well on this task. but performed 2006) and the hippocampus (Thompson and Kim.6. All participants attributed their preferences to appetitive spatial learning in rats and to the data for properties of the patterns themselves. tioned any relationship between the patterns and reward.. in both rats and humans.. none men. toencephalography) in humans (Kirsch et al. Several studies.02. 2002). a more complete dissociation of processing the information types involved in trace and delay condi- tioning than the lesion data. a conditioned preference. aversive conditioning in rats with a discrete cue CS. but Two groups of surgical patients were tested on this not by lesions of the hippocampus (Daum et al. 2003) tex corresponds to the dissociations between the found that delay conditioning evoked only activation processing of these same two kinds of information in the cerebellum. and trace conditioning is resections of the amygdala did not acquire the pref. failed to acquire a conditioned electrodermal After the training trials participants were asked to response but had excellent recall of the experimental say how many times the red ball had appeared in situation. Patients with unilateral 1991. 1996. appetitive and aversive conditioning are real.38 Multiple Memory Systems in the Brain: Cooperation and Competition the trials. 2000). Shortly after of the eye-blink response. and two on 10% of the trials. one with bilateral damage restricted to amygdala as a result of Urbach-Weithe disease.6. The patient with impaired hippocampal each box. Furthermore. and one with damage amygdala-based S-Rf information processing have to both structures resulting from herpes simplex been reviewed. In contrast. 1993). consistent one with hippocampal lesions caused by anoxia with the dissociation of hippocampus-based S-S and secondary to cardiac arrests. These are fewer in number and are encephalitis. This suggests the rat. As is the case for rats. These chosen significantly more often than the 10% pat. This partial dissociation patients with frontal lobe resections acquired normal coincides with the data from studies of eye-blink pattern preferences but were severely impaired on conditioning in the rat. erence for the reward-paired patterns. eye-blink condi- paired patterns that did not depend on cognitive or tioning with the delay paradigm is impaired by declarative information. 3. one lel processing of the information types required for color coincided with the presentation of the US. 1995).. This double preferred patterns and asked why they preferred dissociation corresponds to the dissociation found for them.2. findings dissociate amygdala-mediated processing of terns. a loud horn and the UR was the electrodermal the available data on aversive conditioning in rats are response produced by the startling noise. lesions of the cerebellum (Daum et al. Clark et al. This double dissociation between A study measuring the magnetic flux of specific amygdala-based processing of S-Rf information brain areas as an index of neural activity (magne- and S-S information requiring an intact frontal cor.2. Further experiments are testing on this task. Gabrieli et al. The patients were tested on a usually more difficult to demonstrate than dissocia- Pavlovian conditioning task in which the US was tions between S-S and S-R processing. in humans...6. accurately on the ball counting task. The 90% patterns were both areas performed poorly on both tasks. These designs The patient with impaired amygdala function were not mentioned in the instructions. This was interpreted to mean that they had 3.. A series not consistent with the notion of independent paral- of color slides was shown during conditioning.2 Conditioned fear Bechara and coworkers (1995) compared three 3.3 Skeletal responses acquired a conditioned preference for the reward.02. the subjects were asked a series required to determine whether these apparent differ- of questions about the experimental situation as a ences in the relationships among the systems during test of their declarative memory.3 Summary patients.

and amygdala. After Points at which the outputs of the systems converge reaching a performance criterion. The acquisition of a CCP by rats that important ways.8 Summary and Some responses are unrelated to any spontaneous behavior Outstanding Issues that may occur in the learning situation and. the evidence for CCP learning in the same apparatus. and the double dissociations in both rats behavioral influence of these conditioned responses and humans all point to the idea that different is often indirect. 1994. and a neural sys- tem that includes the amygdala processes S-Rf information. amygdala on the win-stay task (see Figure 14). a simultaneous comparison of the win-stay and CCP The systems are not completely divergent. at least at the level of the hippocampus. 2001). 1994. Rats with caudate In the triple dissociation experiment. Everitt et al. These information-processing specializa- 3. The radial maze. This double dissociation of These findings dissociate the neural system that S-Rf and S-R information processing has been repli. but the reinforcer is not improve performance of the other task. the caudate nucleus and connecting structures process S-R information. the association performance of only one of the two tasks and did between them is strengthened. no evidence of a competitive interaction between the ory. Rats with lesions of the caudate nucleus. which become asso- ciated with contemporaneous stimuli (CSs). Caudate-mediated S-R information were never explicitly trained in that task is a demon- comprises associations between representations of stration of the fact that the caudate and amygdala any response an individual can make and representa. McDonald and Hong (2004) trained them receive input from the cortex. The rats were shows that there is clearly divergence among the placed on the maze for 20 min with all arms open and systems. S-S information. the win-stay task was amygdala lesions were normal on win-stay but failed impaired by lesions of the caudate nucleus but not by to exhibit a CCP. these differ in (amygdala).7 S-Rf versus S-R Information was on in four randomly selected arms. The 3. only a single experiment dissociat. vations of cooperative. lesions of the amygdala. Multiple Memory Systems in the Brain: Cooperation and Competition 39 3.02. However. Observing their existence often parts of the brain are involved in learning that requires requires some form of additional instrumental learn. All of radial maze tasks. processes the S-R information required for win-stay cated and extended.7. 1989. to associations between responses (URs) that are elicited by reinforcers (USs). The light used for win-stay training caudate nucleus. a CCP. Since lesions of either structure affected with the occurrence of a reinforcer.02. If the creation of different kinds of information in the same learning these two representations is temporally contiguous situation. Dickinson. interactions between systems are strong evidence In this section.02. Amygdala-mediated S-Rf information is limited systems in this situation. called systems. The analysis in this chapter concludes these two systems with animals or of any parallel that the hippocampus and related structures process experiments with humans. memory for different kinds of information. all rats were tested could also be shared parts. This is the multiple information processing is described. there is not part of the information that constitutes the mem. performance (caudate nucleus) from the system that Although S-Rf and S-R information both consist processes the S-Rf information required for the CCP of stimulus-response relationships. the other four Processing arms were dark. are largely unobservable internal The historical evidence.. Dickinson and Balleine. no food present. that these parts of the brain. the triple dissociation on the changes in autonomic and hormonal function. function ing amygdala-based S-Rf and caudate-based S-R independently and in parallel. . The obser- ing (Dickinson and Dawson. as summarized unaware of any other direct dissociations between in Figure 7. The author is parallel memory systems hypothesis.1 Win-Stay and CCP Learning tions may result from differences in the internal The S-R and S-Rf systems have been dissociated with microstructure of the systems. Normal rats spent more time in the lit than in the dark arms. and especially competitive. as already described. so the cortex rats with lesions of the dorsolateral caudate nucleus or could be considered a shared part of all systems. memory systems function simultaneously to acquire tions of contemporaneous stimuli. the CCP task lesions were impaired on win-stay performance but was impaired by lesions of the amygdala but not by exhibited a preference for the lit arms.

there is little or or microcircuits within the system will be activated . resulting from the neuroplasticity of their internal Although all information flows through all sys- structures. In the second part of the experiment. the specialization of each system means that of information that flows through the system on future each can represent only one type of information. The rats were placed on the maze with no food available and allowed to move around freely. This is how sentations in each structure will probably differ in experience alters behavior. The graph shows that the normal rats and the rats with caudate lesions spent more time in the lit arms than in the dark arms. For occasions. but by the degree to which the information content of a this does not mean the changes actually represent the learning situation coincides with the information- information). When the the systems. Rats with amygdala lesions performed normally. and four arms remained dark. Although several parts of each system are degree of coincidence is high. the repre- effect it has on thought and behavior. respectively.40 Multiple Memory Systems in the Brain: Cooperation and Competition Dissociation of S-R and S-Rf Information processing Win-stay CCP light food Win-stay training CCP test 90 300 Food-paired Unpaired Percent correct arm choices Time in arms (sec) 80 200 70 100 60 0 Normal Caudate Amygdala Normal Caudate lesion Amygdala lesion lesion lesion Figure 14 Dissociation of stimulus-response (S-R) and stimulus-reinforcer (S-Rf) information processing in the caudate nucleus and amygdala. but the performance of rats with caudate nucleus lesions was impaired. The rats with amygdala lesions did not exhibit this preference. Four randomly selected arms were lit. the coherence of a representation is determined influence neural activity representing information. most neural elements known to have neuroplastic properties. Coherence is a function of two factors. The alterations themselves coherence. The memories are therefore located in processing specialization of a system. altering the output of the system and the any given situation at any point in time. These alterations influence the processing tems. The graph shows the mean percent correct responses over the last 3 days of training for each group. The left side of the figure shows performance of three groups of rats trained on the win- stay radial maze task. Neuroscience 124: 507–513. This pattern of effects shows that rats can process and store amygdala-based S-Rf information (a conditioned approach response to the light CS) during win-stay training. are memories (note that these neuroplastic changes First. The findings dissociate caudate-based processing of S-R information from amygdala-based processing of S-Rf information and are consistent with the hypothesis that the two systems function independently of each other. the same rats were given a win-stay test. shown on the right. Adapted from McDonald RJ and Hong NS (2004) A dissociation of dorso-lateral striatum and amygdala function on the same stimulus-response habit task. a conditioned cue preference (CCP). As information flows through the systems (Figure 7) no definitive evidence localizing specific memories to it may produce temporary or permanent alterations specific structures within a system. even if their win-stay performance is very poor because of caudate lesions.

it should be tested in more different be- the number of training trials) increases the coherence havioral learning situations that can be parsed into of a representation. also be investigated at the level of the neural micro- The data also suggest that the systems have dif. In this case. apparent when a lesion impairs performance on a task 3. This kind of coherent representation tional independence of the proposed systems within produces output with a powerful influence on behav. they interfere with each the outputs of the systems interact? other. The postulated cooperative and competitive sentations of a situation might produce different interactions among the outputs of the systems require behaviors. Eliminating the sing of similar information on future occasions behavior would require impairments of both systems.8. and not all may be activated in the same way. Because the outputs may pro. the hippocampal experience alters neural systems. Specifically. it should be possible to specify how situa- the representations consist of different information. The idea that systems represent information can that recovers after additional training. The present summary has ior. This would be cooperation between the synaptic relationships in the system so that its proces- systems. impairing the function of either system would have no apparent effect. Increased exposure to a situation (by increasing First. the effects of poor representation sufficiently for the output of disabling one or more systems should become the system to influence behavior. 2002. These com- plaints largely focus on behavioral evidence from experiments with nonhuman primates and point out 3. where and how do duce different behaviors. forming an accurate representation 1. 4. 2001. (Gaffan. The outputs of two systems with coherent repre. A given situation might produce changes known to be produced by that system. Impairing the function of either system would 5. and reinforcers. Wise. This would be a competitive interaction further investigation on both the anatomical and func- between the systems. Practice can increase the coherence of even a As more situations are tested. which memories are stored. ior. This list is necessarily limited to a few of the would either support the theory or reveal its most general ones. but specialization of a system. increasing its influence on behav. When there is a mismatch between the emphasized evidence for their independent coopera- information content of a learning situation and the tive or competitive influence on behavior. Even though Ultimately. it will form a less coherent. possibilities for direct facilitatory or inhibitory actions or even incoherent. structure of each system. of one system on another also require further consid- tion. Fewer of its neural elements may be activated. further examination and definition. representation of the informa. eration and investigation. they all tem the study of these neuroplastic processes should be form some kind of coherent or incoherent representa- done in parallel with an examination of behavioral tion of all situations.02. Multiple Memory Systems in the Brain: Cooperation and Competition 41 in the same way. The idea that each proposed system is special- coherent representations produce output with weak ized to represent a different type of information. The implications of this idea for understanding a single exposure). deficiencies. several criticisms of the multiple eliminate the interference and improve performance memory system concept have been published of the behavior produced by the intact system. responses. 1996. requires further investigation on two levels. The theory suggests that ferent learning rates. or no influence on behavior. tional information reaches and flows through each the outputs of both systems might produce the same system and exactly how this information changes the behavior. . Finally. tentatively defined by the triple dissociation expe- The second factor influencing coherence is prac. This may become increasingly predictable.1 Some Outstanding Issues that support for the hypothesis is lacking in these Numerous issues requiring further investigation are species. produces different output. tice. riment. A major question is the degree of genuine func- of the situation. 1996). but the caudate system requires the functional contributions to memory of synaptic and numerous repeated exposures even to represent other changes between and within neurons require situations that correspond to its S-R specialization. coherent representations in two systems. tional levels of analysis. Within each sys- Because all systems receive all information. which in turn changes system can apparently acquire information about a how they process the information that flows through situation with very little experience (sometimes with them. Experiments directly investigating the suggested by the multiple parallel memory systems multiple memory systems hypothesis in monkeys theory. Less 2. Specifically. their elements: stimuli.

Martone M. Neurosci. Biobehav. Behav. J. Cohen NJ and Eichenbaum H (1991) The theory that wouldn’t Neuropsychologia 4: 197–220. Univ. Alzheimer Dis. 70: 314–327. and Robbins TW (1989) Interactions Cook D and Kesner RP (1988) Caudate nucleus and between the amygdala and ventral striatum in stimulus- memory for egocentric localization. Rockland C. 396–408. Neurosci. Neurobiol. Acta Biol. (2003) Cerebellar Chai S-C and White NM (2004) Effects of fimbria-fornix. Neurosci. J. Neurosci. Disord. J. pp. Science 210: 207–209. Rev. 3547–3554. awareness. et al. Maurer R. New York: Dover. 90: between proximal locations. Exp. Chozick BS (1983) The behavioral effects of lesions of the (Warsz. New York: McGraw-Hill. Process. striatal-dependent learning. Int. Neural Biol. 45–79. 1877–1886. Calif. In: Mackintosh N Burgdorf J and Panksepp J (2006) The neurobiology of positive (ed. Washington: American Psychological Dickinson A and Dawson GR (1989) Incentive learning and the Association. Psychol. Damasio AR (1995) Double dissociation of conditioning and Davis M (1997) Neurobiology of fear responses: The role of the declarative knowledge relative to the amygdala and amygdala. 30: conditioning. Everitt BJ. Science 173: Psychol. Untersuchungen zur Cohen NJ and Squire LR (1980) Preserved learning and experimentellen psychologie. 13: 1071–1077. Blodgett HC (1929) The effect of the introduction of reward Devan BD. and Cohen NJ (1992) The hippocampus hippocampus and dorsal striatum. translated 1964]. White B. J. Science 269: 1115–1118. Picconi B. Learn. Comp. Morgan CT. J. die: A critical look at the spatial mapping theory of Ebbinghaus H (1885) Memory: A Contribution to Experimental hippocampal function. 41: 99–112. Mem. Dichgans J. Behav.42 Multiple Memory Systems in the Brain: Cooperation and Competition References Corkin S (1968) Acqusition of motor skill after bilateral medial temporal lobe excision. and Wolfe J (1986) Dickinson A and Balleine BW (1994) Motivational control of Clinical validators: Comparisons of demented and amnesic goal-directed action. Behav. Neuroscience 30: 63–75. Polkey CE. Lutzenberger W. Academic Press. Behav. Gubellini P. Mem. Mem. Neurosci. Dickinson A (1994) Instrumental conditioning. J. and Birbaumer N (1993) Classical conditioning Bagshaw M and Benzies S (1968) Multiple measures of the after cerebellar lesions in humans. Dickinson A. Neurobiol. 118: 770–784. responses evoked by nociceptive leg withdrawal reflex as hippocampus and amygdala lesions on discrimination revealed by event-related fMRI. and Petri HL (1996) Dissociation of Bergmann G and Spence K (1964) Operationism and theory in hippocampal and striatal contributions to spatial navigation psychology. Hillsdale. Exp. Divac I. Jacobson TK. J. Neural Biol. 1103–1107. Rev. 105: Assoc. Psychol.) (trans. motivational control of instrumental performance. Davis DM. 48: 1–14. and Szwarcbart MK (1967) Behavioral Christian KM and Thompson RF (2003) Neural substrates of effects of selective ablation of the caudate nucleus. and Squire LR (2002) Classical CS-UR relation.) Animal Learning and Cognition. Neurosci. Tranel D. Anderson JR and Bower GH (1979) Human Associative Daum I. Exp. and brain systems. learning in the simple T-maze. Brain Res. and Calabresi P the instrumental training contingency on susceptibility to (2001) Dopaminergic control of synaptic plasticity in the reinforcer devaluation. Dimitrova A. and Nissen HW (1960) The Differential effects of estrogen on hippocampal. 30: 173–187. Neurosci. 23: study. 6: 524–531. Hebb DO. and Adams CD (1983) The effect of Centonze D. Exp. 115: 849–857. 49: reward associations: Studies using a second-order schedule 332–343. 10: 427–455. Hippocampus 1: 265–268. eyeblink conditioning: Acquisition and retention. 35B. 100: 5–14. 66: 305–323. 199: 201–209. New York: emotions. Aliakbari S. Q. 1: 5–8. Hilgard ER retention of pattern-analyzing skill in amnesia: Dissociation (ed. Brightwell JJ. Anim. Otto T. Kolb FP. Bernardi G. and White NM (1999) Effects of upon the maze performance of rats. of sexual reinforcement. Eichenbaum H and Cohen NJ (2001) From Conditioning to Colombo PJ. Donahoe JW and Vegas R (2004) Pavlovian conditioning: The Clark RE. Animal Learning and Behavior 22: patients. J. Etienne AS. et al. medial and lateral caudate-putamen lesions on place. 19: 143–160. Psychol. in conditional discrimination. aged rats. Granholm E. [Über das gedächtnis. Psychology. Psychol. Memory. NJ: Lawrence Earlbaum.and Psychol. 9: 382–402. and Séguinot V (1996) Path integration in Colombo PJ and Gallagher M (1998) Individual differences in mammals and its interaction with visual landmarks. Mem. Clin. Learn. Sci. J. McDonald RJ. Chang Q and Gold PE (2003) Switching memory systems during Dimitrova A. 37: 412–422. Manns JR. 107: orienting reaction and their dissociation after 748–756. 17–33. Exp. Cabanac M (1971) Physiological role of pleasure. 20: 175–187. Elles HG. Schugens MM. Douglas RJ and Pribram KH (1966) Learning and limbic lesions. Behav. cue-guided behaviors in the water maze: Relation to Blodgett HC and McCutchan K (1947) Place versus response thigmotaxis. Kolb FP. J. Neurol. Neurophysiol. cAMP response element-binding protein and c-Fos in the Eichenbaum H. and Countryman RA (2003) Conscious Recollection: Memory Systems in the Brain Cognitive strategy-specific increases in phosphorylated Oxford: Oxford University Press. Physiol. Neuropsychologia 6: 255–265. and 132–137. Neurophysiol.) Handbook of Clinical Assessment of 1–18. Devan BD. Psychol. (2004) Cerebellar learning: Changes in patterns of brain acetylcholine release activation during leg withdrawal reflex conditioning: An fMRI in the hippocampus and striatum in rats. In: Poon L (ed. 337–352. corpus striatum: A review. in the water maze. Neurosci. of knowing how and knowing what. spatial memory and striatal ChAT activity among young and Biol. Alzheimer A (1987) About a peculiar disease of the cerebral Daum I. Publ. Ackermann H. 57: 2–36. Exp. hippocampus in humans. Nicholas DJ. and Mizumori SJ (2005) Beach FH. 84: Bechara A. Divac I (1968) Functions of the caudate nucleus. Neurosci. translated by Jarvik L and Greenson H). B. Elles HG. Exp. 63: 183–190. Butters N. Eur. Damasio H. Trends Cogn. and Gray JA (1991) Classical cortex (Uber eine eingeartige Erkrankung der Hirnrinde. 4: 113–134. J. Learn. 35–51. Behav. dorsal striatum. . Goad EH. Adolphs R. Q. Learn.).) 28: 107–120. Channon S. 23: – What does it do? Behav. amygdalectomy in monkeys. Cador M. pp. 3001–3005. conditioning after temporal lobe lesions in man: Impairment 1907. Neurobiol. Rosvold HE.and Neuropsychology of Lashley. Memory. J. Neuropsychiatry Clin.

and Parkinson’s disease patients. Sci. Sydney: Academic Press. task. 463–475. Neurosci. Neurosci. Behav. Hull CL (1943) Principles of Behavior. Huston JP. 8: Semin. Neurosci. memory. et al. 9: 582–587. J. 1051–1054. R. Spatial learning induces presynaptic structural remodeling in Gaffan D (2002) Against memory systems. Physiol. Hirsh R (1980) The hippocampus. Psychiatry 158: 21. Paris: F. Neuroscience 136: 387–395. Fanselow MS (1990) Factors governing one-trial contextual Gross CG. Behav. striatum impair acquisition of a simple membranes. Mueller CC. Behav. Huntington’s. Hirsh R (1974) The hippocampus and contextual retrieval of Frankland PW. Neurosci. and information from memory: A theory. while well-worn route and the path less traveled: Distinct neural sparing conditioned place preference learning. 119: 1563–1571. Hirsh R. Gale GD. New York: Appleton- Gaskin S and White NM (2006) Cooperation and competition Century-Crofts. Behav. Neurosci. Hsiao S and Isaacson RL (1971) Learning of food and water Garcia J and Koelling RA (1966) Relationship of cue to positions by hippocampus damaged rats. Trans. (2005) Impairment basolateral amygdala in the storage of fear memories across in long-term retention but not short-term performance on a the adult lifetime of rats. and dorsal thalamic lesions in rats: Fanselow MS and Gale GD (2003) The amygdala. and Redgrave P (2004) stimulus-response learning: Lesions of the dorsolateral. Psychon. 84: 159–167. J. Heindel WC. 37–86. Philos. 1: Forebrain stimulus-response dependent conditional discrimination and Midbrain. Am. Walicke PA. (2006) Trace eyeblink conditioned eyelid reactions. Godsil BP. conditional operations and Behav. et al. Hajjar K. Anim. Acad. Ferbinteanu J and Shapiro ML (2003) Prospective and Hilgard ER and Bower GH (1966) Theories of Learning. Behav. Hartley T. Neurobiol. Salmon DP. 23: 743–760. O’Brien A. Biol. conditioning in human subjects with cerebellar lesions. Behav. stimulus-response learning: Lesions of the dorsolateral. conditioning. Disord. 27: 453–459. Rev. hippocampal. Neurosci. Franz SI (1902) On the functions of the cerebrum: I. cognition. Behav. Hippocampus 16: 560–570. Anagnostaras SG. 109: 663–668. but Computational models of the basal ganglia: From robots to not dorsomedial. and Robbins TW (2001) The Gill KM and Mizumori SJ (2006) Context-dependent modulation neuropsychological basis of addictive behaviour. 12: Silva AJ (1998) The dorsal hippocampus is essential for 421–444. Parkinsonism Relat. and Butters N Fendt M and Fanselow MS (1999) The neuroanatomical and (1989) Neuropsychological evidence for multiple implicit neurochemical basis of conditioned fear. N. Neuropsychologia 3: 53–68. striatum impair acquisition of a stimulus. and Mosseri A (1978a) Hippocampal mossy lobes in relation to the production and retention of simple fibers motivation states and contextual retrieval. Learn. Everitt BJ. McDonald RJ. Graybiel AM (2004) Network-level neuroplasticity in cortico- striatal interactions underlying reward-related processes. 24: 3810–3815. 37: 877–888. B Biol. retrospective memory coding in the hippocampus. and Disterhoft JF (1995) Intact delay-eyeblink motivational states as controllers of behavior: A study classical conditioning in amnesia. 62: 68–79. Physiol. Behav. Neurosci. conditioned place preference: Further evidence of limbic. Exp. Schoell. Trends Neurosci. and Burgess N (2003) The response-based instrumental discrimination task. McGlinchey-Berroth R. Brain Res. The frontal Hirsh R. 18: 264–270. Neurosci. reinforcement on the acquisition and extinction of Gerwig M. water maze reversal task following hippocampal or Gall FJ and Spurzheim J (1819) Anatomie et Physiologie du mediodorsal striatal N-methyl-D-aspartate receptor Systèm Nerveux en Générale. Hartley T and Burgess N (2005) Complementary memory Featherstone RE and McDonald RJ (2004b) Dorsal striatum and systems: Competition. striatum. memory systems: A comparison of Alzheimer’s. Exp. but Trends Neurosci. cooperation and compensation. (2004) Role of the Holahan MR. and Gillman R (1978b) Fornix fibers and Gluck MA. Soc. 8: 33–38. Dickinson A. 36: 129–138. not dorsomedial. by D(1) receptors: Differential effects in hippocampus and Brain Res. basal ganglia pathways. action and the corpus striatum: Hirsh R and Segal M (1971) Complete transection of the fornix Current developments in the memory-habit distinction. Gaffan D (1996) Memory. bases of route following and wayfinding in humans. Morris KA. Rekart JL. Mem. 985: 125–134. Emrich SM. 170: 7–21. et du Cerveau en Particulier. discrimination task. context discrimination but not for contextual conditioning. and Cohen SM (1965) Caudate. 25: 141–158. Alheid GF. Y. Shults CW. Rev. Brain Res. Vol. Neuron Neuroscience 124: 23–31. Prescott TJ. and Mondadori C (1977) Memory Brain Res. 10: Neuroscience 42: 1–18. et al. 22: 819–827. Featherstone RE and McDonald RJ (2005a) Lesions of the Hatfield T and Gallagher M (1995) Taste-potentiated odor dorsolateral or dorsomedial striatum impair performance of a conditioning: Impairment produced by infusion of an previously acquired simple discrimination task.) The Rat Nervous System. Exp. and Zaborsky L (1985) Basal ganglia. pp. Spiers HJ. N-methyl-D-aspartate antagonist into basolateral amygdala. Sci. Gaffan D (2001) What is a memory system? Horel’s critique Holahan MR. 3rd edn. and reversal of position habit in the rat. Psychol. Carrillo MC. 127: 5–11. Ann. Neuron New York: Appleton-Century-Crofts. Leber B. Physiol. Neurol. cortical. Psychol. Biobehav. Featherstone RE and McDonald RJ (2004a) Dorsal striatum and Gurney K. J. 120: 377–392. Gabrieli JD. 28: 169–170. and Routtenberg A (2006) revisited. 357: 1111–1121. Hippocampus 16: 577–585. and Alternation and Hebb-Williams maze performance. J. dorsolateral striatum impair the acquisition of a simplified In: Paxinos G (ed. Featherstone RE and McDonald RJ (2005b) Lesions of the Heimer L. Taverna FA. Wickens JR. Physiol. 6: consequence in avoidance learning. Filipkowski RK. Behav. facilitation by posttrial hypothalamic stimulation and other . the hippocampal mossy fiber system of two rat strains. Cestari V. Cermak LS. Multiple Memory Systems in the Brain: Cooperation and Competition 43 Everitt BJ. 150: 15–23. Sandoval J. Neurosci. blockade. Biol. Learn. Lond. Behav. Chorover SL. 7: 81–83. 40: 1227–1239. between the dorsal hippocampus and lateral amygdala in Humphreys LG (1939) The effect of random alternation of spatial discrimination learning. Haerter K. 8: 175–182. Maguire EA. fear. Behav. 109: stimulated by the contextual retrieval theory. 293–296. 112: 863–874. J. and Robbins TW (1991) The Graybiel AM (2001) Neural networks: Neural systems V: Basal basolateral amygdala-ventral striatal system and ganglia. Brain Res. 8: 1–22. 439–450. sensory-motor habits. Holt L. Sci.

a path integration system. Deciphering the hippocampal polyglot: The hippocampus as Kirsch P. Erichsen M. 30: 188–202. 36: paradigm: A MEG study. Neurol. striatum. Exp. 58: 69–79. Nerv. and Bohbot VD transection of the fornix. and Sax DS (1984) space in mice requires attention. Maren S and Quirk GJ (2004) Neuronal signalling of fear Kentros CG. 30: 250–264. Neuropsychologia 6: 215–234. Neuropsychologia Conditioned preference in humans: A novel experimental 19: 201–225. and Vaitl D (2003) Cerebellar and hippocampal activation Milner B (1958) Psychological defects produced by temporal during eyeblink conditioning depends on the experimental lobe excision. amnesia and dementia. and Di SG (2004) Trace fear conditioning: 362–377. Hippocampal and caudate metabolic activity associated Rev. 70: 285–295. Neurosci. siecle innovators in neuropsychiatry. Muller RU. 42–48. Comp. and Mahut H (2005) Developmental study Maschke M. 65–77. Neurosci. Squire LR. Petrides M. Learn. Corkin S. White NM. 6979–6985. Paris: Virin. Mem. Burgess N. Cybern. (2000) Impaired preference conditioning after anterior temporal Mahut H. and Zola-Morgan S (1981) Retention deficits lobe resection in humans. 767–793. Kennedy PJ and Shapiro ML (2004) Retrieving memories via Maren S (2005) Synaptic mechanisms of associative memory in internal context requires the hippocampus. Neurol. Heinzmann M. Kudrimoti HS. Am. response preservation in the rat. Rev. Ital. 80: 425–446. Weinberger NM (eds. Miranda R. Science 280: 921–924. 11: 43–58. Ungerleider LG. 87: Cognitive strategies dependent on the hippocampus and 356–372. Neurosci. What can the hippocampal representation of environmental Iaria G. J. Neuropsychologia 10: 65–74. McDonald RJ and White NM (1994) Parallel information Kimble DP (1968) Hippocampus and internal inhibition. 20: 2649–2656. J. striatum and amygdala function on the same stimulus- Kim JJ and Fanselow MS (1992) Modality-specific retrograde response habit task. Burgess N. 23: 3071–3080. and Gluck MA (1994) Probabilistic lesions on recent memory. Ment. 6: 414–417. and O’Keefe J (2002) 1: 143–150. New York: Guilford Press. novel Y-maze procedure. Biol. and Arias JL (2006) Landauer TK (1969) Reinforcement as consolidation. Assoc. Butters N. Payne M. and Teuber H-L (1968) Further analysis of J. pp. Assoc.Rev. 24: the amygdala. Neurosci. 7: Maguire EA. McDonald RJ and White NM (1993) A triple dissociation of Kim JJ and Jung MW (2006) Neural circuits and mechanisms memory systems: Hippocampus. Pike B. Am. Neural Biol. Owen AM. Soc. Rep. Neurosci. Frackowiak RS. Biol. Schienle A. Psychol. McDonald RJ and Hong NS (2004) A dissociation of dorso-lateral Neurosci. Jones B and Mishkin M (1972) Limbic lesions and the problem (1987) edn. Kirsch M. In: Lynch G. Hist. Res. memory systems involving dorsal striatum and Kimble DP and Kimble RJ (1965) Hippocampectomy and hippocampus. Behav. Donnett JG. and Squire LR (1996) A neostriatal study of H. Maine de Biran P (1804) Mémoires sur l’influence de l’habitude. 119: 651–661. Neuroreport 13: mulatta): II. J. Learn. Dagher A. after combined amygdalo-hippocampal and selective Johnsrude IS. Cacucci F. (2002) Limb flexion of the hippocampal formation in rhesus monkeys (Macaca reflex-related areas in human cerebellum. 24 Dissociations between skill learning and verbal recognition in Abs No 316. Assoc. Early ablations do not spare the capacity to 2325–2330. in the control of place cells and head direction cells. Trans. 120: Lashley KS (1950) In search of the engram. Behav. habit learning system in humans. and Everitt BJ (2006) human navigation network. Neural Plast. Neurosci. Owen AM. and Bachevalier J (1984) Memories and LeDoux JE (1993) Emotional memory systems in the brain. caudate nucleus in human navigation: Variability and change Ljungberg L (1992) Carl Wernicke and Sergei Korsakoff: Fin de with practice. classification learning in amnesia. Motiv. Dis. and O’Keefe J (1998) Knowing where and getting there: A Ito R. Neurosci. J. Abstr. Barnes CA. Mishkin M. Neurosci. Exp. McNaughton BL. and spatial vision: Two cortical pathways. (1996) Physiol. approach. Zhao WV. Neuron 47: 783–786.M. Their effects upon hypotheses and frustration. 244–257. processing in the water maze: Evidence for independent Bull. San Francisco: Albion. Moss M. 36: Marchand AR. Res.) Neurobiology of Human Memory and Leutgeb S. Psychol. Neuroscience 124: 507–513. and Bohbot VD (2003) geometry tell us about Hebbian learning? Biol. Psychol. Blanco E. Malamut B. Hartley T. Becker J. ER (2003) Stabilization of the hippocampal representation of Martone M. and White NM (1999) hippocampal resections in the monkey. Rubio S. Gerrard JL. Frith CD. Mangels JA. Rehbein L. Eur. McNaughton BL. and Macko KA (1983) Object vision episodic memory in hippocampal neuronal ensembles. Neurosci. and McNaughton BL (1998) Milner B (1959) The memory defect in bilateral hippocampal Interactions between idiothetic cues and external landmarks lesions. Agnihotri NT. Luck D. Science 256: 675–677. and Behav. Kirkby RJ (1969) Caudate lesions and perseverative behavior. McGaugh JL and Herz MJ (1972) Memory Consolidation. et al. of stimulus-reinforcement associations. 1: 106–120. J. A role for context? Arch. 5: 844–852. 1: Isaacson RL and Kimble DP (1972) Lesions of the limbic system: 23–27. retrieve conditional object-object associations. Leutgeb JK. Brain Res. Exp. Learning. J. J. 4: 451–454. Neuropsychologia 9: cue and place conditioning based on path integration in a 409–424. Zhao WV. 23: 5945–5952. Lever C. Psychiatr. McGaugh JL. Behav. Achenbach C. Biol. 10: 291–301. 142: 251–263. Rev. Publ. Symp. with different navigational strategies. Biol. the hippocampal amnesic syndrome: 14-year follow-up Knowlton BJ. 60: 474–476. Nat. Knierim JJ. 76: 82–96. Arch. 641–650. Hawkins RD. . McNaughton BL. Soc. Neurol. Biobehav. amygdala and dorsal involved in Pavlovian fear conditioning: A critical review. 41: 965–970. Science 309: 619–623. et al. and Kandel memory. Physiol. Behav. 107: 3–22. Barnes CA. Begega A. Mahut H (1972) A selective spatial deficit in monkeys after Johnsrude IS. Behav. Moser EI. Neurophysiol. 199: 173–185. Drepper J. Milner B and Penfield W (1955) The effect of hippocampal Knowlton BJ. Science 273: 1399–1402. Neurosci. and Moser MB (2005) Independent codes for spatial and Mishkin M. Milner B. Behav. Trends Neurosci.15.44 Multiple Memory Systems in the Brain: Cooperation and Competition reinforcers: A central theory of reinforcement. Killiany R. amnesia of fear. Biobehav. 61: 260–270. Res. Selective excitotoxic lesions of the hippocampus and Mahut H (1971) Spatial and object reversal learning in monkeys basolateral amygdala have dissociable effects on appetitive with partial temporal lobe ablations. habits: Two neural systems. 4: 454–482. Robbins TW. Psychiatr.

and Disterhoft JF (1990) Pavlov IP (1927) Conditioned Reflexes Oxford: Oxford Hippocampectomy disrupts trace eye-blink conditioning in University Press. and Kill D (1975) Fornix lesions animal and human brain studies. Neurol. Animal Behavior. and Gold PE (1996) Hippocampal past: Spatial memory in the rat. 9: engram: The lateral amygdala is an essential locus of fear 1465–1472. In: Ellen P and Thinus-Blanc C (eds. Learn. rabbits. Physiol. Neurosci.) functions of the Basal Ganglia. Mizumori SJ and Leutgeb S (2001) Directing place representation Packard MG and Teather LA (1997) Double dissociation of in the hippocampus. Opin. O’Keefe J. USA 93: Olton DS. Neurol. Cahill LF. Fyhn M. USA 91: 8477–8481. Gill KM. Exp. Neurobiol. and Black AH (1975) Slam on the brakes: A conditioning. of fornix and caudate nucleus lesions on two radial maze Schafe GE. 11: 752–756. 40: 243–258. Exp. pp. 12: 347–363. J. Preliminary evidence from unit activity in the Interactive memory systems in the human brain. Behav. affects expression of place and response learning. Arch. intracerebral injection of dopamine agonists. Exp. 3: 230–251. Acad. Neurobiol. Mem. direction. Neurosci. 25: 10010–10014. 25: Cognitive Processes and Spatial Orientation in Animal and 563–593. Neurobiol. J. Man. Nature freely-moving rat. 152–166. discrimination. Yeshenko O. 20: 517–523. memory storage. 14: 381–385. Olton DS and Samuelson RJ (1976) Remembrances of places Ragozzino ME. connections and spatial discrimination. Behav.. multiple memory system hypothesis. 107: 663–694. Mem. Biol. 105: 295–306. Rawlins JNP. Packard MG and Teather LA (1998) Amygdala modulation of 82: 278–298. and LeDoux JE (1997) Organization model of hippocampal function. 124–130. retention in monkeys after amygdala-hippocampus Ribot T (1883) The Diseases of Memory. Poldrack RA. Brain Res. 82: 324–332. Rev. memory systems. Sargolini F. and velocity in Packard MG. Neurosci. 341–373. Sci. Psychol. Science 312: 758–762. Moyer JR Jr.-response’’ question. Neurobiol. Motiv. of intra-amygdaloid circuitries in the rat: An emerging O’Keefe J (1976) Place units in the hippocampus of freely framework for understanding functions of the amygdala. Psychol. Savander V. J. O’Keefe J and Dostrovsky J (1971) The hippocampus as a Poldrack RA. 111: 543–551.dependent memory processes. Physiol. Proc. Nature 297: 681–683. Keightley S. water maze tasks: Further evidence for multiple memory Mitchell JA and Hall G (1988) Caudate-putamen lesions in the systems. Netherlands: Martinus fornix and caudate nucleus lesions on acquisition of two Nijhoff Publishing. Pare-Blagoev EJ. and sodium A possible substrate for procedural knowledge. English translation of resection. Trends Neurosci. Multiple Memory Systems in the Brain: Cooperation and Competition 45 Mitchell JA and Hall G (1987) Basal ganglia. Packard MG and White NM (1991) Dissociation of hippocampal Morris RGM. Sage JR and Knowlton BJ (2000) Effects of US devaluation on Packard MG. Rev.) Cognitive Processes in Neuropsychology 13: 564–574. Clark J. multiple memory systems: Converging evidence from O’Keefe J. Natl. pp. Physiol. multiple memory systems: Hippocampus and caudate- Morris RGM (1981) Spatial localization does not require the putamen. B Comp. et al. 12: 239–261. and White NM (1989) Differential effects entorhinal cortex. amygdala and hippocampus to cued and contextual fear Nadel L. Neurosci. Comp. 65: 66–72. Neuropsychologia 17: 669–682. 29: 61–72. the ‘‘place-vs. Dordrecht. Neurol. and O’Keefe J (1982) and caudate nucleus memory systems by post-training Place navigation impaired in rats with hippocampal lesions. Mem. Process. Neuropsychologia 41: selectively abolish place learning in the rat. Poldrack RA and Rodriguez P (2004) How do memory systems Olton DS and Papas BC (1979) Spatial memory and interact? Evidence from human classification learning. Sci. Prabhakaran V. Can. Peinado-Manzano MA (1988) Effects of biateral lesions of the Muller RU. Behav. 34: 171–175. Learn. 87: 622–643. 106: 439–446. Brunner. Psychol. critique of Altman. learned taste aversions. Nadel L (1991) The hippocampus and space revisited. Neurol. and Gage FH (1978) Hippocampal 4693–4698. 104: 243–252. representation of position. Learn. Rev. Nachman M and Ashe JH (1974) Effects of basolateral amygdala Phillips AG and Carr GD (1987) Cognition and the basal ganglia: lesions on neophobia. Hirsh R. Augmentation by glucose. Deyo RA. Learn. rat may impair or potentiate maze learning depending Packard MG and McGaugh JL (1996) Inactivation of upon availability of stimulus cues and relevance of hippocampus or caudate nucleus with lidocaine differentially response cues. moving rats. Neurosci. Natl. 51: 78–109. and Davis DM (2004) Parallel posttraining intracerebral injections of 2-amino-5- processing across neural systems: Implications for a phosphonopentanoic acid. nucleus. Q. (2001) spatial map. Annu. and McGaugh JL (1994) Amygdala win-stay and win-shift radial maze performance in rats. 114: 295–306. Behav. 106: 274–285. and LeDoux JE (2005) Tracking the fear tasks: Evidence for multiple memory systems. (2006) Conjunctive Sci. 48: 245–251. . Acad. Curr. J. and Bayer’s response-inhibition Pitkänen A. 139: Restle F (1957) Discrimination of cues in mazes: A resolution of 295–308. modulation of hippocampal-dependent and caudate Behav. Neurosci. Brain Res. Walker JA. Neurosci. Psychol. Neurosci. Fowler H. hippocampal function. 1: Experimental Animal Psychology and Packard MG and McGaugh JL (1992) Double dissociation of Ethology. SH. Stead M. et al. instrumental and Packard MG and Knowlton BJ (2002) Learning and memory spatial learning. Garrud P. In: Hulse Striatal activation during acquisition of a cognitive skill. Psychol. Proc. 2: 97–116. Unick KE. 414: 546–550. 69: 163–203. O’Keefe J and Nadel L (1978) The Hippocampus as a Cognitive Poldrack RA and Packard MG (2003) Competition among Map Oxford: Oxford University Press. and Rasmussen T (1960) Learning and 217–228. Brain Res. Les Maladies de la Mémoire (1881) New York: Humboldt Packard MG and Cahill LF (2001) Affective modulation of multiple Library of Popular Science Literature. Milner B. Neurosci. and Gabrieli JD (1999) Olton DS (1978) Characteristics of spatial memory. and Pach J (1996) The hippocampus as a central and lateral amygdala on free operant successive cognitive graph. Learn. J. Doyere V. and Honig WK (eds. 64: 957. J. Gen. acetylcholine release during memory testing in rats: Behav. Nadel L. Hillsdale. Neurosci. Orbach J. Behav. appetite in rats. Anim. Exp. hippocampal and dorsal-striatal memory systems by Mizumori SJ. NJ: Erlbaum. Behav. J. Neurobiol. presence of local cues. Hafting T. Behav. Mem. Vol. Phillips RG and LeDoux JE (1992) Differential contribution of Hippocampus 1: 221–229. 14: 151–162. Seger CA.

USA 93: behavioral correlates of hippocampal neuronal activity. MA: MIT Press. Behav. Neurobiol. and White NM and Milner PM (1992) The psychobiology of McNaughton BL (2005) Self-motion and the hippocampal reinforcers. Budapest: Akademiai Kiado. Rev. Mem. Brain 127: 851–859. 348: 249–260. 459–463. pp. Biobehav. White NM (1989b) Reward or reinforcement: What’s the Neural Biol. Behav. Psychol. Brain Res. and Balleine BW (2004) Lesions of Tolman EC and Honzik CH (1930) Introduction and removal of dorsolateral striatum preserve outcome expectancy but reward and maze performance in rats. 56: 144–155. 36: 221–229. Smith DE. 55: 225–256. and Kalish D (1946) Studies in spatial Young PT and Christensen KR (1962) Algebraic summation of learning II place learning versus response learning. Psychol. J. Krause M. Poldrack RA (2004) Cortico-striatal contributions to Weiskrantz L (1956) Behavioral changes associated with feedback-based learning: Converging data from ablation of the amygdaloid complex in monkeys.) Electrical Activity of the Archicortex. Life Sci. Brain Res. Physiol. Rev. medley of movement. Lavond DG. 4: 257–275. 9: 2737–2763. neuroimaging and neuropsychology. Neurosci. Psychol. 11: Tolman EC (1949) There is more than one kind of learning. Behav. Mem. space and memory.46 Multiple Memory Systems in the Brain: Cooperation and Competition Scoville WB and Milner B (1957) Loss of recent memory after (eds. 7: 228–234. is affected by two independent reinforcement processes. Neurosci. 25: 8085–8096. Bull. temporal factors in spatial learning. Learn. Neurosci. Knowlton BJ. lateral cerebellar cortex aspirations. Gothard KM. White NM and Ouellet M-C (1997) Roles of movement and Thompson RF (2005) In search of memory traces. Behav. Packard MG. Biobehav. Terrazas A. 13: 119–122. Motiv. abolishes conditioned cue preference: Evidence for Tolman EC (1932) Purposive Behavior in Animals and Men. 8(4): 473–479. Rev. Behav. Behav. In: Ingle DJ. Tolman EC. and Eichenbaum H (1989) Spatial and and localization of a memory. Myers CE. 552–557. McKenna JE. demonstration of latent spatial learning in the rat. pp. Thompson WG. 55: water maze tasks. Woodruff-Pak DS. J. 19: 181–189. Learn. Skinner BF (1938) The Behavior of Organisms. and Hicks LH (1980) Effects of Wise SP (1996) The role of the basal ganglia in procedural caudate and cortical lesions on place and response learning memory. 38: 322–347. 102: 586–590. and Black AH (1980) Stimulus control of White NM and Carr GD (1985) The conditioned place preference spatial behavior on the eight-arm maze in rats. Psychol. Brain Res.) Analysis of Visual Behavior. striatal matrix and patches: Mediation of S-R memory and Sutherland RJ and Linggard RC (1982) Being there: A novel reward. Yin HH. stimulus-stimulus associations. New York: Macmillan. Mem. 56: 1–23. Guilford MO. 27: 403–406. Biochem. Warrington EK and Weiskrantz L (1982) Amnesia: A Sherry DF and Schacter DL (1987) The evolution of multiple disconnection syndrome? Neuropsychologia 20: 233–248. Proc. Comp. Yin HH and Knowlton BJ (2004) Contributions of striatal Rev. J. 501–510. Learn. path integration hypothesis. Univ. Curr. Pharmacol. Natl. Ellis-Behnke RG. Sage J. Goodale MA. Neurosci. Ritchie BF. Thorndike EL (1933) A proof of the law of effect. Lipa P. Annu. Rev. J. Eur. Rev. Monogr. Rev. and Maaswinkel H (1997) Vanderwolf CH (eds. Skelton RW (1988) Bilateral cerebellar lesions disrupt Whishaw IQ (1998) Place learning in hippocampal rats and the conditioned eyelid responses in unrestrained rats. Psychol. 13438–13444. in rats. Neurobiol. memory systems. Psychol. Augerinos G. Semin. Tolman EC (1948) Cognitive maps in rats and men. Sutherland RJ (1985) The navigating hippocampus: An individual Brain Res. Suzuki S. Science 77: Yin HH and Knowlton BJ (2002) Reinforcer devaluation 173–175. 174–177. Evidence from spontaneous exploration and spatial learning Hippocampus 7: 465–471. and Thompson RF (1985) Trace Thorndike EL (1898) Animal intelligence. Physiol. learning and expressing a spatial discrimination. subregions to place and response learning. Hippocampal lesions and path integration. Psychol. 13: 181–186. 20: 158–177. Exp. Learn. An experimental study conditioning: Abolished by cerebellar nuclear lesions but not of the associative processes in animals. New York: Whishaw IQ. Neurobiol. Neurosci. and Bazan White NM and McDonald RJ (1993) Acquisition of a spatial NG (2005) Differential induction of c-Jun and Fos-like proteins conditioned place preference is impaired by amygdala in rat hippocampus and dorsal striatum after training in two lesions and improved by fornix lesions. Can. 55: 332–336. In: Buzsaki G and Whishaw IQ. Comp. Rev. Neurol. difference? Neurosci. 43: 443–471. Psychol. Yamamoto T and Fujimoto Y (1991) Brain mechanisms of taste Thorndike EL (1911) Animal Intelligence. Opin. 255–279. tasks in light (allothetic) and dark (idiothetic) tests. 209–220. 49: 381–391. Sci. 84: 75–84. Paul CA. J. 23: 37–42. and Wallace DG (2001) Dead reckoning Appleton-Century-Crofts. Rev. Neurosci. Annu. systems. 11: 1–18. Thompson RF and Kim JJ (1996) Memory systems in the brain Wiener SI. 45: 1943–1957. disrupt habit formation in instrumental learning. J. 8: 39–46. Psychol. 68: 519–549. Physiol. Neurosci. 22: Neurosci. Sutherland RJ and Dyck RH (1984) Place navigation by rats in a White NM (1989a) A functional hypothesis concerning the swimming pool. Grossman S. Gluck MA. Cambridge. spatial metric. J. Thompson RF and Krupa DJ (1994) Organization of memory Winocur G and Mills JA (1969) Effects of caudate lesions on traces in the mammalian brain. hedonic processes. Gen. Teather LA. Barnes CA. Hippocampus 7: Psychol. Acad. 56: 144–155. Physiol. Publ. and Psychol. 94: 439–454. Comp. Psychol. Prog. Neurosurg. 36: 103–107. bilateral hippocampal lesions. (path integration) requires the hippocampal formation: Smith PF (1997) Vestibular-hippocampal interactions. Taube JS (1998) Head direction cells and the White NM and Gaskin S (2006) Dorsal hippocampus function in neurophysiological basis for a sense of direction. and Masnfield RJW Neuropsychologia 11: 271–284. Watson JB (1912) Psychology as the behaviorist views it. Psychol. J. 116: New York: Century. Psychiatry 20: 11–21. . Calif. Hines DJ. Psychol. aversion learning in the rat. 127: 49–69. J. 269–281. 2(4): 1–109. Zola SM and Mahut H (1973) Paradoxical facilitation of object Ungerleider LG and Mishkin M (1982) Two cortical visual reversal learning after transection of the fornix in monkeys. 549–586. 17: avoidance behavior in rats. Shohamy D. Annu.

2.03.03. Agster. Today.2 Overview of the Hippocampal however.1 The Postrhinal Cortex 52 3.4. RI.2.1.3 The Subiculum 62 3.03.3. The terms hippocampal region declarative memory system.03 Anatomy of the Hippocampus and the Declarative Memory System R. 3. D.1. 1994). terminology translates effectively from humans to structures be dissociated? How discrete are such animal models of human memory.03..2 The Parahippocampal Region 52 3. To what extent can the or hippocampal system have the advantage that the function of hippocampal and parahippocampal sub.1. Burwell and K.03. L. memory.03.1. Brown University.2 Overview of the Hippocampal System 47 3.03.2 The Perirhinal Cortex 53 3. but many questions remain concerning the et al.03. monkey.03. storage.03.03. the hippocampal memory system (Eichenbaum ory.1 Introduction functions? How do these structures interact to permit encoding.03. and rodent 51 3.1. together are necessary for human declarative mem.2.03.3. 47 .2.03.1 The Flow of Sensory Information through the Hippocampal System 64 3.1.2 The Hippocampus Proper 61 3.03. consolidation. including rodents. All rights reserved.03. 1991).1 A Short History of the Anatomy of representations of facts and events? What of Declarative Memory additional cognitive functions might be supported? A half century ago.2. In the past two decades. and retrieval 3.2.3 Cross-species comparisons: Human.03.1 Introduction 47 3.1 Nomenclature 47 3.2. The structures that are the topic of this chapter have it is generally accepted that the hippocampal variously been termed the medial temporal lobe formation and the nearby parahippocampal region memory system (Squire and Zola-Morgan.2 The Comparative Anatomy of the Hippocampal System 65 References 65 3.1.3.1 A Short History of the Anatomy of Declarative Memory 47 3. Providence.03.4.2. scientists study. 3.1 Nomenclature formation also play critical roles in memory.03. we have come full circle: It is now apparent System that the cortical areas surrounding the hippocampal 3.03. Scoville and Milner (1957) Understanding the structure and connectivity of described profound memory loss following bilateral these regions is necessary for generating and testing medial temporal lobe resection in the landmark sound hypotheses about the neurobiology of patient HM.3 Entorhinal Cortex 54 3.1 The Dentate Gyrus 59 3.03.2. ing memory and the brain narrowed in on the hippocampus as the critical structure for everyday memory for facts and events.3 The Hippocampal Formation 58 3.4 Conclusions 64 3.03.5 The Parasubiculum 58 3. 2004).4 Presubiculum 57 3. In the following years. 3.03. and the hippocampal region (Witter functional diversity of structures within the so-called and Amaral.1. USA ª 2008 Elsevier Ltd.2 Location of the hippocampal system structures 50 3.

DG. Shown for the human and monkey brain are unfolded layer IV maps of the perirhinal (PER) areas 35 and 36. the entorhinal cortex differentiate it from hippocampal pal formation and the parahippocampal region formation structures. monkey (middle). T. L. subiculum. The lower panel shows unfolded maps of the relevant cortical structures for the human brain (d). Figures adapted from Burwell RD. referred to as the hippocampal formation (Figure 3. or autobiographical memory. entorhinal. the parasubiculum (ParaS) is interposed between the entorhinal and POR/PH (arrows). overwhelmingly reciprocal. Because corticocortical connections in the brain are declarative memory. and Amaral DG (1995) The perirhinal and postrhinal cortices of the rat: A review of the neuroanatomical literature and comparison with findings from the monkey brain. Sub. hippocampus (Figure 1). and CA3). In the monkey and the rat brain. Sobreviela T. Witter MP. The rodent POR is the homolog of the primate PH (see text for details). The dentate gyrus. the included structures are connected by retrohippocampal region. which are components of the parahippocampal region. septal. S. and Gonsalo LM (1995) The human entorhinal cortex: A cytoarchitectonic analysis. The upper panel shows the relevant structures in lateral views of the human brain (a). and the lateral and medial entorhinal areas (LEA and MEA). uni- For research on memory using animal models. and the subiculum (Figure 2). the monkey brain (b). rs. in turn. lateral. the monkey brain (e). Shown for the rodent brain are unfolded surface maps of the PER areas 35 and 36. and subiculum are therefore collectively dentate gyrus. collateral sulcus. and entorhinal cortex (EC). ParaS. M. and rat (right). presubiculum. and the entorhinal cortex primary criterion for inclusion in the hippocampal is considered part of the parahippocampal region. and the rodent brain (f). formation. . Insausti AM. Finally. Comp. medial. ventral. provide a unidirectional input postrhinal cortex in the rodent brain is considered the (a) (b) (c) D/S V/ T POR HF EC R/ T C/ S PH HF PER PER PH HF EC PER EC (d) (e) (f) PER POR PER PH rs 36 ParaS PER 35 (35/36) 36 TF cs LEA 35 TH rs EC ParaS D/L MEA EC PH R C (TF/ TH) V/M Figure 1 Comparative views of the hippocampal system for the human (left). the entorhinal terms episodic or episodic-like memory may be most cortex projects to all portions of the hippocampal appropriate. and parasubicular cortices (Figure 1). rhinal sulcus. largely unilateral pathways beginning with the dentate postrhinal (or parahippocampal). includes the perirhinal. Abbreviations: cs. 355: 171–198. parahippocampal (PH) areas TF and TH. D. such a multisynaptic. formation is the trilaminar character of the structures. Tuñón T. The hippocampal formation includes the proper. temporal. also called the In addition. CA1 projects to the subiculum. In contrast. The structures shown in yellow). the hippocampus proper (fields CA1. dorsal.48 Anatomy of the Hippocampus and the Declarative Memory System These regions are thought to support a type of mem. The connectivity and laminar structure of The hippocampal system comprises the hippocam. The pre. Hippocampus 5: 390–408. PreS. ory that has been variously called episodic memory. and the rodent brain (c). presu- gyrus granule cell input to the CA3 (Figure 3). J. The pyramidal cells. The parahippocampal region. are not shown (but see Figure 2).and parasubiculum. CA2. V. the directional circuit is unique. Insausti R. the postrhinal cortex (POR). Neurol. CA3 bicular. dentate gyrus. parasubiculum. to the CA1.

dentate gyrus. ParaS. London: Oxford University Press. ventral. and rat brain (c) showing the cellular layers of the hippocampal formation structures: the dentate gyrus (green). (d) An unfolded map of the rodent hippocampal formation. M. The presubiculum (light orange) and parasubiculum (dark orange) are also shown at two rostrocaudal levels in panels (e) and (f). Rodent schematics adapted from Burwell RD and Witter MP (2002) Basic anatomy of the parahippocampal region in monkeys and rats. CA2 (purple). monkey brain (b). S. Organization and Role in Cognitive Functions. T. PreS. panels (e) and (f)).and parasubiculum. Also shown are perirhinal areas 36 and 35 (panels (c) and (e)). D. medial. CA3 (blue). lateral. dorsal. V. L. . (a) Coronal sections of the human brain (a). the lateral and medial entorhinal areas (LEA and MEA. parasubiculum. In: Witter MP and Wouterlood FG (eds. Sub. presubiculum. subiculum. temporal. Abbreviations: DG. panel (f)). CA1 (red).) The Parahippocampal Region. Anatomy of the Hippocampus and the Declarative Memory System 49 (a) (b) CA3 CA2 PreS CA2 CA3 DG ParaS DG SUB CA1 CA1 SUB PreS ParaS 1 mm 1 mm (c) (d) CA2 CA1 DG CA1 CA2 SUB CA3 DG Sub CA3 PreS ParaS D/ S L M 1 mm V/ T Figure 2 Comparative views of the hippocampal formation with the pre. and the postrhinal cortex (POR. septal. and the subiculum (yellow).

region areas TG and TGa and the caudal (parahip- layered cortices characterized by reciprocal connections. the verbal term for area 35 was perirhinal.. Koskinsas named the rostral perirhinal/ectorhinal whereas the parahippocampal region comprises six. and orange). which reported more detailed cytoarchitectonic analyses of these regions.and parasubiculum also receive direct nomenclature of perirhinal cortex as designating the cortical inputs.50 Anatomy of the Hippocampus and the Declarative Memory System POR There are also discrepancies in the terminology PER for the perirhinal cortex. area 36 cortex. postrhinal cortex in the rodent brain. The focus of this chapter is the rat hippocampal One practical problem in the comparative anat. PER. 1987) or 35a and 35b (Van Hoesen and entorhinal area. and the verbal term for area 36 was ectorhinal. The ventral or temporal postrhinal cortex). The entorhinal cortex roscience. Von Bonin and Bailey. still shaped like a C. but it is worth noting that there term parahippocampal. a cortical region in the medial the rodent brain (Figure 1(c)). the term perirhinal cortex was used to subiculum. MEA. the 3.e.1. included ParaS PreS the temporal pole in area 36 (von Economo. Thus. 1995. parahippocampal cortex in the primate brain. The term ectorhinal is no longer in brain. Abbreviations: DG. The hippocampus is relatively smaller of the entorhinal. PH. presubiculum. Note that the perirhinal and postrhinal cortices (PER and pocampal) region areas TF and TH.. In the latter nomenclature.03. portion of the region is associated with the fimbria- campal region (and is the homolog of the rodent fornix and the septal nuclei. The entorhinal cortex projects directly combined areas 35 and 36 for both the rodent and to all components of the hippocampal formation and primate brains. Burwell. The structure is cortices. the term is mate brains.. and parahippocampal in the primate brain (Figure 1(b)). PreS. perirhinal. Currently. POR. and all other parahippocampal structures system structures are reciprocal. it seems reasonable to adhere to the and the pre. though shallower and rotated . The upper panel of Figure 1 shows that used in two additional ways. 1947). The perirhinal and postrhinal cortices are the use except in rodent brain atlases. In the rodent brain. perirhinal Pandya. the parahippocampal portion of the structure is associated with the tem- gyrus is the fold or gyrus that contains a large portion poral cortices. In modern POR) have reciprocal connections with CA1 and the terminology. In Brodmann’s (1909) nomenclature. all other components of the parahippocampal region (Figure 3). hippocampal formation comprises three-layered structures Using a different nomenclature. EC (MEA and LEA) Although Brodmann defined area 36 as a very narrow strip of cortex that did not include the temporal pole. blue. 2001) adapted that nomenclature for use in homolog of the parahippocampal cortex in the primate the rodent brain. the term are surprising similarities and interesting differences has only one use (i. designate the combined areas 35 and 36 (Amaral entorhinal cortex. and they project comparative framework for experimental neu- heavily to entorhinal cortex. von Economo and characterized by largely unidirectional connections. ParaS. 2003b). the hippocampus is C-shaped and relatively larger in campal cortex. There was no designa- Figure 3 Simplified schematic of the hippocampal tion in Brodmann’s nomenclature for the caudally system. The dorsal or septal temporal lobe that is a component of the parahippo.2. lateral entorhinal area. within a major recipients of cortical afferents. In the human and monkey brains. 1929. medial et al. First is the parahippo.2 Location of the hippocampal subiculum. green. dentate gyrus. EC. Second. the most commonly used nomencla- ture for memory research in the primate brain is perirhinal cortex comprising areas 35 and 36. subiculum (SUB). LEA. The schematic includes the hippocampal formation (structures in yellow) and the parahippocampal region located region we now call the parahippocampal (structures in red. parasubiculum. The cortex (reviewed in Suzuki and Amaral. was termed TL. 1975). The entorhinal connections with CA1. system about which we have the most detailed ana- omy of these structures is the confusing use of the tomical information. which includes verbal and numeric DG CA3 terms. Burwell and colleagues (Burwell et al. SUB CA1 other classic studies. in the phrase ‘parahippocampal between these structures in the rodent and the pri- region’).

and parahippocampal cortices lies ventral to the ventral temporal area and dorsal to occupy the parahippocampal gyrus and the temporal the medial entorhinal cortex (Figure 2(f )). For example. and entorhinal and perirhinal cortices. Studies in rats. monkey. and the cortex are no longer visible. pole. The perirhinal cortex occupies the temporal pole and continues caudally. the rhinal sulcus is scale. parahippocampal/ In the monkey brain. and rat. relationships of the perirhinal.1.03. It extends along the entire connected with the pulvinar. is located caudal to the cortex. perirhinal and entorhinal cortices (Figure 1(b) and The homology of the rodent postrhinal cortex 1(e)). Area 36 is a larger strip of dysgranular cortex linearly with cortical surface area. in its caudal extent. rostrally. it is sometimes called. (Figure 1). postrhinal/parahippocampal TE of inferotemporal cortex. The layer (lamina dissecans) separating the deep and . cortex is considerably smaller than the perirhinal prising areas TH and TF. Anatomy of the Hippocampus and the Declarative Memory System 51 about 90 clockwise.3 Cross-species comparisons: lies in the medial portion of the anterior parahippo. suggesting that the region scales sulcus. Accordingly. or fissure as lateral posterior nucleus of the thalamus (LPO). The collateral sulcus and ventral temporal association cortex at caudal forms the lateral border of the parahippocampal levels. As in the monkey brain. In the primate brain. The monkey parahippocampal cortex. entorhinal grounds. the relative size differences are also interesting. Aside from the obvious differences in phalic than the rat brain. the surface area of the perirhinal cortex is pole (Figure 1(b) and 1(e)). Suzuki The relative size of the entorhinal cortex. the rostral hippocam. tex is strongly and reciprocally connected with the In the rodent brain. but in the primate brain. The rostrolateral bor. The rodents and primates. posterior parietal cortices. occurs when claustral cells underlying layer VI of the pus is associated with the temporal cortices. cortex also appears to scale linearly with brain size. for cross-species comparisons. the homolog of the lateral surface of the brain. are also similar. Area 36 lies dorsally adjacent the best terminology for the long axis of the hippo. the rat postrhinal cor- ally by visual area V4. All but the most rostral roughly twice that of the postrhinal/parahippocam- part of the lateral border of area 36 is shared with area pal cortex. and entorhinal cortices are similar (smoother) than the human brain and more gyrence. It entorhinal. and rodent campal gyrus and is bordered rostrally and laterally A comparative analysis of the unfolded maps of the by the perirhinal cortex. In the rat brain. human. in all three located lateral to area 35 and including the temporal species. small and is associated with only the most rostral secondary auditory cortex at midrostrocaudal levels. the entorhinal cortex by the insular cortex. Area 35 is a narrow band of agranular cortex that perirhinal cortex accounts for roughly 3% of the occupies the fundus and the lateral bank of the rhinal cortical surface area. the region receives substantial parahippocampal cortex is bordered rostrally by the input from visual associational. is the only prominent sulcus Likewise. monkey. such that the opening is pointing transition from insular cortex to the perirhinal cortex upward. Area TF is larger than area TH and is laterally with the primate parahippocampal cortex is based adjacent to area TH. the perirhinal cortex and provides the caudal border. The region is bordered rostrally In human. The dorsal border of area 36 is In the human brain. and rat brains shows that the spatial tex forms the caudal border of the perirhinal cortex.and presubiculum. its rostrally and septally to include the medial half of the surface area is more than three times that of the temporal pole on cytoarchitectonic and connectional perirhinal cortex. Subcortical connections medially by the para. retrosplenial. cortex. The entorhinal cortex 3. and humans suggest that the tex. monkeys. The perirhinal cortex caudal hippocampus is associated with the septal comprises two cytoarchitectonically distinct strips of nuclei. and caud. laterally by TE. The parahippocampal cor. the monkey parahippocampal cortex is (Figure 1(c) and 1(f)). The entorhinal cortex provides the ventral campus is the term septotemporal. Also. areas 35 and 36. which is less gyrencephalic postrhinal. portion of the perirhinal cortex. The postrhinal cortex is located caudal to gyrus (Figure 1(d)). border of area 35.2. Therefore. perirhinal. Insular cortex is classically is a six-layered cortex characterized by a cell sparse defined as the region overlying the claustrum. the rhinal sulcus is relatively formed by secondary somatosensory cortex. In agranular cortex medially adjacent to area TF. Human. der is formed by the superior temporal gyrus. Area TH is a thin strip of largely on the structural and connectional similarities. com. though it is quite shallow lateral preoptic area (LPO) in the rodent. to area 35. associated with the full extent of the perirhinal cor. the rhinal sulcus. however. monkey. and Amaral (2003a) extended the border of area 36 differs dramatically across species.

the and temporal to describe the long axis because these medial entorhinal area is more caudal and ventral. postrhinal cortex is mentioned. At this level. Usually the postrhinal cortex arises at campal formation is largest in the human brain and the caudal limit of the angular bundle when subicular smallest in the rodent brain. (Figure 1(c)). terms can be applied similarly across all species. though the structure is cells are no longer present in coronal sections relatively larger in the rodent brain. Moving caudally. and the lamina dissecans analysis. It should be noted that in the rat. In the human and monkey brains. the intrinsic connectivity of the entorhinal poral axis in the primate hippocampus is equivalent cortex appears to be organized into intrinsic bands of to the caudorostral axis. structurally. The ventral subregion is characterized by ectopic layer II cells (arrow) that appear near the rostral border with area 36. (a) Drawing of a coronal section of the rat brain at the level of the rostral limit of the postrhinal cortex. The septal subregion has a more differentiated laminar pattern. more highly differentiated as pus in the rodent brain. interconnectivity that form discrete associational networks.2 The Parahippocampal Region suggesting a functional topography. it cells at the perirhinal–postrhinal border near the is as if the hippocampal formation has swung down ventral border with the medial entorhinal cortex and forward. respectively). and the septotem- monkey. dorsal the human brain are also present in the monkey and to the rhinal sulcus and to the medial entorhinal area rat brains (Figure 2). (b) Nissl-stained coronal section showing the septal and temporal subregions of the POR (PORd and PORv. caudal hippocampus in the rhinal fissure and wraps obliquely around the (a) (b) PORd TEv IV V VI II/III ab I PORd VI PORv V II/III MEA I LEA PORv Figure 4 Location and photomicrograph of the postrhinal cortex (POR). In both rat and equivalent to dorsoventral axis. angular bundle. Similarly.03.03.1 The Postrhinal Cortex monkey. For ease of comparative compared to the lateral part. caudal to the perirhinal cortex. the postrhinal primate brain is comparable to ventral hippocampus cortex rises dorsally above the caudal extension of in the rodent brain. The whereas the lateral entorhinal area is more rostral septotemporal axis in the rodent hippocampus is and dorsal (Figure 1(c) and 1(f)). these bands of intrinsic connec- tivity project to different levels of the dentate gyrus. The postrhinal cortex is located near the caudal pole All hippocampal formation structures observed in of the rat brain. There is evi- dence that a similar topography exists for the 3. As previously (Figure 4(a)). arrow). The medial part of the entorhinal the primate brain is comparable to dorsal hippocam- area is. Layers are labeled I–VI. Abbreviations: ab. it is most efficient to use the terms septal is more evident. such that rostral hippocampus in the (Figure 4(b). the hippocampus is situated differently in characterized by the presence of ectopic layer II different species.2.52 Anatomy of the Hippocampus and the Declarative Memory System superficial layers. The absolute size of the hippo. . In the rat. 3.

The are returned. and both projections occupy about one-third of the cortical depth. perirhinal cortex. however. particularly to the lateral The cell layers of the postrhinal cortex have a band. layer IV. Anatomy of the Hippocampus and the Declarative Memory System 53 caudal pole of the brain. anteromedial dorsal thalamic group and the intrala- cally. sulcus. Area 36 has a in ventrolateral orbital frontal cortex. homogeneous look because the packing density of Postrhinal cortex has strong reciprocal connections cells is similar across layers (Figure 4(b)). A signature feature of the peri- tinguish it from the nearby perirhinal cortex. The border of area 36 is best discerned by characteristics input from frontal associational regions largely arises of the granular cell layer. rostrocaudal extent. At its caudal limit. For most of its quarters of postrhinal afferentation arise in neocor. all cortical the patchy layer II in which medium-sized cells are connections are equally reciprocated. area 36 is arises in the posterior parietal cortex. in that rhinal cortex in the rodent and monkey brains is the cortical input to the postrhinal cortex is strongly presence of large heart-shaped cells in deep layer V dominated by visual and visuospatial inputs. thalamic inputs. The precise location of the dorsal jection is reciprocal. The postrhinal cortex projects directly to sections. The dorsal border of the post. V. subdivisions based on cytoarchitectonic features. For the most part. neocortical areas. terms of sensory input.2 The Perirhinal Cortex and radial appearance. Dorsal retro. which crete granular layer of the dorsally adjacent is itself strongly interconnected with visual associa. guished by the large layer II cells and distinct laminar The postrhinal cortex projects strongly to the look of the cortex. In that appear in both area 36 and area 35 (Figure 5). Connections with the temporal hippo- region can be subdivided into dorsal and ventral campus are modest. The (Figure 1(f)). cortex by the absence of the underlying claustrum. caudally by the postrhinal cortex. ance in that the cells in the deep portion of the layer . there is a broadening of the deep layers. the dorsal subregion has a more organized 3. amygdala is very small and is mainly from the lateral rhinal cortex on the lateral surface tends to be at the and basolateral nuclei. There is also input from the border is difficult to distinguish cytoarchitectoni. The remainder is roughly evenly divided the fundus and both banks of the rhinal sulcus between subcortical and hippocampal afferents. Visual association cortex. and VI each the septal CA1 and subiculum. the region rises neocortical connections of the postrhinal cortex dis. The dorsal splenial cortex also provides a strong projection. the perirhinal cortex includes tex. The subcortical afferents are dominated by the which forms the dorsal border of postrhinal cortex. The inputs from the septum are also relatively the postrhinal cortex ventrally and is easily distin. which is due to the conformation of the region at there are strong direct connections with the hippo- the caudal pole of the brain (Figure 4). dorsal to the fundus. The organization of is that the postrhinal cortex projects strongly to the layer V cells into lines gives the region a radial look. The exception organized in clumps or patches. In general. The postrhinal cortex also same dorsoventral level as the parasubiculum on the projects back to the lateral and basolateral amygdala medial surface.2. medial entorhinal cortex. A strong input fairly rudimentary layer IV as compared to the dis- arises in the caudal and ventral temporal area. but the return projection is sub. That pro- broader layer IV. and Retrograde tract tracing studies show that three. which arise predominantly in the has a more differentiated laminar pattern and a lateral posterior nucleus of the thalamus. Another feature of the region is tion cortices. is provided minar nucleus of the thalamus. small and are dominated by the medial septum.03. The input from the by the parasubiculum. Layer VI has a bilaminar appear- stantially weaker. The strongest associational input laminar structure dorsally than ventrally. ventrally by piriform and entorhinal cortex. the postrhinal cortex receives Perirhinal area 36 is located dorsal to the rhinal almost a third of its total input from visual associa. however. The perirhinal cortex arises at the caudal limit of the tion has a distinguishable granule cell layer IV. especially dorsally. subdivision. In addition to these parahippocampal connections. insular cortex and can be distinguished from insular Another difference is that layer V of the dorsal sub. A convenient landmark. layers II–III. In sagittal campus. In coronal with the septal presubiculum and the parasubiculum. The medial entorhinal cortex borders nuclei. generally described as dysgranular cortex. The primary difference between the two subdivisions is that the dorsal por. It region is slightly narrower than in the ventral is bordered dorsally by temporal association regions. Although the region has a more prominent tional regions. sections. The entorhinal projection is weakly reciprocal.

Not only does it provide the major evenly divided between cortical and subcortical conduit for sensory information to the hippocampal . Other smaller inputs arise in septal CA1 and the The entorhinal cortex forms most of the ventral bor. primarily from the endopiriform nucleus. der and can be distinguished from ventral area 35 by the medium to large darkly staining stellate cells of 3. weakly reciprocated. Layers are tial inputs arise in the amygdala.54 Anatomy of the Hippocampus and the Declarative Memory System (a) (b) structures. area 36 receives roughly equal 36 I II I input from olfactory. The strongest projection back to area general characteristic of area 35 is the organization of 35 arises in temporal CA1. amygdala. (a) Nissl-stained coronal section showing PER areas 36 and 35 and the LEA. area 35 receives its stained for parvalbumin. (b) Adjacent section thalamic nucleus. Other abbreviations: AM. a cell-sparse area between layers III and V. VI VI and sensorimotor regions. layer VI has a bilaminate appearance. area 35 receives the larger input from insular cortex followed by temporal association and frontal regions. Areas 36 and 35 each receive only small POR inputs from posterior associational regions.03. and the medial geniculate caudate putamen. There are also subregional similarities and differences in VI VI polymodal association input.2. This feature is most evident below the rhinal sulcus. visual and visuospatial. parasubiculum. Heavy staining of layer II in the LEA strongest subcortical afferents from olfactory struc- provides a useful marker for the PER-LEA border. The perirhinal cortex receives input from Tev nearly all unimodal and polymodal associational VI regions of neocortex. (d) also from the piriform transition area. but it also receives input its cells into an arcing formation that spans all layers. For example. Like the postrhinal cortex. from temporal subiculum and presubiculum. Id Id Perirhinal areas 36 and 35 are also differentiated III by subcortical connections. In (c) (d) contrast. nucleus of the thalamus. these associational connections are lar- gely reciprocal. The II II MEA MEA I afferent input arises largely in the lateral nucleus. In contrast. although agranular cortex. Other substan- Adjacent section stained for parvalbumin. The entorhinal cortex is of considerable interest in The input to the perirhinal cortex is roughly memory research. but stained coronal section showing MEA and POR. As would V V be expected. but I the basolateral and basomedial nuclei also provide Figure 5 Photomicrographs of the perirhinal (PER). The are smaller. Parvalbumin staining also differentiates the MEA-POR border. lateral thalamic groups. Of course. Layers II and III tend to blend together (Figure 5). Area 36 is weakly connected The region lacks a layer IV and is thus considered with hippocampal and subicular structures. Area 36 receives the I II III V AM I II III V largest cortical input from temporal association LEA regions followed by insular and frontal regions. substantial inputs. Substantial thalamic input arises postrhinal (POR) and lateral and medial entorhinal cortices largely in the dorsolateral group and in the reticular (LEA and MEA). CP. Layer V of area 35 has a disorganized these connections may be functionally important. The strongest subcortical III connections of area 36 are with the amygdala. tures. (c) Nissl. and more densely packed than projection arises in area 35 and terminates most the cells in the superficial portion of the layer. look as compared to the radial appearance in area 36. the midline and labeled I–VI. but there are subregional differ- V VI CP V III II III ences. A presubiculum.3 Entorhinal Cortex layer II and by the appearance of the lamina dissecans. Area 35 receives input back from the septal As in area 36. heavily in the so-called lateral band of the entorhinal Area 35 is generally characterized by a broad layer cortex (see following). auditory. whereas area 35 is domi- V V 35 I II/III I II/III nated by olfactory input from piriform cortex. there is also a heavy intrinsic input POR from area 36. The entorhinal projection is I. the perirhinal cortex projects strongly to the lateral entorhinal cortex. darker.

(b) Unfolded map of the rodent entorhinal cortex showing the lateral (LB) in dark green. D. T. the ento- less prominent lamina dissecans as compared to the rhinal projection to the dentate gyrus. The LEA comprises four fields: the dorsal lateral The entorhinal cortex is strongly connected with entorhinal field (DLE). Each field has unique connec- The entorhinal cortex is a relatively large and tional and/or structural characteristics. V. sic connectivity spans the MEA and LEA.. temporal. L. Amaral et al. (c) The unfolded dentate gyrus. the dentate gyrus. entorhinal area (MEA) is subdivided into a caudal graphically organized. . Perirhinal (a) (b) (c) rs rs S DLE LB DIE IB LEA AE VIE MB D/L ME R C CE T V/M MEA Figure 6 Unfolded maps of the entorhinal cortex and the target of the perforant pathway. ME. Briefly. The entorhinal LB projects to the septal half of the dentate gyrus. An also called the lamina dissecans. This connectional 1975. dorsal lateral entorhinal field. lateral. Thus. LEA has a very outside the band of origin (Figure 6(b)) (Dolorfo clumpy layer II as compared to the more homogeneous and Amaral. amygdalo- entorhinal transitional field. the intermediate band (IB) in medium green. and the MB projects to the temporal quarter. the entorhinal cortex has II that thickens into a characteristic club-shaped been divided into two subdivisions roughly equivalent formation. the LEA exhibits a connectivity with the perforant pathway. ventral. Abbreviations: AE. 1997). area are highly interconnected but do not project bottom) by differences in layer II. Interestingly. 1997). medial entorhinal field. important recent discovery about these regions has mark feature of the entorhinal cortex. The sparsely populated layer IV. septal. and medial band (MB) in pale green. Anatomy of the Hippocampus and the Declarative Memory System 55 formation. The intrinsic connections of the entorhinal cortex rhinal areas (LEA and MEA.. dorsal. In general. are organized in a rostrocaudal manner. the differences in entorhinal structure could provide MEA is bordered by the parasubiculum. CE. DLE. (a) Unfolded map of the rodent entorhinal cortex showing the LEA in light green and the MEA in dark green. but a number of recent discoveries also entorhinal field (DIE). 1946). each band of intrin- layer II of the MEA. M. Medially. tional diversity in the hippocampus. The MEA insight into its role in memory. Further parcellation of each subregion is noted by black lines (Insausti et al. The LEA (Figure 5. color coded to denote the terminations of the perforant pathway. and the ventral intermediate tions to the processing of spatial information. is considered a land. S. entorhinal field (VIE). but there are to do with the relationship of these bands of intrinsic subregional differences. respectively (Figure 6(c)). lateral band projects to the septal half of the dentate Some time ago. sitional field (AE). whereas the intermediate and medial bands further subdivided on the basis of structural and project to the third and fourth septotemporal quar- connectional criteria (Van Hoesen and Pandya. the monkey entorhinal cortex was gyrus. medial. 1998). to modern definitions of the lateral and medial ento. caudal entorhinal field. understanding the areal field (CE) and a medial field (ME). such that the 1909. dorsal intermediate entorhinal field. the IB projects to the third quarter. Figure 1(f )) (Brodmann. ters. the amygdalo-entorhinal tran- suggest that the region may make unique contribu. border with the parasubiculum is marked by a layer In rats and other animals. the MEA (compare Figure 5). Krieg. The medial complicated structure. top) is perhaps cells located in each of three bands of the entorhinal most easily distinguished from the MEA (Figure 5. the dorsal intermediate other parahippocampal region structures. DIE. The rat entorhinal cortex topography suggests that functional diversity within has now been subdivided into six fields according to the entorhinal cortex may be in register with func- similar criteria (Figure 6(a)) (Insausti et al.. 1987). and its connections are topo.

that the subcortical afferents are as influential as the rhinal cortex that arises in all layers and all portions cortical afferents. Strong inputs originate thalamus.and parasubiculum. fields tion across the lateral to medial bands. and oc. and cortex also projects to the MEA. (Pikkarainen and Pitkanen. whereas the MEA projects and medial bands. trum. Piriform CA3. cortical areas. The olfactory input arises in the endopiri- from the pre. 2004). The terminations of the layer III projections (a) (b) CA1 LEA Sub CA2 CA3 DG MEA Figure 7 Illustration of the radial and transverse topography of the entorhinal projection to the hippocampal formation. CA2. . The LEA to the LEA. The parietal. projections from frontal. Input from the cingulate. Medial lateral olfactory tract and the central nucleus and orbital frontal regions provide a strong projec. The amygdala input lum (Witter and Amaral. septal stronger to the LEA than the MEA. The LEA receives very strong input from to all amygdaloid structures except the nucleus of the the piriform and agranular insular cortices. In addition.) The Rat Nervous System. In The entorhinal projections to the dentate gyrus. 1999). formation structures including the dentate gyrus. The entorhinal cortex provides modest and MEA receive input from septal nuclei. There is little differentia. form nucleus and the piriform transition area and is lum targets the entire entorhinal cortex.and parasubicu. though reciprocal connections with the pre. San Diego. temporal.56 Anatomy of the Hippocampus and the Declarative Memory System input arises largely in area 35 and terminates prefer. CA: Academic Press. The terminations of the LEA are in light green. to the middle DG molecular layer (Figure 7). (a) Unfolded map of the entorhinal cortex showing the LEA in light green and the MEA in dark green. The rhinal cortex that is positioned closest to the rhinal projections to CA1 and the subiculum originate in fissure gives rise to the major projections to other layer III. Strong projections arise in claus- of the entorhinal cortex. parietal. and the terminations of the MEA are in dark green. including the lateral frontal. lamic input arises primarily in the midline thalamic and temporal presubiculum projects more heavily nuclei and is stronger to MEA than to LEA. terminates in the lateral and intermediate bands. Projections to the medial frontal and radial topography in that the LEA terminates in the olfactory structures tend to arise in the intermediate outer DG molecular layer. The parasubicu. the amygdala. The dorsal tha- presubiculum projects more heavily to the MEA. In: Paxinos G (ed. and dorsal tion arises in the lateral band. The terminations of the layer II projections exhibit a cipital cortices. and it is possible MEA. A very narrow strip of the ento. cingulate. entially to the lateral band of the LEA. parietal. tion. the inputs are relatively small. whereas the projection to CA1 and subiculum (sub) exhibits a transverse topography. and CA1 of the hippocampus proper. 3rd ed. (b) Line drawing of a hippocampal section perpendicular to the long axis. amygdalohippocampal area and is stronger to LEA tions. arises in all nuclei except the central nucleus and The entorhinal cortex has neocortical connec. and occipital The entorhinal cortex projects to all hippocampal cortices is relatively weak. There is a heavy return projection to peri. Adapted from Witter MP and Amaral DG (2004) Hippocampal Formation. See text for details. though the strongest projec. contrast. 2004). through weaker than perirhinal and postrhinal than MEA. The projection to the DG exhibits a radial topography. the entorhinal cortex projects cortices. and occipital cortices. postrhinal input arises in all portions of the region The entorhinal cortex has widespread connec- and terminates primarily in the lateral band of the tions with subcortical structures. cingulate. the lateral band receives moderate to strong and CA2 originate in layer II of the entorhinal cortex. but the projection the subiculum (reviewed in Witter and Amaral. olfactory structures. CA3.

and layer V. Cells in layer III of the subiculum. Because the presubiculum and this dorsal presubiculum are also extensive. Layer V is very thin the MEA projects to the proximal CA1 and distal and contains pyramidal cells. and cell-sparse gap are two layers. The organization of the CA1 and subicular thicker and contains a mixture of cell types. 2004). Regarding parahippocampal connectivity. and hippocampal parahippocampal Brodmann (1909) are both included in the presubi. Deep to layer II is a dark band that contains layers III. lum. Layer III from the subiculum terminates in layer I. and the temporally Whereas cells in layer II tend to form clusters. Layer VI is moderately The presubiculum is bordered dorsally by retrosple. similarities. (b) Nissl-stained coronal section showing the PreS and ParaS. dentate gyrus and all fields of the hippocampus Layer II of the six-layered presubiculum is thick proper. cells directed projection is relatively weak. 2(f). Layer II stains moderately darkly (Figure 8(c)). The input in layer III have a more homogeneous look. The presubiculum projects to superficial layers ing pyramidal cells (Figure 8(b)). (c) Adjacent section stained for acetylcholinesterase (AChE). is sometimes called the postsubicu. densely packed. medially by the subiculum. It is reciprocally connected with the subicu- and contains small. Area 48. The presubiculum has extensive associational. (a) Drawing of a coronal section of the rat brain at the level of the angular bundle (ab). spar. ventrolaterally by the parasubiculum (Figures 2(e). . Layer VI is slightly subiculum. Anatomy of the Hippocampus and the Declarative Memory System 57 exhibit a transverse topography such that the LEA dissecans of the parasubicular cortex.4 Presubiculum cell-sparse gap. is separated from the deep layers by a narrow. AChE provides an excellent marker for these regions. projections back to deep layers of the LEA and MEA As it turns out. the most dorsal extension of the and temporal parts of the presubiculum are highly presubiculum. acetylcholinesterase (AChE) is an roughly reciprocates the forward projections. and 8(a)). The inset designates the areas shown in panels (b) and (c). round. to lightly stained in AChE preparations. The projection to the septal sub- are even smaller.2. the 3.03. and a good marker for the parasubiculum. Areas 48 and 27 according to commissural. connections (Witter and Amaral. the sely populated gap that is continuous with the lamina presubiculum projects to superficial layers of the (a) (b) (c) ab ab PreS PreS ab POR MEA 1d ParaS ParaS LEA Figure 8 Photomicrographs of the presubiculum (PreS) and parasubiculum (ParaS). and darkly stain. excellent marker for the presubiculum. Layer II is outlined for the PreS. iculum is moderately strong. and the combined layer II/III is outlined for ParaS. Connections with the contralateral lum. Deep to this projects to distal CA1 and proximal subiculum. it may be more appropriate to designate The presubiculum provides a weak input to the the area collectively with a single term. though commissural component exhibit considerable cytoarchitectonic connectivity may be stronger ventrally than dorsally. The septal culum. interconnected. AChE is also nial cortex. and also darkly staining.

Septal presubi. the anterior thalamic projection to the para- a very heavy input from the MEA portion of the subiculum provides a pathway by which the anterior medial band. primarily the anterior thalamic band providing the heaviest return projection. the Interestingly. The entorhinal cortex activates the dentate III contains large. moderately darkly gyrus via the perforant pathway. inputs are from the dorsal thalamus. and the anterior cingulate cortex layer II of the entorhinal cortex. weak input from the LEA. In AChE preparations. that project dorsally and ventrally. The heaviest neocortical input to the presubicu. The presubicular– layer V is moderately darkly stained. The entorhinal pro- (Witter and Amaral. A broad. and the deep layers by a broad lamina dissecans. Extrinsic connections of the parasubiculum are gets the same nuclei. Parahippocampal nuclei. layers Cajal. in layers I and III of the contralateral homotopic torhinal divisions. anterodorsal. Septal presubiculum projects projections are heavier and more extensive than much more heavily to the MEA than the LEA. dentate gyrus. Commissural projections terminate temporal presubiculum projects heavily to both en.58 Anatomy of the Hippocampus and the Declarative Memory System parasubiculum.2. sometimes arises mainly in the subiculum and terminates in termed the postsubiculum) projects massively to layer I and superficial layer II. Other than the input from the ante- mamillary nuclei of the hypothalamus. dorsomedial structures. the anterodorsal. basal. the parasubiculum exhibits lum arises in the granular retrosplenial cortex.03. densely packed. but the connections with the entorhi. Temporal presubiculum projects projections to the hippocampal formation. There is also a strong few. Finally. 2004). 3.5 The Parasubiculum For most of its rostrocaudal extent. the parasubiculum is grouped together partly because of the sequential interposed between the postrhinal cortex and the activation pattern that was identified several decades medial entorhinal area. splenial cortex and visual cortex. and accessory basal nuclei. and nal cortex are by far the strongest.3 The Hippocampal Formation lum is bordered by presubiculum dorsally and the medial entorhinal area ventrally (Figure 2(e) and The structures of the hippocampal formation are 2(f)). the only other subcortical afferents arise in the amygdala from the lateral. and these inputs the presubiculum has reciprocal connections with the are quite weak. rior thalamus. and the laterodorsal even heavier than that to the MEA. The presubiculum receives subcortical input inputs arise mainly from the MEA. The Golgi for darkly staining a small number of neurons . incoming information at very early stages. specifically. combined layer II/ ago. and is also a modest presubicular input. but the dorsal ones. lamina dissecans and layer V are lightly stained. This layer is separated from pathway from the denate gyrus activates CA3. entorhinal projection is bilateral. The parasubiculum projects selectively to prefrontal areas. Temporal presubiculum receives noted. The only neocortical afferents arise in retro- cholinergic input arising from septal nuclei. and virtually thalamus can affect hippocampal processing of nothing from the perirhinal and postrhinal cortices. the parasubicu. but substantial connections with other parahippocampal weaker inputs arise in prelimbic cortex. the mossy fiber staining pyramidal cells. the parasubicular projection to POR is anteroventral. The struc- heavily to the LEA and the MEA and moderately ture projects directly to the molecular layer of the heavily to perirhinal areas 36 and 35. As has been previously of the lateral band. The ventral minates in layer III. There nuclei including the anteroventral. A massive return projection tar. The hippocampal input to the parasubiculum The dorsal extension (Brodman’s area 48. and almost exclusively ter. This is especially interesting given culum receives heavy input from postrhinal cortex that the parasubiculum receives strong inputs from and a moderately heavy input from the MEA portion anterior thalamic nuclei. There are also return postrhinal cortex. who used a technique developed by Camillo I and II/III are darkly stained (Figure 8(c)). Some of and VI can be distinguished from one another and the earliest and most famous studies of the structure tend to run continuously with deep layers of the of the nervous system were conducted by Ramón y medial entorhinal area. with the medial from the thalamus. 3. Like the presubiculum.03. laterodorsal nuclei. Septal presubiculum also provides region. Layers V the CA3 Schaffer collaterals activate CA1. At more caudal levels. a moderately heavy input to the postrhinal cortex. largely directed to The parasubiculum has associational connections medial entorhinal cortex. The primary subcortical jection is much heavier to MEA than to the LEA.

the longitudinal. c-e. the laminar structure is perpendicular to the radial axis. each of which is a conductive device. Indeed. G. now clear that those pyramidal cells belong to CA3. Thus. a. and the part closest to CA1 is distal (Figure 10). nervous system is made up of countless separate units. h. pro. the major hippocampal and dentate associa- tional projections extend along the septotemporal axis as well as the transverse axis.3. however. b. subiculum. CA1 pyramidal ule cell has a small number of primary dendrites (one cells. collaterals of alvear fibers. the part axes. In this terminology. corpus callosum. the part of CA1 closest to CA3 is proximal. very densely packed. C. F. soma. recurrent collaterals. dentate gyrus can be considered the proximal limit campus. Cajal’s elegant studies and drawings. campus. Along the transverse axis. Although the lamellar hypothesis shaped research on the hippocampus for years to come and continues to influence modern concepts of hippocampal function. fimbria. i. the molecular layers are superficial and the layers on the opposite side of the principle cell layers are deep. or the part of CA3 next to the dentate gyrus was some- nerve cells composed of dendrites. we use the term of the CA3 lying in the V of the dentate gyrus is proximal CA3. D. The beginning of the CA3 principle cell layer protrudes into the poly- morphic area of the dentate gyrus. as well as the identity of these cells. to four) that are covered with spines. B. g. oval cells that have a dark angular bundle. He further proposed that times called CA4. or slabs. The dentate granule cell layer contains small. and occasionally in modern reports. The molecular layer lies outside the V. and so on. the Because of the complex architecture of the hippo. In earlier formation by Ramón y Cajal (1909). modern neuroanatomical research has revealed that the hippocampal projections are much more diver- gent than is suggested by the lamellar hypothesis. which is vided the basis for the neuron doctrine. They further proposed that this lamellar organization would permit strips. it is and conducted to distant locations through axons. Based primarily on electrophysiological data and mapping of vasculature. transverse. and radial position relative to the dentate gyrus. With modern techniques for information is received on the cell bodies and dendrites defining connectional characteristics. retrosplenial area. Anatomy of the Hippocampus and the Declarative Memory System 59 in the brain. cingulum. Similarly. of the hippocampus to function as independent units. 3. it is helpful to describe its structure in terms and transverse locations designated according to of three axes.1 The Dentate Gyrus The dentate gyrus is three-layered cortex whose principle cell layer is V shaped (Figure 10). E. Anderson and colleagues (1971) proposed that the hippocampal formation was organized in parallel lamellae stacked along the longitudinal axis. Ammon’s horn. Schaffer collaterals. appearance in cell stains (Figure 10(a)). K. septotemporal for the longitudinal axis of the hippo- including the rodent hippocampus (Figure 9). This conformation has generated some confusion over the border between CA3 and the dentate gyrus polymorphic Figure 9 Drawing of the circuitry of the hippocampal layer. and axons. dentate gyrus. Finally. As previously discussed. Each gran- perforant path fibers. and the poly- morphic layer lies inside the V. subicular cell.03. H. Cajal proposed that the nomenclatures. axon entering the cingulum. The dendrites . Abbreviations: A. cingulum fibers. orthogonal to the long axis of the hippocampus.

its terminal so CA2 field coincides with the perforant path terminations in sr the dentate molecular layer. between the granule cell layer and the polymorphic CA1 layer. parvalbumin slm sr (Figure 10(c)). with its soma in the molecular layer. and the stratum oriens (so). it is possible to visualize the three sublayers gcl sr of the molecular layer (Figure 10(b)). Axo-axonic cells have a (b) dendritic tree of radial branches extending through the molecular layer. The most prominent types are the gcl ml basket cell and the axo-axonic. the stratum radiatum (sr). One such cell type is the formation. There are a number of interneurons with somata in the poly- morphic layer that make inhibitory connections on Figure 10 Photomicrographs of the hippocampal the dentate granule cells. CA3 also innervates the inner molecular layer. GABA. CA2. which outer stratum lacunosum-moleculare (slm).60 Anatomy of the Hippocampus and the Declarative Memory System (a) extend into the molecular layer all the way to the hippocampal fissure. the molecular contains the stratum lucidum (sl). gcl These large multipolar cells are so named because of sr ml their large dendritic spines. the so-called thorny sl gcl excrescences. but the best characterized is the mossy cell. and the outer sublayer stains orange. the perforant path input terminates. and parvalbumin. extensive axonal arbor in the outer two-thirds of stained section showing the same regions. (a) Nissl-stained coronal section showing the hilar perforant-path (HIPP) associated cell. One interesting type of (c) GABA-immunoreactive cell. The dashed line layer perforant-path (MOPP) associated cell. (b) Adjacent section stained for has a dendritic tree in the polymorphic layer but heavy metals using the Timm’s method. The demarcates the border between the DG and CA3. The inner third sl ml of the molecular layer. The solid dendritic tree and axonal arbor of this cell occupy line shows the location of the hippocampal fissure and the outer two-thirds of the molecular layer. there are several cell types. next to the lightly colored so pol CA3 granule cell layer. In material stained for heavy metals using the Timm’s slm method. Though the molecular layer mostly contains dendrites. The CA fields contain the associational pathway (HICAP) related cell. Thus. there are a few cell types that stain for VIP. The middle gcl DG sublayer stains yellow. terminates. (c) Parvalbumin. . sl pol The basket cell interneurons are quite large with a CA3 so gcl single. where demarcates the border between DG and CA1. A subset of these cells is also immunoreac- tive for the calcium-binding protein. The axon arborizes extensively CA1 in the granule cell layer. Layers of the DG the molecular layer where the perforant path input are the outer molecular layer (ml). The mossy fiber axons from the gran- CA3 so pol DG ule layer terminate on these spines. and CA1. has an axon restricted to the outer CA1 two-thirds of the molecular layer. Another type is the hilar commissural- and the polymorphous layer (pol). most of which so are immunoreactive for gamma-aminobutyric acid CA2 sr (GABA). the granule cell layer (gcl). The polymorphic layer contains a number of cell slm types. cell. or chandelier. so CA2 The molecular layer contains mostly dendrites sr from cells of the granule and polymorphic layers. aspiny apical dendrite extending into the mo- DG lecular layer and several basal dendrites extending into the polymorphic layer. stains a reddish brown. which dentate gyrus (DG) and hippocampus proper. comprising fields CA3. the gcl. In the subgranular region.

immunoreactivity. but there are also interneurons with long hippocampal circuitry that does show a lamellar pro.3.03. The mossy fiber axons can be easily identified basal dendrites form a thick arbor in the stratum in Timm’s preparations in which they stain very oriens. of the septal nuclei.2 The Hippocampus Proper axonic cells and the basket cells are conveniently The hippocampus proper consists of three fields. the hypothalamus. cholecystokinin. CA2. The primary hypothalamic input is dendritic arbor extending into the stratum oriens. and some with tems in the brainstem. ule cell mossy fiber projection to CA3. The CA3. dentate gyrus. the stratum lucidum. in the medial septal areas tend to terminate septally. eventually reaching the stra. There is also a heterogeneous group of basket darkly (Figure 10(b)). the Ammon’s horn fields CA3. and the brain pyramidal cells. The cholinergic input larger and the layer thicker compared with CA1. There are many other interneuron . parvalbumin. Axon vasopressin. role in the regulation of local circuits in the hippo- gic input is from several of the raphe nuclei and also campus proper. This projection is probably gyrus. Hippocampal interneurons differ in terminates in the polymorphic region. The dentate gyrus receives input arbors of pyramidal cells in CA2 are mixed. similar to those described for the located between cell-sparse layers. It is topographically organized such that cells in other ways. This is the one component of the short axons. immunoreactive for somatostatin. Most hippocampal interneurons have the transverse axis. calbindin. and connectivity. discussion. Superficial to (O-LM) cells have a dense axonal arbor restricted the principle cell layer are the stratum radiatum and to the stratum lacunosum moleculare. CA3 has an addi. but they may also be The only output of the dentate gyrus is the gran. from the pontine nucleus of the locus coeruleus and Interneurons undoubtedly play an important terminates in the polymorphic layer. tum lucidum. We mention a few types here. Like the dentate gyrus. tree is localized to layers that receive recurrent tional thin layer. The serotoner. similar to CA3. The oriens lacunosum-moleculare ciple cell layer is the stratum oriens. For example. collaterals of each cell project to the full extent of and calretinin. Anatomy of the Hippocampus and the Declarative Memory System 61 The entorhinal cortex provides the only cortical superficial to the principle cell layer and deep to the input to the dentate gyrus through the perforant stratum radiatum. axons that project outside the hippocampal forma- jection pattern. morphology. and the stratum lacunosum-moleculare and a basal morphic layer. they are ents of the hippocampus proper. The consisting of about seven collaterals that terminate in apical dendrites are radially oriented and span all the dentate gyrus polymorphic layer before entering superficial layers to the hippocampal fissure. which lies just collaterals. from the supramamillary area and terminates in a CA3 cells proximal to the dentate gyrus have smaller narrow band of the molecular layer just superficial dendritic trees than the cells distal to the dentate to the granule cell layer. these fields are cell bodies in stratum oriens and innervate principal a three-layered cortex consisting of a principle layer cell dendrites. with large arbors. similar to CA3. There is substantial subcortical input from The principal cell layer consists. it lacks mossy fiber input. pathway. synaptic properties. positioned to receive excitatory input from all affer- Proximal to distal from the dentate gyrus. but for a full CA3 pyramidal layer. The dendritic the stratum lacunosum-moleculare. The noradrenergic input is small arbors. cells whose apical dendrites extend into the stratum moleculare and whose extensive basal dendrites span the entire depth of the stratum oriens. Minor dopa. Hippocampal in- and the substantia nigra. and CA1 Some hippocampal interneuron cell types have (Figure 10). In addition to the CA3 projection. terneurons are GABAergic. some from each of the modulatory neurotransmitter sys. from the septal region originates in the medial septal The small and often-overlooked field CA2 contains nucleus and in the nucleus of the diagonal band of larger pyramids. but overall. Each pyramidal cell has an apical dendritic arbor and cells in lateral septal structures terminate extending upward through the stratum radiatum temporally. The axo- 3. Deep to the prin. but is similar to CA1 Broca. The projection terminates in the poly. the One prominent interneuron type in the pyramidal mossy fibers give rise to an associational connection cell layer is the chandelier or axo-axonic cell. The fibers travel along or within the tion. similar to CA1. the dendritic arbors of cells in excitatory and appears to target both granule cells CA3 are larger than those in CA1. The pyramidal cells in CA3 are stem modulatory systems. see Freund and Buzsaki (1996). primarily. neuropeptide Y. minergic inputs arise in the ventral tegmental area laminar location. The dendritic and interneurons.

parasubi. presubiculum. Cortical CA3 cells tend to terminate more superficially in input arrives from the perirhinal. CA1 receives weak Extrinsic connections of CA3 are not robust noradrenergic input from the locus coeruleus and except for the substantial projection to the lateral weak serotonergic input from the raphe nucleus. however. The pyramidal cell 3. is to the subiculum. which are considered to be . and medial prefrontal of the stratum oriens. in CA3. The intrinsic projections are similar The subiculum is widely considered the output to those of CA3. nucleus accumbens. CA3 does not. the septal half projects more heavily minor input from the amygdala basal nucleus. distal subiculum. Distal CA1 projects to dentate gyrus) tend to project to levels of CA1 that proximal subiculum. whereas distal CA3 cells tend to torhinal cortices. the hypothalamus. There is also a Field CA1 exhibits only a weak associational/ small collateral projection to the polymorphic commissural connection. The input from the amygdala is in general the CA3-CA3 associational projections more substantial. projections of proximal tical input from extrahippocampal structures.3.and parasubiculum. In this way. Distal CA3 cells (farther from the dentate nates in midproximodistal subiculum. although the topographies may not structure of the hippocampal formation. horizontal trilami. extrinsic connections. and infralimbic cortex. Inputs from piriform cortex have also projects more heavily to perirhinal cortex. especially to distal CA1. a feature that is in striking layer of the dentate gyrus.03. in the raphe nucleus. The projection to CA1 underlying some of the putative functional differ- is called the Schaffer collateral projection. and the basal Field CA2 can be differentiated from CA3 by the nucleus of the amygdala. There robust extrinsic projections. CA2 also pro. is cholinergic and arises in the medial septal nucleus Of the hippocampal CA fields. but available evidence suggests The CA3 pyramidal cell axons are highly col. Other intrahippocampal jections to CA3/CA2 are called the associational connections. The associational projections exhib. and the projections to contralateral rons project to CA3 and to the polymorphic layer of structures are called the commissural projections. Temporal the locus coeruleus. the dentate gyrus. but in stratum oriens. ences in CA3 and CA1. the vides a small collateral projection back to the dentate pre. The major projection from CA1. septal nucleus. In addition. ipsilaterally and contralaterally. lack of mossy fiber input and the associated thorny excrescences (Figure 10(b)). the medial entorhinal area. that the region is differentiated from CA3 by lateralized and project to all CA fields both hypothalamic input from the supramamillary area. Amygdala terminate extensively along the septotemporal axis input arises in the basal and accessory basal nuclei. Not much is known specifically about CA2 nar. Like CA3. This difference has been interpreted as culum. The pattern of the There is a prominent input from the nucleus reuniens terminations of the commissural projections mirrors of the thalamus that terminates in the stratum lacuno- those of the associational projections. and en- stratum radiatum. A noradrenergic projection arises in entorhinal area. CA1 has the more and the nucleus of the diagonal band of Broca. retrosple- terminates primarily on the GABAergic interneurons nial cortices. whereas the temporal half of CA1 tum radiatum. it differs from its parahippocampal neighbors. The cortical projections is a GABAergic component of the projection that include the perirhinal. entorhinal. and a serontonergic input arises levels also project to the anterior olfactory nucleus. and proximal CA1 projects to extend farther in the septal than temporal direc. the pro. sum moleculare. the lateral been reported. Temporal CA3 receives a cortex.3 The Subiculum layer also stains more intensely for parvalbumin (Figure 10(c)). CA1 interneu- projections. For example. contrast to the robust associational network present project to the subiculum. (Witter and Amaral. to postrhinal cortex. The mid-CA1 projection termi- tions. Subcortical input to CA1 is grossly terminate more deeply in stratum radiatum and in similar to the subcortical input to CA3 but differs in stratum oriens. or entorhinal cortex. the details. and radial trilaminar cells. however. retrosplenial cortex. and that projection graphy such that proximal CA3 cells (closer to the exhibits a strict topography.62 Anatomy of the Hippocampus and the Declarative Memory System types including the bistratisfied. postrhinal. preinfralimbic. The septal input is weaker and terminates it complex transverse and radial topographies. are extensive. 2004). like CA3. gyrus) tend to project farther in the temporal Field CA1 receives substantial cortical and subcor- direction. The major subcortical input to CA3 There is also a weak dopaminergic input. gyrus. however. postrhinal. The Schaffer collateral projection exhibits a topo. be the same. and which terminates in the stratum oriens and the stra. In general.

and ventral pre- subiculum. The latter projec- lum are more diverse. and the apical dendrites extend cortex projects relatively more strongly to septal into the molecular layer. cially area 35. . to the anterior thalamic complex. interneurons. subiculum. Temporal subiculum a pyramidal cell layer containing the principle cells. The projection exhibits a topogra. proximal part of the temporal subiculum projects to Connections of the subiculum with other hippo. There are also local associational connec. The basal dendrites terminate in the deep part strongly to temporal subiculum. entorhinal intermediate and medial bands project The border with CA1 is further demarcated by the more strongly to temporal subiculum. there are documented subiculum. rosplenial projection is reciprocated. targets the posterior and basolateral nuclei. The proximal part of the ever. the population of inter. and the columnar network that is roughly interconnected. The ret- pyramids are numerous smaller cells. pathway fibers contact GABAergic cells that stain The diverse subcortical projections target the sep- for parvalbumin. and the by a principle cell layer that is more loosely packed. mus also projects to the subiculum. how. Anatomy of the Hippocampus and the Declarative Memory System 63 input structures. stratum lacunosum moleculare of field CA1. septum. though they cannot be distinguished mor. though the return findings suggest that there are two populations of projections are modest (O’Mara et al. The pro- tions confined to the pyramidal layer and the deepest jection to the nucleus accumbens is topographic such part of the molecular layer. All septotem- neurons appears similar to that observed in field CA1 poral levels of the subiculum project to the lateral (Witter and Amaral. though the details subiculum. Temporal subiculum receives projection is to deep layers of entorhinal cortex. The pyramidal cells are subiculum. The entorhinal lateral band situated CA1 and the distally situated presubiculum projects more strongly to septal subiculum. input from the nucleus reuniens. The projection is not truly lamellar. that the proximal part of the septal subiculum pro- gests that the bursting pyramidal cells form a jects to rostrolateral nucleus accumbens. The presubiculum projections to projection cells. Although both types are retrosplenial cortex. 2004). but available senting varied types of interneurons. and bursting cells tend to be septal part of the subiculum projects to the ventral located deep in the layer. The subiculum also projects heavily to large and of uniform shape. Perforant evidence suggests that the frontal projections are not. The available data sug. midproximodistal CA1 projects to mid. the caudomedial nucleus accumbens. mammillary nuclei. prelimbic. the hypothalamus. pyramids. but the projection arises primarily in the The associational connections of the subiculum proximal part. 2001). It terminates in the medial preoptic area phy such that proximal CA1 projects to distal and the ventromedial. The distal and be located superficially. but the best-characterized tion is reciprocated. cells project to the entorhinal cortex. The amygdala projec- minate in all layers. are not well described. The The subiculum also receives a substantial input from subiculum can be distinguished from the proximally the entorhinal cortex. probably repre. There is also widening of the middle layer of the subiculum. the nucleus accumbens. and distal CA1 projects to proximal connections with thalamic nuclei. Septal subicular input to the postrhinal cortex is pus proper and the dentate gyrus. There is no commissural tion arises primarily in the temporal subiculum and projection. but in general. the subiculum is a equally strong. modest input from the perirhinal and postrhinal cor- The principle cell layer contains large pyramidal tices. The most prominent neocortical projections are to phologically. Among the infralimbic. The hypotha- campal structures is limited to input from CA1. nucleus of the thalamus. as any part of CA1 projects to about one-third septal subiculum projects to the anteromedial of the septotemporal extent of the subiculum.and parasubicular cortices. Septal input arises mainly in the extend temporally from the point of origin and ter. nucleus of the diagonal band. Perirhinal cortex projects relatively more cells. Electrophysiological the pre. and the thalamus. espe- three-layered cortex with a deep. polymorphic layer. Not much is known about subicular tal complex. which is continuous with the moderate input to perirhinal and postrhinal cortices. lamic projection also arises primarily in the temporal which is massive. dorsomedial. the amygdala. provides massive input to the entorhinal cortex and and a molecular layer. and anterior cingulate cortices. Like the CA fields of the hippocam. it is possible that only the bursting frontal areas include the medial orbital. Finally. Available evidence Septal levels of the subiculum provide substantial suggests that the anteroventral nucleus of the thala- input to the lateral and medial entorhinal cortices. and the postrhinal of the principle layer. is modest. but input to perirhinal cortex. So-called regular spiking cells tend to retrosplenial and prefrontal cortices. but the distal part projects The parahippocampal connections of the subicu..

ponent. primary way station for sensory information on its Longitudinal intrinsic connections integrate across way from the neocortex to the hippocampus. but there are also direct tory information and then passed on to entorhinal neocortical inputs to the entorhinal cortex. visuospatial auditory. olfactory Septal sub LEA MEA Septal DG/CA3 Septal CA1 Visual. The pre. . auditory. we have attempted to schematize the flow of sensory information through the hippo- 3. CA field. and somatosensory regions. auditory. tial. and the MEA. SS. of the neocortical input arrives by way of the peri.03. LEA. along with a small input from hippocampal structures. cortex. Presubiculum are involved in the preprocessing of sensory informa. band somatosensory. Visual. Much modalities before transmission to perirhinal area 35. olfactory olfactory tial. pri- connections suggest that parahippocampal structures marily to the lateral band of the MEA. Presubiculum somatosensory POR Parasubiculum Visual. and visual The distinct patterns of cortical afferentation to para. somatosensory. region. Medial Intermediate Lateral Visual. postrhinal cortex. somatosensory. visuospatial. DG. olfactory olfactory Auditory. Black arrows indicate the primary pathways by which sensory information traverses the hippocampal memory system. parasubiculum. visuo. especially the septal com- evidence that there is also substantial functional diver. SS. dentate gyrus. Auditory. and auditory association cortex. auditory. lateral entorhinal area. MEA. olfactory visuospatial. PER Area 36 Visual. somatosensory. retrosplenial. The view that parahippocampal structures Subicular input is integrated with direct visuospatial have different functions is consistent with emerging input to the presubiculum. auditory. Visual. medial entorhinal area. association regions. That information is the topography of the parahippocampal–hippocampal integrated and transmitted to entorhinal cortex. band visuospatial.4. The subiculum and parasubiculum are also considered postrhinal cortex receives visual and visuospatial input structures for the hippocampal memory system. Visual.03. primarily the lateral band of the LEA. olfactory auditory. Beginning with the input through the Hippocampal System structures. olfactory olfactory Visual. visuospatial. perirhinal cortex. visuospa- somatosensory. the postrhinal cortex. olfactory olfactory olfactory Temporal DG/CA3 Temporal CA1 Temporal sub Figure 11 Diagram of the flow of sensory information through the hippocampal memory system. POR. Visual. visuospatial. That information is forwarded to the sity among hippocampal formation structures. band spatial. auditory. Abbreviations: CA.64 Anatomy of the Hippocampus and the Declarative Memory System 3.4 Conclusions In Figure 11. Visual. visuospatial. is targeted by the subiculum in a topographical man- tion provided to the hippocampus. auditory. Note that only the stronger connections are indicated and that emphasis is on the feedforward connections as opposed to the feedback connections. auditory. PER. auditory. perirhinal area 36 receives sensory input The entorhinal cortex is widely recognized as the from visual. Visual. and that there is ner such that septotemporal levels of the subiculum functional diversity within the parahippocampal map onto septotemporal levels of the presubiculum. visuospatial SS. Visual. subiculum. This polymodal input to area 35 is joined by olfac- rhinal and postrhinal cortices. Visuospatial Area 35 Visual. Visual. somatosensory. auditory somatosensory. Visual. input from posterior parietal. visuospatial. olfactory somatosensory. sub. the intrinsic connections. visuospa. visuospatial. auditory.1 The Flow of Sensory Information campal memory system.

projects lar to the lateral to medial bands of intrinsic massively to the postrhinal cortex. layer III. Curr. the available evidence distribution of cortical efferents. Hippocampus 5: 390–408. Herrero MT. Hippocampus. J. Neurol. Anderson M. I. Neurol. and Cohen NJ (1994) Two functional represented in both the rodent and the primate brain. cortical afferenta. J. and Gigg J (2001) The projections to CA1 and the subiculum originate in subiculum: A review of form. the para. 6: 347–470. (1995) The human entorhinal cortex: A cytoarchitectonic way is similar in the primate and rodent brains. the rat and monkey hippocampal memory systems projects to postrhinal cortex and the MEA. 398: pocampal region and the hippocampal formation are 49–82. Many of the connectional principles are also con. Neurol. Comp. Brain Sci. Finally. Brain Res. and Amaral DG (1995) The perirhinal and postrhinal cortices of the rat: A review of the 3. in the monkey and the lateral entorhinal area in the Ramón y Cajal S (1909) Histologie du systáeme nerveux de rat project to the border of the CA1 and subiculum. perirhinal cortex. Krieg WJS (1946) Connections of the cerebral cortex. parasubiculum. Neurobiol. CA1 and distal subiculum (Witter. Cytoarchitectonic organization. Organization and Role in Cognitive Functions. Liepzig: Barth. components of the hippocampal memory system. tions to the dentate gyrus. organization. Comp. Eichenbaum H. layer II of the entorhinal cortex. 437: poral axis. hippocampus. J. Comp. monkey and the rat. Anderson P. the terminations of the projec. Neurol. medial LEA. rat: Borders and cytoarchitecture. Commins S. 264: 326–355. 17: 449–518. Taking into account the differences in brain Freund TF and Buzsaki G (1996) Interneurons of the hippocampus. Opin. J. the perforant pathway projec. served. In: Witter MP tory information is predominant in the temporal and Wouterlood FG (eds. Exp. 1986. Insausti R. Insausti AM. Hippocampus 7: 146–183. 355: 171–198. there appears to Grosshirnrinde in ihren Prinzipien dargestellt auf Grund des Zellenbaues. but the degree of processing of different modalities is weighted differently along the septo. which receives direct. and Witter MP (1997) Entorhinal cortex tion of the parahippocampal structures is similar of the rat: Cytoarchitectonic subdivisions and the origin and across species.2 The Comparative Anatomy neuroanatomical literature and comparison with findings of the Hippocampal System from the monkey brain. l’homme and des vertâebrâes. The rostral entorhinal cortex the hippocampal formation in rat. and Skrede KK (1971) Lamellar organization of hippocampal pathways. 13: gated but follows a pathway that includes the 222–38. reaches the full septotemporal extent of the hippo- campus. and Cowan WM (1987) The entorhinal tal hippocampal formation. the O’Mara SM. Olfactory information is less segre. Neurol. Additionally. Neurol. Also in both. Behav. B. Neurosurg. Dolorfo CL and Amaral DG (1998) Entorhinal cortex of the rat: As indicated earlier.) The Parahippocampal Region. cortex of the monkey: I. and temporal hippo. 1993). physiology and function. J. Anatomy of the Hippocampus and the Declarative Memory System 65 The most septal component of the presubiculum. Sobreviela T. Squire LR and Zola-Morgan S (1991) The medial temporal lobe Amaral. London: Oxford University Press. J. Insausti R. be functional diversity in the processing of sensory Burwell RD (2001) The perirhinal and postrhinal cortices of the information that is organized along the septotem.4. . and olfac. but modest. Burwell RD and Witter MP (2002) Basic anatomy of the predominant in the septal hippocampus. connectivity observed in the rodent entorhinal cor- subiculum. The albino rat. Visual and visuospatial input is processed and elaborated along a pathway that Amaral DG (1993) Emerging principles of intrinsic hippocampal includes the postrhinal cortex. Thus.03. Comp. are excellent models for the medial temporal lobe To summarize. 64: 129–55. In the analysis. 3: 225–229. the evidence suggests that both and visuospatial input along with its subicular input. Structure of the cortical areas. all components of the parahip. and accessory basal nuclei of the amygdala to transverse topography. Organization of intrinsic connections. Brodmann K (1909) Vergleichende Lokalisationslehre der campal formation structures. septal presubiculum. Otto T. basal. Prog. Paris: Maloine. information from all modalities memory system in the human brain. and CA2 originate in 84: 277–323. the sep. J. The intrinsic connections of the memory system. and lateral MEA. Burwell RD. Tuñón T. Witter MP. Bliss TV. visual tex. and Gonsalo LM suggests that the architecture of the perforant path. parahippocampal region in monkeys and rats. Insausti R. Neurol. Neurobiol. In addition. References temporal axis. Comp. Science 253: 1380–6. Amaral DG. the monkey entorhinal cortex also exhibit patterns simi- area sometimes termed the postsubiculum. and then back to postrhinal cortex Comp. Taken together. 403: 229–260. medial entorhinal area in the rat project to proximal Psychiatry 20: 11–21. Pikkarainen M and Pitkanen A (1999) Projections from the tions originating in entorhinal layer III exhibit a lateral. visual and visuospatial information is 17–41. size and sensory processing needs. the caudal entorhinal cortex in the monkey and the Scoville WB and Milner B (1957) Loss of recent memory after bilateral hippocampal lesions. lateral MEA. CA3. 1993.

J. 3rd edn. CA: Academic Press. 463: 67–91. Neuroscience 120: 893–906. Neurol. San Von Bonin G and Bailey P (1947) The Neocortex of Macaca Diego. London: Oxford University Press. Biol. Urbana: University of Illinois Press. Van Hoesen GW and Pandya DN (1975) Some connections of the Hippocampus 3: 33–44. parahippocampal cortices? A historical overview of the 203: 67–82. formation. nomenclature and boundaries applied to the primate medial Witter MP (1993) Organization of the entorhinal-hippocampal temporal lobe. rhesus monkey. In: Paxinos G (ed. Med. Mulatta. 95: 1–24. Adv. I. . system: A review of the current anatomical data. with emphasis on the septotemporal organization Suzuki WA and Amaral DG (2003b) Where are the perirhinal and of its intrinsic and extrinsic connections. Witter MP (1986) A survey of the anatomy of the hippocampal Comp. Cytoarchitectonic and chemoarchitectonic organization. Temporal lobe afferents.) The Rat Nervous System. Exp. entorhinal (area 28) and perirhinal (area 35) cortices of the Witter MP and Amaral DG (2004) Hippocampal Formation.66 Anatomy of the Hippocampus and the Declarative Memory System Suzuki WA and Amaral DG (2003a) Perirhinal and Von Economo C (1929) The Cytoarchitectonics of the Human parahippocampal cortices of the macaque monkey: Cerebral Cortex. Brain Res.

The major distinction is between declarative memory. Nondeclarative memory refers to a collection difficulty in learning new facts and events (antero- of nonconscious knowledge systems that provide for grade amnesia). neurological amnesia.3 Anatomy 68 3.02. trauma.4. amnesia) refers to difficulty in learning new informa- 3.1 Motor Skills and Perceptual Skills 74 3.1 Introduction the study of patients with memory disorders (amne- sia) and from animal models of amnesia (See Chapter Memory is not a single entity but is composed of 3.1 Impairment in Declarative Memory 70 3.02.04. Significant informa. Schacter and Tulving.04.5.04. University of California at San Diego.2 Anterograde Amnesia 70 3. patients tion about how memory is organized has come from fully recover. 3. USA ª 2008 Elsevier Ltd. Patients with neurological amnesia the capacity of skill learning.4.04.04.04. 1994). and certain other ways of and events that were acquired before the onset of interacting with the world.5 Spatial Memory 73 3.04. The terms ‘explicit mem. Declarative or disease that damages the medial temporal lobe or memory refers to conscious knowledge of facts and diencephalon. 1992.5 Adaptation-Level Effects 76 3.5 Spared Learning and Memory Abilities 73 3.5. or no anterograde amnesia.04.04.04. Neurological amnesia causes severe events.2. Squire and Y. Shrager.04 Declarative Memory System: Amnesia L.04.5. the also typically have some difficulty remembering facts phenomenon of priming.04.04. Neurological amnesia is characterized by a loss of lel memory systems.6 Functional Amnesia 77 3. Long-term memory can be divided into several paral. 1. Functional amnesia is charac- which can be studied in isolation from perception terized by a profound retrograde amnesia with little and other intellectual abilities.2. habit formation.04. Functional amnesia is a psychiatric 67 .5.2.5.04.2.4 Priming 75 3. All rights reserved.5. amnesia (retrograde amnesia).5.2 Artificial Grammar Learning 75 3.09).7 Habit Learning 76 3.7 Summary 77 References 77 3. It presents as a different pattern of antero- The brain is organized such that declarative mem.5.04.04.5.4.2 Nondeclarative Memory 74 3. grade and retrograde memory impairment than ory is a distinct and separate cognitive function.04. ory’ and ‘implicit memory’ are sometimes used as well Functional amnesia is rarer than neurological and have approximately the same meanings as amnesia and can occur as the result of an emotional declarative and nondeclarative memory.2. respectively. Squire. It occurs following brain injury declarative and nondeclarative memory.4 Retrograde Amnesia 72 3. CA.2 Etiology of Neurological Amnesia 68 3. San Diego.04.04.3 Category Learning 75 3.1 Immediate and Working Memory 73 3.3 Remembering versus Knowing and Recollection versus Familiarity 71 3.4 The Nature of Amnesia 70 3.04.04.2.4.04. 3. R. tion or in remembering the past.1 Introduction 67 3. Amnesia (neurological amnesia and functional several separate systems (See Chapters 1.2.6 Classical Conditioning 76 3.4. In some cases.

1986). encephalitis. The amnesic surgery. ultimately the anatomy of normal memory. entorhinal. Many patients the monkey (Figure 1) and in the rodent. and parahippocampal (patient R.) to ask what brain areas are associated with these had surgery in 1953 to treat severe epilepsy. FMRI reveals activation during these tasks of cortex). The common factor in all of these conditions is lesions of the medial temporal lobe. tered to amnesic patients can also be administered to dial temporal lobe cortices were removed bilaterally healthy individuals while their brain activation is mea- (the entorhinal cortex and most of the perirhinal sured. as estimated from MRI scans.) One well-studied case (H. chronic alcohol abuse. head condition is associated with neuronal death and tissue injury. medial temporal lobe structures have become more specific. who became amnesic as a result of viral and the diencephalic midline.P. and the rupture collapse.M. cause amnesia. poral lobe. The same tasks that are adminis- the hippocampus and much of the surrounding me. temporal lobe neurons (Rempel-Clower et al. To understand the anatomy of human amnesia.).2 Etiology of Neurological and W. learning and remembering in the same structures that. right-sided damage affects memory for nonverbal As questions about amnesia and the function of material (e.B. These observa- tions suggest that a reduction in hippocampal volume Neurological amnesia results from a number of con. damage limited to the hippocampus itself total intracranial volume. of the brain – the medial aspects of the temporal lobe. memory for faces and spatial layouts). severe and persistent amnesia. For volumes of the adjacent medial temporal lobe struc- example. One can also calculate the is sufficient to cause moderately severe amnesia.H. virtually all of the hippo- material-specific amnesia. 2000). volume of the hippocampus itself as a proportion of In humans. it is important to calculate the volumes of ity of memory impairment is exacerbated by lateral temporal cortex and other regions that might be additional damage outside of the hippocampus.3 Anatomy being studied. but the tissue does not disappear altogether and repair of an anterior communicating artery aneur. quanti- tative information about the damage in the patients 3. in global amnesia.g. it resulted in a when damaged. while (Stefanacci et al. it has become vital to obtain detailed.04. affected. Most of memory processes. The sever. ditions including Alzheimer’s disease or other likely indicates the near complete loss of hippocampal dementing illnesses (See Chapter 3. (Animal studies later elucidated Functional MRI (fMRI) of healthy individuals who the critical anatomical components of this memory are engaged in learning and remembering allows one system. postmortem neurohistological analysis (patients L. 1996).M. volume reduction the disruption of normal function in one of two areas as measured by MRI is more dramatic. has damage to all of the perirhinal cortex.. In the case of patients with models of amnesia provide information about the restricted hippocampal damage.) proved to have lost virtually all the neurons Amnesia in the CA fields of the hippocampus. in one carefully studied case of amnesia tures (the perirhinal.’s seizures. patient E. 3.. Two such patients whose brains were available for detailed. Although the surgery was successful in redu.. the only significant damage was a cortices) in proportion to intracranial volume. left-sided campus. anoxia. and most of the parahippocampal cortex damage affects memory for verbal material. reduction in hippocampal volume of about 40%. of approximately 40%.29). bilateral lesion confined to the CA1 field of the when there is extensive damage to the medial tem- hippocampus (Zola-Morgan et al..04. because fibers and glia remain. In addition. see following. It is important to characterize patients in this Thus. the adjacent cortex. In patients with larger ism.M. one can calculate the neural connections and structures that are damaged. For example. and unilateral damage results in all of the entorhinal cortex. and It is also possible through structural magnetic reso. An animal .. infarction. Specifically.68 Declarative Memory System: Amnesia disorder. severe amnesia results when damage extends way in order to address the kinds of questions now beyond CA1 to the rest of the hippocampus and to being pursued in studies of memory and the brain. Last. cing the frequency of H. and no particular brain structure or region with restricted hippocampal damage have an average is known to be damaged. ischemia. Bilateral damage results encephalitis. animal nance imaging (MRI) to detect and quantify the models of human amnesia have been established in neuropathology in amnesic patients. single-case studies are not nearly as useful as group studies involving well- Well-studied cases of human amnesia and animal characterized patients.

As a memory (Squire and Zola-Morgan. dentate gyrus. 1991). a memory that depends on representations in both TE and PG) can be revivified even when a partial cue is presented. TE and PG) and to support the strengthening of connections between these regions. convergent activity must occur along projections from these regions to the medial temporal lobe. anatomically related structures monkey in the early 1980s (Mishkin. one can now study memory in rodents and important structures are the hippocampal region (hip. the distributed representations in neocortex can operate independently of the medial temporal lobe. The consistency learning.) model of human amnesia was established in the complex) and adjacent.g. The networks in the cortex show putative representations concerning visual object quality (in area TE) and object location (in area PG). succeeded in identifying the system work have also made it possible to pursue parallel of structures in the medial temporal lobe essential for studies in simpler animals like rats and mice. LeDoux.g... mem. Declarative Memory System: Amnesia 69 PG V1 TE TF/TH PRC EC CA3 DG CA1 S Figure 1 Schematic drawing of primate neocortex together with the structures and connections in the medial temporal region important for establishing long-term memory. Squire (entorhinal cortex. ory impairment is exhibited on the same kinds of tasks 1996) and for the enhancement of declarative memory of new learning ability that human amnesic patients by emotion (Adolphs et al. the gateway to the hippocampus. memory for a whole event (e. where widespread efferent projections return to neocortex. Projections from neocortex arrive initially at the parahippocampal gyrus (TF/TH) and perirhinal cortex (PRC) and then at entorhinal cortex (EC). long- term memory. using impairment and its anatomical basis. perirhinal cortex. is not a part of the fail. The hippocampus and adjacent structures are thought to support the stabilization of representations in distributed regions of neocortex (e. first in the dentate gyrus (DG) and then in the CA3 and CA1 regions. and parahippo- and Zola-Morgan. 1983). . If this disparate neural activity is to cohere into a stable.. between the available neuroanatomical information minations. 1994. These lines of the animal model. 1982. and subicular relevant to the human condition. Further processing of information occurs in the several stages of the hippocampus. The fully processed input eventually exits this circuit via the subiculum (S) and the entorhinal cortex. The same animals succeed at tasks of motor skill declarative memory system itself. aspects of emotional learning (Davis. Subsequently. They also do well at learning pattern discri. The result. The amygdala. Damage to the medial temporal lobe system causes anterograde and retrograde amnesia. Once sufficient time has passed. which share with motor skills the factors of from humans and the findings from monkeys have incremental learning and repetition over many trials. have some confidence that what one learns will be pocampus proper. 1997). The severity of the deficit increases as damage involves more components of the system. considerably illuminated the description of memory Systematic and cumulative work in monkeys. Following lesions of the campal cortex). (This diagram is a simplification and does not show diencephalic structures involved in memory function. although critical for bilateral medial temporal lobe or diencephalon.

memory impaired in amnesia. the anterior thalamic Amnesia is characterized especially by profound dif- nucleus. major distinction is between declarative and nonde- comparing performance on memory tasks following clarative memory. amnesia resembles medial temporal lobe amnesia in and general intellectual capacity. scious knowledge that the impairment becomes rhinal cortex.04. pocampal cortex especially impairs spatial memory.70 Declarative Memory System: Amnesia Anatomical connections from different parts of the have intact immediate memory and memory for a neocortex enter the medial temporal lobe at different great deal of information from the past.2).. It is also important to note that the is either true or false. Last. The important structures include the mediodorsal thalamic nucleus. Damage limited to the hippocampal region Declarative memory can be brought to mind as a causes significant memory impairment. Amnesic individuals exhibit significant memory even on difficult tests that require the ability to dis- impairment. they 2006). Declarative memory poral lobe. However. and the mammillo-thalamic tract. language and social skills. 1994a. whereas damage to the parahip. In some patients with memory impairment. amnesia can also occur in the absence of other cognitive deficits and without any change in personality or social skills. a way to model the external world. especially points (Suzuki and Amaral. learning and memory (see section 3.04. declarative mem- medial temporal lobe might make qualitatively dif. amnesic patients can appear quite strongly to the perirhinal cortex than to the parahip. Shrager et al. b). and intact perceptual abil- For example. occur as part of a more global dementing disorder order. visual association cortex projects more ities. the mamillary ficulty in new learning. and in this sense it ory impairment. when the pattern of sparing and loss. In fact. normal on casual observation. functional system. midline diencephalon. Indeed.04. Declarative memory provides to the adjacent cortex increases the severity of mem. attention. This work has shown that the severity of refers to the capacity to remember the facts and events memory impairment depends on the locus and extent of everyday life. The stored representations are discovery that larger medial temporal lobe lesions flexible in that they are accessible to multiple response produce more severe amnesia than smaller lesions is systems and can guide successful performance under a compatible with the idea that structures within the wide range of test conditions. The possibility that different medial temporal lobe struc.g. . criminate between highly similar images containing patients have intact ability for some forms of new overlapping features (Levy et al. 2005. It is only when one pocampal cortex. forming and maintaining associations between arbitra- damage to the perirhinal cortex especially impairs rily different kinds of material (e.4. It is important to appreciate that amnesic patients are tures make different contributions to memory has not impaired at all kinds of long-term memory.4.4 The Nature of Amnesia circumscribed form of amnesia.. Also. Another important brain area for memory is the 3. the hippocampus lies at the evident. It is the kind of memory that is meant of damage within the medial temporal lobe memory when the term ‘memory’ is used in ordinary language.. Monkeys referred to as anterograde amnesia. the internal medullary lamina. and monkeys with mammillary nucleilesious that includes other cognitive deficits such as impair- exhibit a modest impairment. patients have intact intellectual functions and intact perceptual functions. In this more 3. Further. visuospatial abilities. The been addressed in a number of studies in monkeys. Because diencephalic ments in language.1 Impairment in Declarative Memory structures that precedes it in the hierarchy.04. learning to object recognition. whereas the parietal cortex projects interrogates their capacity for new learning of con- to the parahippocampal cortex but not to the peri.2 Anterograde Amnesia diencephalon. system. these two regions likely damage is restricted to the medial temporal lobe or form an anatomically linked. Declarative memory is the kind of damage to different components of the medial tem. In addition. intact structures play different roles in declarative memory. Yet. despite their memory deficit. which raises childhood. end of the medial temporal lobe system and is a recipient of convergent projections from each of the 3. Amnesia can with medial thalamic lesions exhibit an amnesic dis. ory is especially suited for rapid learning and for ferent contributions to memory function. patients with neurological the question of whether these medial temporal lobe amnesia can have intact intelligence test scores. associate two different words).5. and this impairment is nuclei. but damage conscious recollection.

where a list of words is presented and partici.3 Remembering versus Knowing time and place (episodic memory). tered or not). Thus. One needs that includes information about the context in which only to know about certain facts. they are presented with one objects (visual). for- ability). tion is also available about the original learning ory ability) and event learning (or episodic memory context (Wixted and Stretch. Both episodic memory and tice. memory is impaired pants try to decide (yes or no) if each word had regardless of the kind of material that is presented been presented in a recent study list (Squire and and the sensory modality in which information is Shimamura. In contrast. or cued recall (e. Tulving. During test. amnesic patients were mal studies show that both R and K responses are taught new factual knowledge in the form of three. remembering and knowing responses are closely semantic memory are declarative. tory. Knowing (K) is meant to refer contrast. asking for recall of 1998). presenting an view that episodic memory and semantic memory item and asking whether it was previously encoun. Remembering (R) is meant to refer to the circum- 1983).g.g.. MEDICINE cured HICCUP). related to the strength of a memory and that items Formal experiments have compared directly the given K responses are often items for which informa- ability to accomplish fact learning (or semantic mem. olfactory). are similarly impaired in amnesia (Squire and Zola. and quan.. visual. and they are asked to (Levy et al. one need stance when an item elicits a conscious recollection not remember any particular past event. Formal experiments The memory impairment in amnesia involves both have also demonstrated recognition memory impair- difficulty in learning factual information (semantic ment for auditory stimuli. in a cued recall task. episodic memory is thought to then asked to recall the passage immediately and require these cortical sites and the medial temporal after a 12-min delay.g.04.. The amnesic patients understand what underlies these deficits. during testing. 2004). in a in cortical storage sites as a consequence of experi- standard test of free recall. titative brain measurements are needed in order to MEDICINE cured ______). For instance. recognition (e. In recalling this type of information. showing that their impairment spans the Amnesic patients are impaired on this task as well. impaired in amnesia. zero segments (Squire and Shimamura. Thorough landmarks that identify the particular time and place studies of healthy individuals indicate that in prac- when an event occurred. were similarly impaired on tests of fact memory ent data support the idea that declarative memory is (what word completed the sentence fragment) and separable from other brain functions. amnesic word sentences (e. and odors (olfactory) word from each pair (ARMY). The term ‘seman- and Recollection versus Familiarity tic memory’ is often used to describe declarative memory for organized world knowledge (Tulving. 1985). audi- tests. 1986). but one’s life (Tulving... usually recalling ( Janowsky et al. Taken together.. They are lobe. three tasks. Episodic memory. 1983). 2004). beyond the medial temporal lobe. visual and olfactory domains. In these cases. In one experiment. Unlike semantic memory. damage might extend sented with the instruction to complete each laterally. in the item was learned. Amnesic patients with damage lobes to work together with the frontal lobes in order to the medial temporal lobe do well at immediate to store when and where a past experience occurred recall but are impaired at the delay. memory) as well as difficulty in learning about spe- cific episodes and events that occurred in a certain 3. 2005a.g. Last.4.b).17. recognition memory of viduals study a list of word pairs. That is. The patients were impaired on all recall the word that was paired with it (TABLE).. In one such experiment. In any case. regardless of whether memory is tested by Squire. the data favor the free (unaided) recall. patients and control participants were given a . designs (visual). presented. fragment with a word that had been studied (e. sentence fragments were pre- et al. is autobiographical memory for the events of to a circumstance when an item appears familiar. such as ARMY– amnesic patients was assessed for line drawings of TABLE.g. 1989). Declarative Memory System: Amnesia 71 visual perceptual deficits have been described (Lee Then. 1995). Semantic knowledge is thought to accumulate an item when a hint is provided). pres.. on tests of event memory (what specific events Amnesic patients are impaired on tasks of new occurred during the testing session) (Hamann and learning. Last. memory for the original learning context is not avail- episodic memory stores spatial and temporal able (See Chapter 2. indi. the memory deficit in amnesia is global and Amnesic patients are also impaired on recognition encompasses all sensory modalities (e. participants are read a ence and with support from the medial temporal short prose passage containing 21 segments. For example. 1986).

or disproportionately.. 2006). Last. 1881). where the delay between initial learning and a lesion patients with hippocampal damage were impaired can be manipulated directly. whereas familiarity The severity and extent of retrograde amnesia is does not require any recollection of the original determined by the locus and extent of damage. after a lesion. grade amnesia covering decades.. life. is 3. For amnesia. long-term memory. lesions can spare old weak memories while tion test. and then a short interval In addition to impaired new learning.g. That is.e. limited retrograde amnesia covering a few years iarity can be supported by the adjacent cortex within prior to the onset of amnesia. severely amnesic patients can produce 2003). 2007). Accordingly. The extent of retro- (K). that responded to familiar images during a recogni. there was a familiarity age of the memory that is critical. weaker memories (Wixted. memories for the facts and were similarly impaired for R and K responses events of childhood and adolescence are typically (Knowlton and Squire. In one study. Indeed. variable that combines estimates of recollection and Because the study of human retrograde amnesia is familiarity. the memories of healthy individuals with respect to A distinction closely related to remembering and the number of details. Trace con- ditioning is a variant of classical conditioning in which the condition stimulus (CS). patients with medial temporal lobe damage have extensive retro- damage limited to the hippocampus should be selec. 1995. rabbits given trace eyeblink conditioning. electrophysio. signal in the hippocampus regardless of whether any These same points can be illustrated by a study of recollection had occurred) (Rutishauser et al. and type of memory loss is referred to as retrograde retention of the conditioned response is much poorer .72 Declarative Memory System: Amnesia recognition test 10 min after studying words. The patients were impaired for both R and K grade amnesia can be relatively short and encompass responses. When damage tively. 2003. retrograde amnesia An alternative view is that recognition decisions sometimes can be ungraded and extensive and are based on a unidimensional strength-of-memory include the facts and events of early life. the original learning experience.. participants indicated whether they graded. It has sometimes been proposed that Patients with restricted hippocampal damage have recollection relies on the hippocampus. Recollection and the number of prompts needed to begin a narra- involves remembering the contextual associations of tive (Bayley et al. That is. 2005b). the clearest data about retrograde amnesia gra- associated with strong memories and familiarity with dients come from studies using experimental animals. the duration of the narratives. term memory that has not yet been converted into a logical recordings from patients being evaluated for long-term memory. In normal rabbits. Findings from such on a recognition memory test. In this view.04. damage to the (e. or it can be more extensive and that was tested after 1 week. showing that it is the when recollection failed (i. as encephalitis and head trauma. the evidence suggests that detailed autobiographical narratives of their early remembering and knowing are two different expres. were operative in the absence of the does not reflect simply the vulnerability of a short- hippocampus (Wais et al. impaired at recollecting occurs beyond the brain system that supports declara- information and less impaired at recognizing material tive memory. while famil. such as a tone. and they performed like a control group only 1 or 2 years. Thus. These memories were indistinguishable from sions of declarative memory. Even then. information acquired in the distant remembered it (R) or whether they knew that the past (remote memory) is spared relative to more word was presented but had no recollection about it recent memory (Ribot. were acquired before the onset of amnesia. retrograde amnesia iarity. where they gave con. The results gradients of retrograde amnesia can occur within indicated that both processes. remote epilepsy surgery found neurons in the hippocampus memory can be even better than recent memory. the patients cover decades. Patients with large the medial temporal lobe. knowing is recollection and familiarity. These familiarity signals were present even disrupting strong recent ones. First. 2006). Retrograde amnesia is usually temporally each word. This forgetting occurs gradually after training. experience. which can result from conditions such when it can be supported by familiarity.4 Retrograde Amnesia presented and terminated. also see Manns et al.4.. a capacity for recollection is likely to be tests. as in the case of remembering and based almost entirely on findings from retrospective knowing.. studies make three important points. recollection and famil. Second. That is. intact. temporal fidence judgments (scale of 1–6). 500 ms) is imposed before the presentation of medial temporal lobe also impairs memories that the unconditioned stimulus (US). Third..

2001). memory tests that have no obvious spatial compo- nent. the same regions that were initially involved in the either spatial or nonspatial. It is clear that spatial memory is impaired refers to a heterogeneous collection of abilities. During the process of con. use alternate and novel recent memory illustrates that the brain regions routes to describe how to travel from one place to damaged in amnesia are not the permanent reposi. The areas of neocortex events. Thus. memory in amnesic patients showed that the patients Immediate memory refers to what can be held actively were similarly impaired on tests of spatial memory in mind beginning the moment that information is and tests of nonspatial memory.4. 1971). Amnesia rodent (O’Keefe and Dostrovsky.5. needed for the long-term storage of spatial knowl- solidation. Instead. perceptual and cognitive skills.04. These find- tion after learning.M.. difficulty with the test of spatial memory (Cave and Nevertheless. 1999). One well-studied patient after training had no effect on the weaker 30-day-old with large medial temporal lobe lesions and severe memory (Kim et al. Memory is at that point sup. 1988). both spatial and nonspatial. 1986). and phylogenetically early forms of Similarly. complete aspiration of the hippocam. In these cases. Declarative Memory System: Amnesia 73 30 days after training than after only 1 day. whereas the same lesion made 30 days spatial knowledge is intact. though. asked to navigate to a destination in a virtual envi. ing. Further. It is the information that forms the focus of .. Amnesic patients are impaired on which afford the capacity to acquire knowledge non- tests that assess their knowledge of the spatial layout consciously. As is the case with nonspatial memory. memories borhood while imagining himself oriented at some undergo a process of reorganization and consolida. the neocortex is always important. these important for long-term memory are thought to be structures are not repositories of long-term memory. another. There was no special received. Nondeclarative memory memory. It is also sensitization.11. 1996. other location (Teng and Squire. the noted patient H.04. priming. simple classical condition- ronment (Maguire et al. pus 1 day after training abolished the strong 1-day. It is a striking feature of amnesia that many kinds of learning and memory are spared.5 Spared Learning and Memory phalon are also important during initial learning and Abilities during consolidation. all of in human amnesia. temporary memory capacities that operate shortly fore focused especially on the status of spatial after material is presented. was impaired at experience-dependent behavior like habituation and recalling object locations (Smith. amnesia (E. such as recall or recognition of items (Squire 3. memories are vulnerable if there is edge and does not maintain a spatial layout of learned damage to the medial temporal lobe or diencephalon. environments that is necessary for successful naviga- After sufficient time has passed. impairs only long-term declarative memory and tial idea has been that the hippocampus creates and spares immediate and working memory. Discussions of amnesia have there. Nondeclarative memory includes motor of an environment. and they are also impaired when skills.) was able to mentally navigate his The sparing of remote memory relative to more childhood neighborhood. Accordingly. formal experi- ments that directly compared spatial and nonspatial Amnesic patients have intact immediate memory. adaptation-level effects. that amnesic patients are impaired on through performance rather than recollection. storage and retrieval tion. Furthermore. Immediate memory and to support spatial memory (See Chapter 2. the available data support the view of memory no longer require the participation of that the hippocampus and related medial temporal these brain structures. lobe structures are involved in learning new facts and ported by neocortex. Memory is not a 3. processing and analysis of what was to be learned. 1978). memory is expressed clear. remote old memory. habits. but the structures of the medial temporal lobe and dience.04. 1991). 1995). Squire. and point correctly to locations in the neigh- tories of long-term memory. 3.. Spiers et al. during which time the neocortex ings show that the medial temporal lobe is not becomes more important.P. as well as uses spatial maps and that its predominant function is nondeclarative memory. an influen. O’Keefe working memory can be viewed as a collection of and Nadel.1 Immediate and Working Memory and Shimamura.5 Spatial Memory unitary faculty of the mind but is composed of many Since the discovery of hippocampal place cells in the parts that depend on different brain systems.

and the or after a long delay.04). digits. Yet. Memory is expressed through perfor- rehearsal. ness. Indeed.g. he could quite limited.5. the term encompasses several they can hold it in mind by rehearsal.1 Motor Skills and Perceptual The intact capacity for immediate and working Skills memory was well illustrated by patient H. In these situations. at least not in the same left to himself for several minutes. will remember fewer items than their healthy counterparts. when a list of eight or spared in amnesia. Rather. only a minute or two later. the amygdala. One would say different kinds of memory. It is ate memory capacity (typically. when the cerebellum and on processes intrinsic to neocortex capacity of working memory is exceeded. and it operates outside of aware- more items must be remembered) or when informa.. patients (Figure 2). 3. Nondeclarative forms of in this case that the patients have carried the contents memory have in common the feature that memory is of immediate memory forward by engaging in explicit nonconscious.M. was describe what has been learned. In addition to the central role of the amygdala in emotional learning. when One can learn how to ride a bicycle but be unable to he was asked to remember the number 584. but it is not itself a brain- remember the material for minutes. and amnesic Long-term memory Declarative Nondeclarative (Explicit) (Implicit) Facts Events Skills & Priming & Simple Nonassociative habits perceptual classical learning learning conditioning Emotional Skeletal responses responses Medial temporal lobe Neocortex Cerebellum Reflex diencephalon Striatum Amygdala pathways Figure 2 Classification of mammalian long-term memory systems. and it is independent of the even the knowledge that memory is being influenced by medial temporal lobe system (See Chapter 2.2.74 Declarative Memory System: Amnesia current attention and that occupies the current stream in mind. it is also able to modulate the strength of both declarative and nondeclarative learning. This rehearsal-based activity is referred to mance and does not require reflection on the past or as working memory.5. Intact immediate memory explains why amne- sic patients can carry on a conversation and appear quite normal to the casual observer.2 Nondeclarative Memory amount of material to be remembered is not too Nondeclarative memory refers to a collective of non- large (e. then patients can conscious knowledge systems. if the 3. after his of thought. The taxonomy lists the brain structures thought to be especially important for each form of declarative and nondeclarative memory. tive memory is distinct from declarative memory. This is because the learning of schemes and holding the information continuously motor skills is largely nondeclarative. The past events. H.04. in the ability to repeat back a short string of schemes for holding the number in mind. Nondeclarative forms of memory depend tion must be recalled after a distraction-filled interval variously on the neostriatum. or as long as systems construct. and he was able to sense that one might recall riding a bicycle on a par- retain this information by working out mnemonic ticular day with a friend. .04. difficulty for amnesic patients arises when an amount The following examples illustrate that nondeclara- of information must be recalled that exceeds immedi. a three-digit number). This type of memory is reflected. for not remember the number or any of his mnemonic example.M. The capacity of immediate memory is attention had been directed to another task.

For example. After viewing these letter not encountered recently. even though the lying prototype. in which they exposed to several exemplars of a single category. sented intermixed with new drawings. In formal experiments. suppose that strings.04. participants are told for the first time that the line drawings of a dog. as member of the learned category more readily than measured by their decreased reaction times for press. participants identify the proto- above each key was illuminated. hammer. The task was to press one of to whether they are or are not members of that four keys as rapidly as possible as soon as the location category.04. the items used during training. 1992. despite being impaired at time occurs independently of whether one remem- recognizing the letter strings that were used during bers having seen the items earlier. amnesic Amnesic patients performed as well as control par- patients learned to read mirror-reversed words at a ticipants. remember the task itself.2. These include such skills as reading mirror. If in a later test these same pictures are pre- individuals often report that they are simply gues. Even though aloud. participants are presented encounter with an item results in a representation with a series of letter strings that are generated of that item. after they had finished the mirror-reading task. Amnesic patients but now the dog. with the instruction to and that their task is to decide for a new set of letter name each item as quickly as possible.5. The first grammar–learning task. Declarative Memory System: Amnesia 75 patients can learn these skills at a normal rate. amnesic Perceptual skills are also often intact in amnesic patients and control participants were shown a series patients. In one 3. In addition. all of which were distortions of an under- as individuals with intact memory. Amnesic patients are was changed. Then. even though the patients were severely normal rate and then retained the skill for months. though they items that were used to train the category (Knowlton had learned it normally (Reber and Squire. and Squire.4 Priming 3. as a learned the sequence. amnesic prototype of the category..5. Typically.2. In one experiment. 1993). and airplane are letter strings were formed according to a set of rules presented in succession. 1994). and airplane are named classify items as grammatical or nongrammatical as about 100 ms more quickly. of dot patterns formed by distorting a randomly reversed print and searching a display quickly to find generated pattern that was defined arbitrarily as the a hidden letter. impaired at recognizing which dot patterns had Yet. grammatical or nongrammatical. 1994.2 Artificial Grammar Learning Priming refers to an improved ability to identify or Another kind of learning that is intact in amnesia is produce a word or other stimulus as a result of its the learning of artificial grammars. Having seen a series of dot patients acquired perceptual skills at the same rate patterns.2. and in some cases they could not remember that memory system was able to account for the they had ever practiced the mirror-reading skill on a dissociation between categorization and recognition previous occasion (Cohen and Squire. amnesic patients and control participants In tasks involving category learning.04. strings whether each one is formed by the same set of about 800 ms are needed to produce each name rules as the set they had just studied. as did the normal participants. The amnesic patients type. It is worth noting that it they could not remember the words that they had has been suggested that a model assuming a single read. despite a severe deficit in recognizing the no declarative knowledge of the sequence. despite having poor and Squire.3 Category Learning experiment. the new sing. the amnesic patients had little or category. In an artificial prior presentation (See Chapter 3. they are able to classify new letter strings as drawings will still require about 800 ms to name. responded successively to a sequence of four illumi. hammer.12). even when the pro- ing a key when it was illuminated. the reaction times increased for both also able to classify items according to a learned groups. and that representation then allows it to according to a rule system that specifies what letter be processed more efficiently than items that were sequences are permissible. participants try to classify new items according nated spatial locations. been presented for training. Strikingly. Knowlton exhibit this effect at full strength. (Nosofsky and Zaki. memory of seeing the items earlier. . 3. The improved naming well as healthy individuals. participants then were able to patients did not remember what items were discriminate new dot patterns that belonged to the encountered during the task and sometimes did not training category from other dot patterns that did not. When the sequence totype itself was not presented. Amnesic patients the initial training (Knowlton et al. For example. or central tendency of the category. 1996). 1980). 1998). participants are performed a serial reaction-time task.5.

. a brief interval of In another experiment.7 Habit Learning 3. the tone comes to elicit an eyeblink in would produce these words if they had not been anticipation of the airpuff. which included. conditioning. In a typical mind. or nondeclaratively as judged as heavier than they would be if the light.76 Declarative Memory System: Amnesia The dissociation between intact priming and difficulty remembering their prior experience accu- impaired recognition memory in amnesic patients rately (Benzing and Squire. campus dependent because it requires the acquisition Amnesic patients exhibited intact fluency (the ten. In one of the prim.2.P. first set of objects with one hand and then make amnesic patients are disadvantaged. tive forced-choice and yes–no recognition. and retention of conscious knowledge during the dency to label those words that were identified more course of the conditioning session (Clark and Squire. These results relationship acquired differential trace conditioning. 1994). can be particularly compelling. It is reflexive and automatic and the words MOTEL and ABSENT. 2005). After a number of (The probability was about 10% that participants pairings.5. for example. E. a habit. who is so severely 3. One of the best-studied examples of classical condi- ing tests. For example.. healthy individuals will solve weighted objects had not been presented. However. Participants had a strong tendency (30–50%) conditioning procedure. Amnesic many trial-and-error learning tasks quickly by sim- patients show this effect to the same degree as ply engaging declarative memory and memorizing healthy individuals. participants were tioning in humans is delay conditioning of the shown a short word list. such as experience with one set of stimuli influences how a learning to make one choice rather than another. Then.) In addition. word stem completion. In trace conditioning.5. whereas there was no correlation fluency are not sufficient to increase recognition between awareness and conditioning performance on a performance.5. participants saw words 500–1000 ms is interposed between the CS and the slowly clearing from a mask. second set of stimuli is perceived (e. 1989). This form of conditioning requires the hippocam- each word as quickly as possible and then make a pus (McGlinchey-Berroth et al. Formal recognition judgment (old/new) about whether the experiments suggest that trace conditioning is hippo- word had been presented in a preceding study phase. However. a mild airpuff directed to the eye. experience with light-weighted ganglia). 3. Two tests of priming were given. E. rate as healthy individuals. In this circumstance. they have can also be constructed that defeat memorization . awareness of the temporal contingency between performed entirely normally on the two priming the tone and the airpuff is unrelated to successful tests but performed at chance (50%) on the recogni. they were depends solely on structures below the forebrain.6 Classical Conditioning amnesic that he exhibits no detectable declarative Classical conditioning refers to the development of memory (Hamann and Squire. through memorization.5 Adaptation-Level Effects Habit learning refers to the gradual acquisition of Adaptation-level effects refer to the finding that associations between stimuli and responses. two parallel tests of and retain the tone–airpuff association at the same recognition memory for words were given: alterna. and is a lobe as the result of herpes simplex encephalitis. sustained an association between a previously neutral stimulus complete bilateral damage to the medial temporal (CS) and an unconditioned stimulus (US). They tried to identify US. quintessential example of nondeclarative memory. 1997). delay conditioning task.P.g. Only those who became aware of the CS-US tion was impaired (Conroy et al. even when they experience the which responses are correct. tion tests. For example. support the idea that priming depends on brain struc. One study investi- gated priming in patient E.2. tasks judgments with the other hand. a tone repeatedly precedes to produce the words that were recently presented.04.. There was a correlation between measures of aware- tures independent of the medial temporal lobe. with including the cerebellum and associated brainstem instructions to form the first word that comes to circuitry (Thompson and Krupa.. and ness taken after the conditioning and trace conditioning they show that the combined effects of priming and performance itself. 1997).04. Many tasks can be acquired either declara- objects subsequently causes other objects to be tively.2. In both groups.04. their heaviness Habit learning depends on the neostriatum (basal or size). but their recogni. eyeblink response. quickly as old) and intact priming.P. shown the fragments MOT___ and ABS___. 1998). Amnesic patients acquire presented recently.

Shimamura AP. and occurs in association with some prior psychiatric Davis M (1994) The role of the amygdala in emotional learning. Schul R. Affect. even when the task aging studies of healthy volunteers. and related can be learned declaratively by healthy individuals. Rev. Sometimes. the disorder lasts longer. 109: is lost. Hopkins RO. and Thompson RF (1995) Hippocampectomy ability to learn new information is usually intact. Neurosci. individual to individual. Behav. Neurosci. Cogn. In such a case. 5(1): 14–20. and Squire LR (2005a) Robust habit operates outside of awareness and independently of learning in the absence of awareness and independent of the medial temporal lobe. and that monkeys with lesions of the neostriatum Adolphs R. studies in experimental animals continue to reveal In this case they succeed by engaging habit memory. Neuropsychologia 27(8): 1043–1056.. the cause of the systems: The role of awareness. and Squire LR (2005b) The in amnesia (Bayley et al. Science 280(5360): 77–81. Hamann SB and Squire LR (1995) On the acquisition of new in a confused or frightened state. As more is learned about the concurrent discrimination task. Despite profound impairment in memory impairment in patients with frontal lobe lesions.6 Functional Amnesia Benzing WC and Squire LR (1989) Preserved learning and memory in amnesia: Intact adaptation-level effects and learning of stereoscopic depth. 538–547. especially autobiographical memory and even 1027–1044. memory. 4(3): 291–300. Neuron 38(1): 135–144. but not remotely. though factual information about the patient’s life Janowsky JS. and Squire LR (2003) Successful recollection of remote autobiographical memories by amnesic patients with medial temporal lobe lesions. Nature 436: 550–553. Neurosci. neuroscience of memory. Healthy individuals can learn all eight better diagnosis. but in most cases functional Conroy MA.46). Behav. more opportunities will arise for achieving object pairs. Mem. Hopkins RO. Clark RE. particular brain structure or brain system is impli. and the lost memories return. Thus. Cohen NJ and Squire LR (1980) Preserved learning and retention disorder must be due to abnormal brain function of of pattern-analyzing skill in amnesia: Dissociation of knowing some kind. Frascino JC. 1992). sia. Semantic or fac. Memory for the past declarative knowledge in amnesia. 850–854. by making the outcomes on 3. Hamann SB and Squire LR (1997) Intact perceptual memory in the absence of conscious memory. 36: 225–266. Bayley PJ. Behav. The impairs the memory of recently. Science 210(4466): 207–210. and Babinsky R (1997) Impaired declarative memory for emotional material following bilateral (basal ganglia) are impaired.7 Summary each trial probabilistic. Learn. It is known that this task is acquired at a normal (slow) References rate by monkeys with medial temporal lobe lesions. history. the Kim JJ. treatment. and Squire LR (2005) On the contribution of perceptual fluency and priming to recognition amnesia is preceded by physical or emotional trauma memory. or the objects. humans appear amygdala damage in humans. the instructions. Its presentation varies considerably from how and knowing that. insights about what memory is and how it is orga- This situation is nicely illustrated by the eight-pair nized in the brain. nature of memory disorders and has led to a better It is also true that severely amnesic patients who understanding of the neurological foundations of have no capacity for declarative memory can gradu. Clark RE and Squire LR (1998) Classical conditioning and brain cated in functional amnesia. amnesic patients and healthy individuals learn at the same The study of amnesia has helped to understand the gradual rate (Knowlton et al. Although no hippocampus. personal identity (See Chapter 2. Behav. 111: tual information about the world is often preserved. Behav. Often. Neuron 46(5): 799–810. 109(2): 195–203. 103(3): Functional amnesia. and Squire LR (1992) Intact artificial pieces of the past remain unavailable.04. neuroanatomy of remote memory. for example. Hopkins RO.04. Gold JJ. 2005a). is a dissociative psychiatric disorder that involves Cave CB and Squire LR (1991) Equivalent impairment of spatial and nonspatial memory following damage to the human alterations in consciousness and identity. and about how memory viduals to learn the correct object in each of eight works. to have a robust capacity for habit learning that Bayley PJ. neuroim- ally acquire trial-and-error tasks. Cahill L. the ability to recall information about the past. Hippocampus 1: 329–340. also known as dissociative amne. which requires indi. acquired disorder often clears. Declarative Memory System: Amnesia 77 strategies. Neurobiol. Experimental studies in patients. Neurosci. even though at the start of each session they cannot describe the task. and Squire LR (1989) Source may be unavailable. Severely amnesic eases and disorders that affect memory. Neurosci. 3. the patient is admitted to the hospital Int. and prevention of dis- pairs in one or two test sessions. grammar learning in amnesia: Dissociation of classification . patients acquire this same task over many weeks. Ramus SJ. the medial temporal lobe structures that are damaged Bayley PJ.. and sizable Knowlton BJ. trace eyeblink conditioned responses.

Neurosci. 675–677. and Squire LR (2005) Intact visual Stefanacci L. Squire LR and Zola-Morgan S (1983) The neurology of memory: Lee AC. Rev. Hopkins RO. Buffalo EA. Science 253: 1380–1386. Spiers HJ. Behav. Hopkins RO. pp. (2005b) Perceptual deficits The case for correspondence between the findings for in amnesia: Challenging the medial temporal lobe human and nonhuman primate. (2003) Recognition memory and the human hippocampus. Phillips J. Hippocampus 8: 205–211. Annu. medial-temporal lobe amnesia. and amnesia. and Amaral DG (1996) theory of recognition memory. individuals: An exemplar-based interpretation. Neurosci. Oxford University Press. the reciprocal connections between the monkey entorhinal McGlinchey-Berroth R. J. Science 262(5140): 1747–1749.) The ‘mnemonic’ view. Zola-Morgan S. 43(1): 1–11. Psychol. New York: Appleton-Century-Crofts. 22(1): 169–181. detection interpretation of remember/know judgments. Wixted JT. in bilateral. patient H. Brawn CM. (2005a) Specialization in Squire LR and Zola SM (1998) Episodic memory. J. 40: 385–398. and Staunton H (1996) E. 11(6): 794–796. New York: 9: 247–255. Reed JM. with and without awareness. Brain 124: Psychol. parahippocampal cortices of the Macaque monkey: Cortical Manns JR. Mem. Learn. Neuron 49(6): 805–813. et al. exemplar-specific information. Oxford: Oxford University Press. A neuroanatomical and neuropsychological profile of patient Maguire EA. 21(3): 699–710. Zola SM. Exp. Murray EA. et al. Wixted JT (2007) Dual-process theory and signal-detection Rempel-Clower NL. cortex and the perirhinal and parahippocampal cortices. Psychol. Neurosci. Schacter DL and Tulving E (1994) Memory Systems. Levy DA. J. Squire LR. Lee AC. memory system. Sci. 14: 1856–1877. J. monkeys. Brain Cogn. Thompson RF and Krupa DJ (1994) Organization of memory Nosofsky RM and Zaki SR (1998) Dissociation between traces in the mammalian brain. 114(1): Three cases of enduring memory impairment after bilateral 152–176. 866–77. Hippocampus. Philos. Cogn. 26(8): 2235–2240. and episodic memory deficits in a virtual town. 99(2): 195–231. 1(4): 217–229. ago is intact after hippocampal damage. J. Learn. Spiers H. Neurosci. O’Keefe J and Dostrovsky J (1971) The hippocampus as a Tulving E (1985) How many memory systems are there? Am. Soc. Psychol. and Squire LR afferents. 11(4): 616–641. Trans. Neuropsychologia 34: 993–1001. 12(1): 61–66. 100(6): Schuster. damage limited to the hippocampal formation. Neuropsychologia. Gabrieli JD.78 Declarative Memory System: Amnesia learning and explicit memory for specific instances. Neurosci. Mem. Maguire EA. scenes. Knowlton BJ and Squire LR (1996) Artificial grammar learning Squire LR (1992) Memory and the hippocampus: A synthesis depends on implicit acquisition of both abstract and from findings with rats. Cambridge. B Biol. Schmolck H. Neuron 49: Reber PJ and Squire LR (1994) Parallel brain systems for learning 459–466. New York: Levy DA. Knowlton BJ and Squire LR (1994) The information acquired Smith ML (1988) Recall of spatial location by the amnesic during artificial grammar learning. In: Deutsch JA (ed. Nature 400: Mishkin M (1982) A memory system in the monkey. 3: 172–179. Sci. Squire LR and Shimamura AP (1986) Characterizing amnesic LeDoux JE (1996) The Emotional Brain. Cogn. 20(18): 7024–7036. and Squire LR (2006) The O’Keefe J and Nadel L (1978) The Hippocampus as a Cognitive hippocampus supports both the recollection and the Map. Cogn. and humans. Topographical disorientation following unilateral temporal lobe Suzuki WA and Amaral DG (1994a) Perirhinal and lesions in humans. Burgess N. Neurosci. 6(10): 2950–2967. New York: Simon and patients for neurobehavioral study. recognition memory in patients with hippocampal lesions. Learn. Suzuki WA and Amaral DG (1994b) Topographic organization of Neuron 37(1): 171–180. J. Mem. Bussey TJ. Exp. Mem.M. 199–267. Lond. Tulving E (1983) Elements of Episodic Memory. semantic the medial temporal lobe for processing of objects and memory. Sci. Burke T. Carrillo MC. 34(1): 171–175. Preliminary evidence from unit activity in the Psychol. 298: 83–95. (2001) Unilateral Knowlton BJ and Squire LR (1995) Remembering and knowing: temporal lobectomy patients show lateralized topographical Two different expressions of declarative memory. Bull. 350: 497–533. Kitchener EG. Squire LR and Zola-Morgan S (1991) The medial temporal lobe Learn. Mamelak AN. Psychol. 7: 178–183. R. 20(1): 79–91. Mem. J. Ribot T (1881) Les Maladies de la Memoire [English translation: Rev. and Schuman EM (2006) Single-trial amnesia and the medial temporal region: Enduring memory learning of novel stimuli by individual neurons of the human impairment following a bilateral lesion limited to field CA1 of hippocampus-amygdala complex. and Squire LR (2006) Intact Parallel brain systems for item memory and category visual perception in memory-impaired patients with medial knowledge. Psychol. temporal lobe lesions. 15(6): 782–797. Brain Res. Gold JJ. P. the hippocampus. Neurol. Learn. . and Amaral DG (1986) Human Rutishauser U. freely-moving rat. 17: categorization and recognition in amnesic and normal 519–549. and Squire LR (2000) discrimination of complex and feature-ambiguous stimuli in Profound amnesia after damage to the medial temporal lobe: the absence of perirhinal cortex. Hopkins RO. Knowlton BJ and Squire LR (1993) The learning of categories: Shrager Y. Neurosci. spatial map. et al. Buckley MJ. Rev. J. and Disterhoft JF (1997) Impaired trace eyeblink conditioning Neurosci. Exp. Wixted JT and Stretch V (2004) In defense of the signal 16(16): 5233–5255. Comp. Wais PE. Squire LR. Rev. MA: MIT Press. Psychol. and Squire LR (2004) Impaired odor Academic Press. Diseases of Memory]. J. Mem. Behav. Hopkins RO. familiarity components of recognition memory. 2476–2489. Teng E and Squire LR (1999) Memory for places learned long 111(5): 873–882. Learn. Shrager Y. Physiological Basis of Memory.

1. by clarifying the relationships between memory. 3.04. the ability to remember facts 3.05.05.02 for an over.1 The Dm Approach 84 3.05.1. cognitive/representational. See Chapter 3. phenomena measured in these tests.2 Intracranial Dm Effects 86 3.1 Direct Comparisons between Recollection and Perceptual Priming 92 3. In 79 . and 3. while rehearsed and/or or subjective criteria (See Chapter 3. Paller. as listed in Table 1.2 Identification of Neural Correlates of Conceptual Priming 93 3. Expressions of declarative memory tend to coincide Analyses of memory in healthy individuals and in with the potential for making the metamemory judg- patients with memory impairments have revealed a ment that memory is being expressed – the awareness set of distinct memory functions that can be assessed of remembering. ety of noninvasive neuroimaging techniques. (1890: 648) termed secondary memory. remembering can be fractionated into a set ory phenomena.05.05. Taxonomies of memory have thus helped to memory phenomena that take place when informa- guide research into fundamental questions about mem.3 Characterizing ERPs 81 3. 3.05. we must seek a mind after a delay. and by showing how The neural substrates of remembering can be neurocognitive processes produce memory behavior examined in healthy human volunteers using a vari- and associated conscious experiences.1 Introduction Amnesic patients have specific impairments in declarative memory. other categories of mem- Rather. are not impaired of component processes that are expressed in differ.4 Advantages and Disadvantages of Using ERPs in the Study of Human Memory 82 3.2 The ERP Technique 80 3.05.3.1. however. For these reasons. tion that was initially encoded is later brought back to ory.4 Using ERPs to Contrast Memory Subtypes 91 3.4.05. ent combinations under different circumstances. L.2.4.05 Neural Substrates of Remembering – Electroencephalographic Studies J.05. as assessed in recall and Remembering does not refer to a unitary ability. (For a sum- ing the component processes in both cognitive and mary of research on primary memory or working neural terms.33.3 ERPs and Memory Retrieval 87 3. neural.4 Recognition with Pure Familiarity 90 3. A.1.1 Identifying Correlates of Recognition 87 3.1 Memory Subtypes and events from the past. declarative mem- using different memory tests.2.5 Future Contributions of ERP Studies to Memory Research 94 References 95 3. 3.05.1 Introduction 79 3.05. Beyond taxonomies. our emphasis here is on view). emphasizing Information can also be held in awareness for an either behavioral.05. USA ª 2008 Elsevier Ltd. In contrast.05.3. recognition tests.05.13.2 ERPs and Memory Encoding 83 3. Nonetheless.2 Recollection and Source Memory 88 3. in amnesia (See Chapters 2.1.05.1 Memory Subtypes 79 3.05.3.3 Postretrieval Processing 89 3. All rights reserved.05.) cognitive and neural descriptions. IL.12). Voss and K. manipulated.05. which is what William James comprehensive understanding of memory by describ. extended period of time.05. Evanston.3. Northwestern University. Various theoretical ory is usually regarded as fundamentally distinct from schemes have been used to categorize the memory other expressions of memory.

as well as the following) Perceptual priming Speeded or more accurate responses in Preserved if performance is not a priming test. We ongoing EEG signals can vary in magnitude on the emphasize memory processes that contribute to order of tens of microvolts over a few seconds. category learning. but we also include the related can be extracted so as to observe signals much smaller memory phenomena of priming (see the section than 1 mV. Psychol. These experiments cognitive processing unfolds. signal-averaging methods applied to EEG Our goal in this chapter is to examine how ERP recordings can be used to produce ERPs. ERPs elicited by stimulus events are connected to an amplifier system.e. EEG recordings from have made significant headway in identifying neuro. 13: 49–55. including cognitive skills accomplished without reliance on (e.1.g. ERPs thus reveal electrical signals produced during the course of stimulus processing. causing deficits facts (i. particular. locked to a particular class of stimuli is averaged in this given that dissociations between priming and way...05. classical conditioning. Curr.g. which is conventionally referred to as . Signal averaging can be performed with fields that vary moment to moment. recordings of event-related potentials voltage changes over time and can provide indica- (ERPs) have been used to monitor the activity of tions of the functioning of networks of neurons as the brain during memory tasks. based on item-specific contaminated by declarative memory or perceptual representations Conceptual priming Speeded or more accurate responses in Preserved in some cases. When these respect to any class of repeated event occurring at a electric fields are sampled via recording electrodes known time.. When a set of EEG responses that are time- titled ‘Using ERPs to contrast memory subtypes’). processing. ERPs declarative memory. Dir. based on association. episodic memory and in new learning and in remembering semantic memory) information acquired prior to onset of amnesia Immediate memory Information available while kept in mind Preserved if performance is not by continuous rehearsal (e. with permission. EEG signals unrelated to stimulus processing declarative memory have provided important clues tend to decrease because they do not occur at a con- about why declarative memory is distinctive.2 The ERP Technique such signals are not averaged out due to temporal Neurons in the human brain generate electric variability. reading mirror-reversed text) declarative memory (i. not most and motor skills skills learned outside laboratory circumstances) Adapted from Paller KA (2004) Electrical signals of memory and of the awareness of remembering. Sci. viewed as a time series of voltage changes following troencephalographic record – the EEG – shows stimulus onset.. verbal supported in part by retrieving working memory) declarative memories for the rehearsed information Nondeclarative memory (a large Generally preserved. particularly specific or conceptual across stimulus domains representations Skills Behaviors that improve gradually with Preserved when skill acquisition is practice.e. and in specifying processes engaged in association To examine EEG activity associated with stimulus with different memory feats. habit learning. but with some category that includes notable exceptions nonassociative learning. but further a priming test. to the extent that 3. the resultant elec.80 Neural Substrates of Remembering – Electroencephalographic Studies Table 1 A memory taxonomy based on findings in amnesic patients Type of memory Behavioral outcome Findings in patients with amnesia Declarative memory Recall and recognition of episodes and Impaired storage. electrodes placed on the scalp are used in a variety of cognitive processes that are responsible for memory clinical and research contexts. investigation is required. sistent time relative to the time of stimulus onset. Whereas research has shed light on various processes.

3 Characterizing ERPs terization of any of the individual components. difference-centered approaches have become components. Valid conclusions can after a click. tion of multiple components. on comparing two conditions distinguished by a task vals without making a priori assumptions about the manipulation. Luck. the identification and measurement of In ERP investigations of human memory functions. see misleading to assume that only a very small number Rugg and Coles. however. and most processes are associated with reliably different elec- importantly. Given this strict known ERP components to a focus on the thorough definition that requires a component to be identified understanding of relationships between cognitive with a unique neurocognitive function. 2005). the entire waveform is most likely Accordingly. the analysis on neurocognitive functions that can be tory evoked potentials produced during the first 10 ms manipulated across conditions. Although component amplitude and occur over a time interval of several identification can be informative. goal was to understand an ERP component per se. In lated based on theory-driven goals concerning specific practice. Other ERPs. Ideally. the experi- and the neural operations performed in response to mental manipulations play a critical role in focusing the stimuli.1. each bearing a unique and systematic prevalent. tive transactions over an extended time interval. central analysis question in any experiment is often ERP waveforms can nonetheless be quantified in to determine whether two ERPs differ reliably from several different ways. many ERP investigations in cognitive composed of the summation of neural activity from neuroscience no longer exemplify the strict compo- many distinct sets of neurons in the service of many nent-centered approach. However. When this approach is followed. relationships to experimental parameters.05. 1995. deflections are based on the summa. mining whether specific aspects of the ERP display a fication. There may ing with different methodologies. bringing together a variety of meth- be adequately measured by examining a deflection. as is bility of the EEG signals in question and on the likely the case in many of the interesting paradigms presence of other EEG signals and EEG artifacts of cognitive neuroscientists choose to study. amplitude. whereby experimental variables are manipu- relationship to a unitary neurocognitive function.’ ERP waveforms thus consist of a series of components that might be present. response factors. of components have been produced. but with an positive and negative deflections. distribution. significant progress. points. or . such as some associated with convincing correspondence with ERP components that cognitive processing. Putative components can be one another. it can be various sorts (for further methodological details. the component structure of an neurocognitive functions. Some ERPs. the experimenter may identified based on a combination of factors. conclusions are orthogonal to the challenge of deter- noise ratio sufficiently high to permit reliable quanti. a specific ERP components can be problematic. When a large ERPs can be most readily measured by examining number of ERP components occur simultaneously. effects of experimental variables in psychological Indeed. Neural Substrates of Remembering – Electroencephalographic Studies 81 ‘time 0. Typically. a complex ERP wave. it would be operations and neural events has facilitated a greater unwise to accept the assumption that each deflection dialogue between ERP experimenters and those work- corresponds to a particular component. stimulus factors. ods in cognitive neuroscience has been responsible for In other cases. given that some hundreds of stimulus repetitions to obtain a signal-to. form would be decomposed into a series of ERP Instead. including ask whether two or more hypothetical psychological latency. the timing and emphasis instead on differences between experimental waveshape of which vary with the nature of the stimuli conditions. averaging over in memory paradigms when a large number of compo- 30–100 stimulus events is required to observe reliable nents overlap with each other in the same time range. wherein a chief experimental different functions. Accordingly. This shift from a focus on ERP waveform is difficult to discern. such that the amplitude and latency of the composite peak does not provide a valid charac- 3. the positive and negative deflections (peaks and specific ERP components cannot always be isolated troughs) that occur at various poststimulus time from one another and separately characterized. it may not be feasible hundreds of milliseconds. depending on the amplitude and relia. such as the brainstem audi. In the context of be some cases when a hypothetical component may memory research. when subjects engage a wide variety of cogni- experiments. Experimental contrasts are often based ERPs can also be quantified for specific latency inter. polarity. which themselves are unknown. have very small amplitudes and require be drawn based on such analyses. can be several microvolts in have been described previously. In other words. trical signals.

amplitude will vary with a particular wave shape.05. Various ERP source-modeling procedures can never- theless be used when considering the anatomical location of the sources of ERPs.82 Neural Substrates of Remembering – Electroencephalographic Studies some combination. At any given latency. between these conditions can be visualized as a difference pensive compared to other neuroimaging methods. The time course of the difference in amplitude ERP methods are noninvasive and relatively inex. In most circumstances. Using ERPs in the Study of Human Memory although some investigators plot amplitudes in the opposite manner. Despite straightforward procedures for solving this so-called forward problem. Condition B Across a recording epoch. or by averaging over latency intervals of As just described. Models of electric currents and volume conduction of the head can be used to estimate the scalp topography that would be produced by activity at a certain loca- tion in the brain. represent the distribution of ERP differences between other methods are preferable for precise neuroana.1. A difference between ERPs can ERP grand average waveforms then be described and displayed (Figure 1). Condition A minus condition B Another important facet of an ERP difference is the distribution of the potential field across multiple electrode locations on the scalp. When an ERP difference between conditions is characterized in this manner. so both identifies the electrode location used for the waveforms locations are relevant for determining ERP charac. it is therefore extremely helpful to bring Figure 1 Visualizing ERPs. the 200–400 ms 600–800 ms inverse problem of determining the brain sources based only on the scalp topography is not soluble. These waveforms are plotted with time shown on the x-axis going from left to right. The head is tomical information. wave (middle). 1985. A and B. because many different configurations of intracranial generators can produce the same field on the scalp. Such +1 µV inferences involve many assumptions. This topographic information can help investigators make inferences regarding the responsible neural generators. or it may appear to encompass a different Difference waveform: time interval with a unique wave shape. 3. the +5 µV difference can have a positive or negative polarity. The black circle location and a reference electrode location. Waveforms averaged across multiple sources of evidence to bear on understanding experimental subjects for two experimental conditions. When theorizing about ERP sources Difference topographies in the brain. and with voltage on the y-axis. Here. Black dots indicate time-series of voltages between a scalp electrode the locations of recording electrodes.4 Advantages and Disadvantages of positive potentials are plotted in the upward direction. 0 1 2 3 Wilding. it may appear to correspond to a systematic enhance. shown. Condition A An ERP difference may be statistically significant over a certain time interval. These topographic maps schematically distributions on the scalp. some limited information about interest (gray shading) for many scalp locations and generating images of the voltage differences across the relevant neural sources can be extracted from ERP scalp (bottom). These images can thus demonstrate both temporal teristics. although drawbacks of inferring brain sources based on scalp recordings µV have been heavily debated (McCarthy and Wood. 2002. 200 400 600 800 1000 ment of one ERP deflection over a discrete time ms interval. with anterior scalp regions toward the top. the relevant brain structures or systems. approximated by a circular shape. are considered at a single electrode (top). 1997. ERP waveforms comprise a viewed from above. Kutas and Dale. . conditions A and B for the two latency intervals. 2006). Here we will emphasize results from and spatial characteristics of an ERP effect. Urbach and Kutas.

in the sense trial to trial. the signals by conducting analyses in the frequency EEG is sensitive to activity produced by a restricted domain. trials decreased. informing accounts of the relevant neural processing nomena amenable to examination. monly used paradigms concern memory for artificial tive events or. memory literature in the future and may also provide trode locations. during which the success of dictable manner or blocking experimental trials. Other oscillations can be recording electrodes are placed. ERP knowledge learning over many years. Less work has been devoted to this stimulus presentation and temporal blurring across challenge. because this recording method is used in a is unsurpassed by that of any other technique in majority of the relevant memory studies. Düzel et al. A that the EEG may not include signals from some of small but steadily increasing body of literature con- the active neurons engaged in relevant processing. tally created afterward.. The focus of most ERP studies of memory has been noise ratios and arguably are most useful. In general. Klimesch et al. ERP findings have better signal-to. These fields those available from analyses in the time domain. Critical memory processes often unfold within the first second after Experiments examining long-term memory generally exposure to a stimulus. The most com- signals may be better for isolating brief neurocogni. Such meth- influence observed ERP effects in accordance with ods hold promise for substantial contributions to the how currents are conducted through tissue to elec. 1998) are feasible with the laboratory. for defining a series of events in a laboratory setting.. 2003). Additional infor- Like other neuroimaging methods. must thus be generated by sets of neurons that are Oscillations that occur in phase with stimulus onset oriented together in such a way to produce electrical are generally apparent in EEG records and are thus fields that will conduct to distant locations where called evoked rhythms. Much neural activity time-locked to an event but be out of phase from may be electrically silent at the scalp.2 ERPs and Memory Encoding millisecond-by-millisecond basis. followed after randomized trial orders. it will ultimately be im- such procedures do not adequately constrain the portant to examine relevant processing that can take timing of relevant cognitive events.g. studies. Delivering trials in a pre- some delay by a test phase. 2001. Neural Substrates of Remembering – Electroencephalographic Studies 83 recordings using a mastoid or average-mastoid refer. stimulus events produce set of neurons. because the circumstances of acquisition can be .. cerns memory-related cyclic EEG and MEG activity A key advantage of ERP methods is that they (e. These studies are advantageous time. which can potentially be monitored by ERPs on a 3. other neuroimaging modal- reference that can be used during recording or digi. Some studies have randomized event-related designs with short inter. into neural activity that are complementary to duced by their activity can summate.05. examined autobiographical memories formed outside trial intervals (Burock et al. tion of ERPs makes them well suited to examinations of neural events responsible for human memory. as memory storage and retrieval is evaluated. and ERPs can allow these employ an encoding phase. quency-domain analyses of EEG and MEG activity thus hold great promise. further neuroanatomical insights. 2006. Rather. The superior temporal resolu. provide measures of neural activity with very high Prospects for combining ERP methods with fre- temporal resolution. such as viewing specific neurocognitive events that occur closely together in words or images. as potentials generated near this location monitoring near-infrared optical signals. Extended intertrial required during these stages. especially. For example. Like any cognitive neuroscience. place at various times between initial encoding and when cognitive events can be tightly time-locked to later retrieval. Given that declarative intervals can also be undesirable to the extent that memories change over time. Although ERPs provide temporal resolution that ence. ERP mea- required in positron emission tomography (PET) sures can be collected both at encoding and at test. during which subjects processes to be resolved in real time and with attempt to commit items to memory. or memories for general semantic functional magnetic resonance imaging (fMRI). can severely limit the range of memory phe. ities that also provide high temporal resolution.. ERPs do not mation can also be extracted from MEG or EEG provide a full view of neural activity. Although on explicit memory for episodes. these are called induced rhythms. the mastoid reference is not including magnetoencephalography (MEG) and inactive. These neurons are ones activated reliable EEG oscillations that may reveal insights synchronously such that extracellular fields pro.

it is possible that the artificial Observing reliable Dm effects generally requires nature of this research will become less problematic in that multiple criteria are met. During an encoding phase. neurophysiological difference measures found by In many Dm studies. 2002). a substantial polarization will be present first reported with ERPs by Sandquist and colleagues in successful versus nonsuccessful encoding opera- (1980) and can be traced back to earlier work using tions. Fabiani et al. Figure adapted from Paller KA and Wagner AD (2002) Observing the transformation of experience into memory. Dm has been observed in many paradigms.05. . These effects that can be demonstrated using these methods (e. component. Forgotten Leaflet. Dm effects will 3. Logically.. 1987). item variability in encoding strength such that a suffi- cient number of items are subsequently remembered and subsequently forgotten. Dm potentials observed with different for items later remembered. 6: 93–102. subsequently remembered and subsequently forgotten. Semantic elaboration is task. In the past two differences reflect variations in a known ERP decades.. positive potentials maximal sorting trials on the basis of subsequent memory over parietal regions were found to be greater for performance (Paller et al. Dm.. including some inter- placing limitations on interpretations. This term provides later-remembered items than for later-forgotten a convenient way to refer to the various phenomena items at approximately 400–800 ms. A comparison between neural correlates of these two conditions thus yields neurophysiological differences computed on the basis of subsequent memory performance. Subsequent-memory analyses can be including tests of recall and recognition (reviewed in conducted with different encoding requirements.1 The Dm Approach be greater when ERPs index a larger difference One way to isolate brain events responsible for suc. 1999. memory experiences. Indeed.. Performance on these tests is used to classify study-phase stimuli and corresponding neural measures into two categories. 1986). memory tests for these stimuli are administered. Sci. A ferent types of memory tests.g.. Wagner et al. have sometimes been attributed to ERP components Dm for free recall. Memory test for events Swamp. The term Dm has been used to refer to presented for encoding. Ideally. temporal dynamics must also be skin-conductance methods. between mean responses in these two conditions.. stimulus.g. and deeper semantic elaboration Encoding Retrieval Events Remembered Baboon. Test-phase behavioral data used to sort Subsequently remembered encoding phase ERPs Subsequently forgotten Figure 2 Schematic of the Dm or subsequent-memory methodology.. Furthermore. Dm cessful encoding is to acquire ERP responses during analyses can thus gain power when confidence or encoding and sort trials based on subsequent memory other graded measures of retrieval success are consid- performance (Figure 2). Dm for recognition.. and subject parameters can presum.. As laboratory extent that these operations partially determine what studies move closer to accurately simulating real-life information will later be remembered.84 Neural Substrates of Remembering – Electroencephalographic Studies carefully monitored and controlled.2. many sorts of suitable. well known to be effective for producing strong ep- ably index various neurocognitive operations to the isodic memories. Paller and Wagner.. such that processing that influences later neural signals can be used in subsequent-memory memory performance is well time-locked to stimuli analyses. Dm for pure such as P300 or the late positive complex (e. with permission. neuroimaging measures are recorded while subjects attempt to remember stimuli.. dif. Karis familiarity) and also avoids prejudging whether the et al. This general method was ered. 1984. During the subsequent retrieval phase.. Trends Cogn. and different retention favored cognitive hypothesis about these Dm find- intervals – and all of these parameters may influence ings is that they reflect superior elaborative encoding the results.

Figure adapted from Yovel G and Paller KA (2004) The neural basis of the butcher-on-the-bus phenomenon: When a face seems familiar but is not remembered. with permission. As shown in Figure 3. and relatively more positive ERPs were found to predict later memory. 1990). whereas a smaller right- lists of unrelated words by generating novel associa. a differ depending on the nature of the elaborative robust Dm that was bilaterally symmetric at posterior processes. 2004).0 0. whereas Dm for familiarity was smaller. briefer.. and restricted to right posterior locations. 1990. The magnitude of these potentials (a) Frontal +4 µV Parietal 200 400 600 800 ms Later familiarity Later occupation Later forgotten Later forgotten (b) Familarity – forgotten Occupation – forgotten µV 2. 2000). however. Friedman. retrieval possible during the subsequent memory test tials. sensitive to the amount of information bound into a In a recent study with faces. results revealed dis. ERPs were recorded when subjects studied these face/occupation pairs. it appears that left-frontal tex compared to elaborating on the inherent meaning potentials starting at approximately 500 ms can be of a single word. Creating novel associa. tions likely requires a larger contribution from Based on another type of Dm demonstrated in a working memory processes supported by frontal cor. and the scalp topography of these effects appears in (b). Neuroimage 21: 789–800. The nature of Dm effects due to elaboration can (Yovel and Paller.0 400–600 ms 600–800 ms 400–600 ms 600–800 ms Figure 3 Representative Dm effects for two categories of memory. in that frontal slow waves rather scalp locations was found to predict later recollection than late parietal-maximum positive potentials have (when subjects could recall information previously been observed when subjects attempt to remember associated with the face). Neural Substrates of Remembering – Electroencephalographic Studies 85 often corresponds with larger late posterior poten. These Dm effects are displayed at two representative electrodes in (a). few studies with words.0 3. information). with anterior oriented toward the top. .0 0. Comparing the two effects. Yovel and Paller (2004) categorized trials according to whether faces were forgotten or remembered with reference to the face alone with no contextual retrieval (familiarity) or remembered with retrieval of both the face and the paired occupation (recollection). Dm for recollection exhibited a bilateral topography with larger amplitudes spanning a longer time interval. sided Dm predicted later familiarity (when subjects tions to create relationships among items (Fabiani recognized the face but could retrieve no additional et al. Scalp maps are of the head viewed from above. memory trace at encoding (reviewed in Friedman tinct Dm effects depending on the type of memory and Johnson. Two different memory experiences can be assessed after people learn to associate novel faces with randomly assigned occupations.

to dissect the neurocognitive processes that support given that such operations can lead to better encod. A fruitful approach for future parahippocampal. including detailed perceptual analysis. Based on ory test. leagues (2002). Results thus identified the together due to the poor temporal resolution of specific contribution to encoding of mnemonic opera- fMRI. recordings were made from two areas Further research is needed to clarify how Dm varies within the MTL: the hippocampus and an adjacent as a function of stimulus materials and type of mem. In an ERP study reported by Fernandez and col- semantic elaboration. for example. By virtue of this directed-forgetting that precedes the onset of a stimulus has also been manipulation. 2002). Gonsalves and Paller (2000b) showed that ERPs at 3. Trials were categorized real-time temporal resolution of the ERP technique according to whether or not subjects subsequently with superior spatial localization. tonic brain activity ally forget others. seizure medication for many years to try to control ory for viewed pictures. The usefulness of remembered subsequently. tigation (Reber et al. together with adjacent encoding operations. Although encoding condition and stimulus onset by several hundred milliseconds. that is critical for the formation of declarative mem- cesses to determine the relative contribution of ories. rote rehearsal. however. In contrast. the hippocampus. revealing that these posterior ERPs sampled from a limited number of brain regions. and to precisely specify the relationship research with amnesic patients and with nonhuman between these electrical measures and specific animals. entorhinal. as were predictive of whether or not a false memory for electrodes are placed only where required for clinical that item would happen in the test phase. has been considered as essential circuitry directly manipulate hypothetical mnemonic pro. and so on.. large and more number of individuals who have typically taken anti- widespread ERPs were predictive of accurate mem. This approach has claimed (erroneously) to have seen a picture of the limitations. formation can be exemplified by a recent fMRI inves- tion of strategic encoding processes. future studies could examine the gamut of effec- extensively investigated. One hypothesis is that this approach for advancing explanations of memory these sustained positive potentials reflect the opera. memory formation. in conjunction with widespread neocortical . generalizability can be ques- revealed encoding activity that was partially respon. ERPs were found to reflect the vivid Electrodes implanted into the brains of patients who visual imagery when subjects visualized the referent are candidates for surgery to remove an epileptic of a word. cortical region near the rhinal sulcus. trieval were correlated in this study. ERPs thus purposes. Procedures are thus needed to separate fMRI tions guided by the intent to remember a word and correlates of these dissociable cognitive events. Results can On the whole. and perirhinal cortical neuroimaging studies. fMRI (2006). supported by left inferior prefrontal cortex. the probability of successful subsequent memory re- These electrophysiological findings have an impor.2 Intracranial Dm Effects encoding can reflect mistaken memories as well as accurate ones. Subjects were instructed In addition to stimulus-evoked neural processing to intentionally remember some words and intention- that leads to later remembering. tioned because recordings are made from a small sible for the later mistake. In this Although anticipatory Dm activity has not been way. whereas MTL activity was preferentially asso- stimulus-locked neural activity could be blurred ciated with success. Dm effects may index a group of nonetheless be used to inform theoretical accounts of encoding operations that lead to superior memory. For intention to remember were identified in addition to example. the neural basis of memory formation. one compelling hypothesis tive encoding operations in order to use Dm analyses is that it reflects working memory control operations.2. and this activity was found to be relevant focus provide a special opportunity to combine the to false remembering. both inferior prefrontal cortex and the medial tem- quently forgotten items in the interval preceding poral lobe (MTL).86 Neural Substrates of Remembering – Electroencephalographic Studies may be proportional to the amount of information different operations to Dm effects. abnormal electrical activity in the brain. ing of a declarative memory. ERPs over the front of the head were more Dm effects commonly include increased activity in negative for subsequently remembered versus subse. in that activity can only be visualized object. fMRI correlates of the differential linked to the efficacy of memory encoding. in a study reported by Otten and colleagues standard Dm effects based on retrieval success. would be to regions. prefrontal tant implication for investigations of encoding using activity was preferentially associated with encoding other methods such as fMRI. anticipatory as well as effort. mental imagery.05. In addition.

2002. tested. 1999. 2.. theories that posit two distinct recognition processes: tion and desynchronization was found via frequency.23). attempting to elucidate the specific memory processes 1998. such detail is retrieved. and McCarthy. EEG synchronization in various other has occurred in the past along with the retrieval of brain locations has also been observed to correlate specifics regarding the prior occurrence. the most Under some experimental conditions. Squire et al. The remember/know sible if some information was initially encoded. electrophy- with the extent to which words were processed siological responses are recorded while recognition is semantically. of any specific detail... 2005. in that source memory in each MTL region. Further studies are of knowing in the absence of the ability to bring to required to replicate and extend these various find. 2001). Engel Measures of neural activity obtained when people et al. 2001).. recollection and familiarity (See Chapters 2. Subjects are cued to introspectively classify their recognition of old 3. Trautner et al. 2004. extensively in attempts to identify neural correlates of recollection and familiarity.3 ERPs and Memory Retrieval stimuli as ‘remember’ if specific study-phase detail is simultaneously brought to mind or as ‘know’ if no Whereas veridical memory performance is only pos. Neural Substrates of Remembering – Electroencephalographic Studies 87 regions that are ultimately responsible for memory one can engage in the conscious experience of remem- storage (See Chapter 3. Paller that give rise to old/new effects. Furthermore. recollection and familiarity) qualitatively different memory processes (Eldridge are largely realized at retrieval. domain analyses of the same intracranial data (Fell Recollection involves the recognition that an event et al. For example. episodic memory (sometimes termed episodic mem- These results support the notion that distinct ory effects or old/new effects) are commonly MTL regions perform unique mnemonic operations.1 Identifying Correlates of predicted subsequent free recall of visual words. limited to item memory.g. familiarity denotes recognition of prior patterns in the hippocampus have been shown to occurrence that remains unsubstantiated by retrieval vary as a function of subsequent memory perfor. In addition. Recognition The rhinal Dm peaked approximately 400–500 ms after word onset and was thought to be associated In most ERP studies of memory retrieval.3. The hippocampal Dm. et al. Viskontas et al.. 2004). identified by contrasting ERP responses elicited by Other intracranial ERP studies have demonstrated a old items to those elicited by new items.g.05. whereas regions. mind any additional information. concerns the spatiotemporal context and various poral dynamics of interactions across multiple brain other features that can support recollection. respectively. 2006). in contrast. subjects must discriminate old binding of multiple features into memory following (repeated) items from novel items. as well as to explore the tem. the response categories have been taken as generic nature of memory is also a function of events taking indices of recollection and familiarity. Furthermore. 1988... place at the time of retrieval.. This is the time when et al. it is possible that results . Guillem et al. 2002).05..17. interesting distinctions between types of memory remember/know responses may correspond to vary- (e. will be very important for delineating the distinct An experimental procedure known as the contributions to memory formation dependent on remember/know paradigm has been employed different brain processes and regions. Familiarity can lead to a feeling mance (Cameron et al.. thereby with subsequent memory performance (Sederberg guiding the conscious experience of remembering. however.... Allison et al. In a recognition test for episodes studied during started later and was taken to reflect the successful an encoding phase.. ERP measures of memory formation. 2006). Potentials recorded from the rhinal region (entorhi- nal and/or perirhinal cortex) and hippocampus 3. single-unit firing In contrast. bering that can approximate reliving a past event. declarative memory and priming) and between ing degrees of memory strength instead of memory processes (e.03. a remember episodes are often interpreted in light of complex pattern of rhinal-hippocampal synchroniza. Much effort range of phenomena that may also reflect important in ERP studies of memory has been devoted to memory functions (Heit et al. ERP correlates of semantic analysis. in familiarity for a stimulus might be driven by retrieval combination with findings from these other methods. Grunwald et al. Recollection and ings in order to elaborate on the information familiarity are connected with the concepts of source processing steps that are performed by neurons memory and item memory. 2003. 1999.

. mental design have substantiated the association recognition confidence). repeated from earlier in the experiment. Late parietal ing of these ERPs during these years. given that recollection may be sup. particularly when word mean- 400–800 ms over much of the scalp.88 Neural Substrates of Remembering – Electroencephalographic Studies obtained from this procedure are highly influenced An early study that convincingly associated by nonmnemonic variables. subjects were cognizant of specific contextual information with respect to some 3. late parietal ERPs with recollection. is not always a good indicator and was only present for words that were successfully of familiarity. and asked to recall the original gender. 2000). 2000. These findings are outlined in the debriefing showed that subjects noticed that words following three sections. ponents related to recollection and familiarity is to the same level of priming (Table 1) was observed on test memory for specific source information based on an implicit memory test of word identification for the experimental context at encoding. In general. contextual retrieval (Smith in Friedman and Johnson. however. behavioral evidence obtained at ory retrieval. Unlike typical old/new ported by retrieval of information other than the ERP effects. endorsed a variety of hypotheses concerning their 1999). both encoding tasks. 1987). and processes that do not con. focused attention on letter information. with methods. memory paradigms have also been used to associate 1989).. relative ories linked with conscious remembering (reviewed familiarity (Rugg. from the encoding phase appeared during the word- identification test. 1989). Halgren and Smith. Despite this lack of consensus about the mean.g. a common ERP amplitudes are often found to be greater for assumption was that the effects included modulation remember compared to know responses and know of two ERP components: N400 potentials and P300 responses compared to new trials.2 Recollection and Source Memory of the words in the word-identification test that The most consistently reported finding in recogni. and Halgren.. there have not been convincing demonstrations that . Early investigations of these old/new ERP effects and other types of stimuli such as faces (Paller et al. 1985). Compared to recognized between ERPs and recollection and extended the items (hits).. Further studies using the same strategy in experi- ory strength based on various behavioral indices (e. including associations with signals of the successful retrieval of episodic mem- memory strength ( Johnson et al. In other words. Importantly. types of mnemonic processes associated with mem. Confidence in results can be Behavioral results showed that this manipulation increased with convergent support from multiple influenced recall and recognition performance. For example. Furthermore. even though this was irrelevant to their task. These effects ing had been encoded deeply. differences generally increases with increasing mem.3. differences associated with priming because priming Distinct ERP effects have been linked to three was matched between the encoding conditions. this effect could not be attributed to specific source information in question. These late parietal ERPs can thus be taken as functional significance. 1995). 2000a). In contrast. a useful approach to separating ERP com. Incorrect source retrieval with correct based on encoding task began at a latency of 500 ms recognition. This ERP difference collection. The authors tion studies is that ERP amplitudes to old items are thus inferred that incidental recollection took place greater than those to new items from approximately during the test phase. such as the capacity to ERPs with recollection utilized a levels-of-processing introspect accurately.05. ERP correlates of recollection. potentials (e. identified in the test phase. and corresponding differences were interpreted as correct source retrieval can be used to indicate re. Paller. applying this method. and that ERPs were typically show a maximal difference over midline or sensitive to the differential processing associated with left parietal scalp locations. auditory modality (Gonsalves and Paller. words presented in the rejections) tend to elicit smaller ERPs.g. such as with memory judgments based on superior memory following semantic encoding that source information. 1990). 1992). ERPs recorded during the im- subjects may encode words spoken by either a male plicit memory test were compared between the two or female voice and later be tested with visual words conditions defined by the task assigned at encoding. Results from remember/know as well as source- tribute to recognition judgments (Rugg and Nagy. The amplitude of these recollection. required visual imagery compared to encoding that Indeed. Great care is thus needed in manipulation at study (Paller and Kutas. both old items that are forgotten (misses) results to the use of other encoding tasks and memory and new items that are correctly identified (correct tests (Paller et al..

each also confirmed associations between successful epi. patients exhibited impaired conscious recognition as Behavioral results showed that recognition was better for well as reduced or absent late positive amplitudes Remember faces than for Forget faces. In the highly demanding for both successful and unsuccessful retrieval test. in that they tend to be similar retrieval of perceptual detail.. 2006). Brain Res. Experimental manipulations of retrieval demands.g. Instead. Amnesic information when cued by the corresponding face at test.. 2006a). however. these frontal potentials are thought identical format. prior to encoding created a temporary state of and Yamada S (1999) Brain waves following remembered amnesia..g. partly due to the (Figure 4). 2006). 2000. Remember face was presented with a short biographical vignette. or imag- ined actions (Senkfor et al. 2002). Neural Substrates of Remembering – Electroencephalographic Studies 89 the distribution of late parietal ERPs differ between remember and know responses. Late frontal potentials were larger mnemonic processing such as further evaluation.. Grabowecky M.. watched. manipulation objects were to be endorsed as old regardless of any of working-memory contents. admin. 1996. 7: 519–531. spatial location (Van Petten et al. Ranganath C. 1999). At encoding. objects were presented again at test in either the ning approximately 500 ms after stimulus onset and identical format or perceptually altered (Ranganath extending for up to several seconds..05. In addition. Subjects were highly accurate at recollecting this sodic retrieval and late positive ERPs. Wilding and Rugg.. both recol. 600 200 In addition to corroborating the connection between recollection and late posterior potentials.. and these positive potentials at prefrontal scalp locations begin. In one study. whereas in the less-demanding test. and/or postdecisional format alterations.3 Postretrieval Processing operations indexed by these potentials. Similarly. faces) were greater than for faces that were encoded with Results from memory-disordered patients have the instruction to forget (Forget faces).3. Thus. 700 100 and incorrect source judgments greater than new Remember faces– Forget faces trials. subjects responded ‘old’ only to objects in the attempts. Curran et al. background figures Figure 4 Late positive potentials index conscious recollection. and stimulus color (Cycowicz et al. Bozic VS. disrupted (e. faces index conscious recollection. absence of a direct connection with retrieval success.. ms after the presentation of faces that were previously 2000).g. Experimental 500 300 contexts have included: speaker gender (Senkfor 400 and Van Petten. whereas priming did (e. Smith. in the highly demanding test than in the . –1 µV 3 µV 1999). encoded with the instruction to remember (Remember among others. They do not tests that differed in the demands placed on effortful index retrieval success. Brain lection and late parietal potentials were severely Res. indicating that these ERPs may index a neurocognitive operation that differs only quantitatively between recollection and familiarity conditions (e. 1998). without qualitative differences in scalp distri- bution (e. and following this manipulation. evidence taken from a variety of experimental para- digms thus converges on the conclusion that recollection is a distinct expression of memory that The functional significance of these late frontal reliably occurs with a particular ERP signature potentials was difficult to decipher. Another memory phenomenon may be indexed by images of common objects were encoded. were useful for clarifying the cognitive 3. 0 ms correct source judgments elicit greater late parietal amplitudes compared to incorrect source judgments. 1993). positive potentials from 300 to 800 (Guo et al... These potentials and Paller. not differ. Data are from Paller KA. performed. Cogn. many experiments have targeted ERP correlates of source and item memory in order to understand contextual memory in its own right. these late positive potentials index istration of benzodiazepine drugs to healthy subjects recollective processing uncontaminated by priming. Olichney et al. Taken together. Trott et al. Subjects performed one of two often display a right-sided distribution. 2001).g. Posterior. to index effortful retrieval processing.

Figure adapted from Ranganath C and Paller On the other hand. 2006) and have been found not to scale with the objects that were perceptually altered from one in the study amount of recollection during tests of source memory phase (old/different). 2006b). there is reason to doubt this KA (1999) Frontal brain potentials during recognition are modulated by requirements to retrieve perceptual detail. N400 repetition effects in amnesic patients (Olichney supporting the view that late frontal potentials track et al.. the frontal N400 old/new effect is generally processing.. compared between a highly demanding (specific) Rugg et al. are sensitive to recognition success but not to manip- These late frontal potentials were maximal over the left ulations that effect recollection. with permission from Elsevier. in the sense that a patient cess likely mediated by prefrontal cortex. 1998a. Thus.. anterior scalp. same Beginning with the work of Düzel and colleagues Old/ (1997).90 Neural Substrates of Remembering – Electroencephalographic Studies Specific test negative potentials peaking approximately 400 ms General test Specific – General poststimulus.4 Recognition with Pure Familiarity potentials might instead reflect the operation of con- Attempts to associate the memory experience of ceptual implicit memory processes that can be familiarity with specific ERPs have been most con. Düzel et al. and so thus appeared to track retrieval effort as manipulated the inference made has been that they therefore across these two recognition tests. sia disrupts familiarity. a pro.b). A reasonable generalization is that amne- strategic processing that accompanies retrieval. In troversial. special steps are necessary to isolate the to indicate that familiarity is generically indexed by contribution of separate but potentially co-occurring . Olichney and col- leagues (2000) thus suggested that frontal N400 3. 2007. Yonelinas et al. frontal N400 potentials corresponding ERP topography was computed over all conditions for the latency interval from 500 to 1200 ms. Specifically. generalization about frontal N400 potentials. Hemispheric with severe amnesia does not behave as if people loci of late frontal potentials can also vary as a function and various objects they encounter feel familiar. engaged incidentally during recognition testing. A direct challenge to this fragile gations have confirmed this interpretation (Ranganath interpretation arose with the identification of intact et al. This generalization also stands on sound empirical panying an increase in the complexity of the retrieved footing (Knowlton and Squire. but in other experiments have also been described in the extant ERP literature as a general found to be bilaterally symmetric or maximal over right correlate of familiarity. following ms semantically deep vs. with a right-to-left shift accom.. information ( Johnson et al. Subsequent investi.g. Whereas late 4 µV posterior potentials are greater in circumstances in 0 400 800 1200 –0. 2000..e..g. In addition. one might expect patients with amnesia to exhibit reduced N400 old/new effects if these effects indeed index familiarity. more positive ERPs) for old com- Old/ 1. of retrieval demands. semantically shallow encoding Figure 5 Late frontal potentials associated with retrieval tasks).. 1997. Woodruff et al.g. 1998). 2002). These potentials behave as a neural correlate of recollection. in contradistinction to recollection New + (reviewed in Curran et al.3. Waveforms (left) showed a relative positivity for the specific test compared to the general test for all three familiarity as indexed by the remember/know stimulus classes in the experiment: objects that were paradigm (e. many researchers have proposed that the different frontal N400 old/new effect indexes recognition with familiarity.05. Nolde et al. 1995. frontal N400 poten- recognition test and a less-demanding (general) recognition tials have been associated with phenomenological test. 2000). for a review). index familiarity. Wilding.. an N400-like potential with reduced amplitudes (i. the logic of interpretation has generally been that frontal N400 less-demanding test for both categories of old stimuli potentials reflect recognition memory and do not as well as for new stimuli (Figure 5). the majority of findings used to argue in favor of this association are indirect (see following and Paller et al. and thus are widely anterior scalp. ERPs to drawings of common objects were found to be insensitive to such manipulations (e. Results from many studies have been taken general... 1997.. and entirely novel object pictures. 2000..3 µV pared to new items. The (e.. especially at anterior locations. In sum.3 µV which recollection is greater (for example.. Ranganath and Paller. 1998). perceptually identical to one in the study phase (old/same). because Neuron 22: 605–613. Leynes. 2000).

positive than to new items. independent of conscious N400s (Tsivilis et al. More evidence is needed to determine if when seen in an unusual context. Neural Substrates of Remembering – Electroencephalographic Studies 91 memory phenomena to neural correlates of recogni. 2003.33). Indeed. Like FN400 potentials. plicit test of memory. such as on the bus. in contrast. recognition with recollection co. as described earlier. 2. specific reference is made to which repeated faces engage conceptual priming. faces with different racial Here we will highlight one distinction that has and ethnic features) and failed to replicate this pat. but rather.. identifying the occurrence of these variable contribu- lar study and found statistically significant tions and thereby disentangling the operation of topographic differences. The butcher’s face reports of familiarity and does not scale with recol- may seem extremely familiar yet not be identified lection.05. but otherwise closely resemble FN400 tion memory. no reference is most studies of familiarity experiences during face made to learning episodes. Yovel and Paller (2004) used a variation of the remember/know paradigm to seg- regate trials for separate analyses of recollection and 3. Priming Results from a limited number of studies can be is a form of implicit memory that is indexed behav- taken as tentative evidence that familiarity may be iorally as faster or more accurate responses on indexed by potentials occurring earlier than frontal specialized priming tests.g. 2004. occurred with late positive ERPs that were much Multiple memory systems or processes make variable larger in amplitude. spanned a longer time interval. produce memory abilities. 2005). Diana et al. tive interpretations of frontal N400 potentials are On the other hand. verbal versus facial stimuli. Curran and Dien. we must first with late posterior ERPs and recollection with larger come to understand these separate components before and more anterior ERPs.e. familiarity was associated distinct memory components.. Contrasts between these able objects. these potentials indeed index familiarity as opposed whereas in the butcher’s shop. ERP investigations are especially well-suited for MacKenzie and Donaldson (2007) conducted a simi.. Friedman. neous mixture of faces (i. 2007) used a more heteroge. These potentials occur between 100 and ing (also called item-specific implicit memory). it is plausible that characteristics implicit memory known collectively as priming.4 Using ERPs to Contrast familiarity. been particularly amenable to investigation with tern. co-occurred with late positive ERPs that were maximal at midline Memory performance undoubtedly reflects the opera- parietal locations. The most common types Duarte et al.12. compared Memory Subtypes to correct rejections of new faces.. of memory search (e. memory is processing associated familiarity with late posterior demonstrated via a change in performance in a cer- ERPs. Diana et al. such as various forms of priming and the initiation episodic and semantic information that uniquely iden. as in the with familiarity based on indirect evidence: the effect classic example described by Mandler (1980) as the is present in association with phenomenological butcher-on-the-bus phenomenon. . However. familiarity is more to other potentially co-occurring memory phenome- likely to occur together with memory for additional na. In an im- although this idea deserves further study. as discussed in detail in the next potentials – frontal ERPs to old items are more section. 300 ms. have been shown to be preserved in amnesia. In sum. disrupted in cases of amnesia. memory for study episodes. In of the people shown could influence the extent to an explicit memory test.02). Several studies have examined the phenomenon of these earlier frontal potentials have been associated familiarity without recollection using faces. tasks. Of course. instead attributing an N400 effect to familiarity ERPs.. tion of a variety of neural systems (See Chapter 3. tifies the specific person. consciously remembered.. that between explicit memory and forms of for faces. Recognition with familiarity. not with the earlier frontal N400 potentials tain task due to a prior event that may or may not be described in prior studies that used words or name. 2004. many types of implicit memory viable. that heterogeneity of Performance on explicit memory tests is typically face stimuli plays a crucial role. of priming tests are used to measure perceptual prim- 2005). or whether alterna. contributions to performance on different mnemonic and showed a slightly more anterior topography. 2001. Further studies are needed to determine two broad categories of memory phenomena have whether this divergence can be explained by showing been very prominent in memory research over that familiarity entails different neural events for the past two decades (See Chapters 3. a third study we can work out how their interactions ultimately (Curran and Hancock. remembering information learned earlier.

dissociations identified in amnesic patients. While maintaining central fixation. ERP were thus associated with conscious remembering or other neuroimaging results cannot be unequivo. selectively influence the operation of distinct compo. Paller and colleagues (2003) used a condition also occur. Experimental manipulations that mately 200–400 ms after face onset (Figure 6(b)). whereas perceptual In either implicit or explicit memory tests.. tated or more fluent perceptual processing of the subjects attempted to discriminate between two subtly physical features of repeated items. subjects in which faces were encoded only to a minimal extent indicated using strict criteria whether they recog- such that priming occurred in the absence of recogni. ERPs can priming was associated with a relative ERP negativity reflect neurocognitive processes responsible for both over anterior recording electrodes from approxi- types of memory.e. This pattern of neuro- cally associated with one type of memory versus imaging findings complements neuroanatomical another. Perceptual priming for these faces. and were recognized sense.. Subjects viewed each face for 100 ms at a central were categorized as showing priming if the subject location while simultaneously a yellow cross was produced the word at the completion stage but failed shown unpredictably in one of the four quadrants to endorse it as an old word (i. tials (Figure 6(a)). ciated with conscious memory for faces and ERPs cesses. conceptual priming..e. sented in an explicit memory test (i.05. tions from recognition processes would be negligible.. association-specific prim. distinct from different types of yellow crosses. Isolating neural correlates of perceptual priming Evidence for the independence of implicit and uncontaminated by those of conscious remembering explicit memory can also be derived from contrasts is problematic because of the difficulty of prevent. whereas contribu- investigation. In this minations or backward masking. 2000). however. neural mea. novel-information priming. 1999. Schott priming tests. was new-association priming. the results imply that implicit access to memory is supported by processing within a network of brain regions that is 3. recognition tests. but subjects were encouraged to guess if they could In order to isolate ERP correlates of perceptual not remember a studied word so that priming might priming. Spatiotemporally distinct ERPs of opposite polarities nents of memory are thus essential. Trials tion. Similarly. nized the word from the encoding phase. without disruptive perifoveal visual discri- implicit memory. Recognition- such as ERPs are liable to be influenced by multiple related neural correlates included late positive poten- memory processes as well. Moreover.4. and this processing can assess memory. Paller. it is important to note that neural measures associated with perceptual priming. priming-without- . between neural correlates of encoding responsible for ing subjects from recalling prior episodes during later perceptual priming versus recollection. versus perceptual priming. (Paller et al. observed behaviorally on two implicit memory tests.. nying recognition. memory tests may not be ‘process-pure. ven if behavioral measures of followed by an ingenious two-stage procedure to priming are not obtained.e. Faces in another condition were presented for a longer uals. A different set of mechanisms subsequent test. these faces could conceivably reflect neural events although this is a topic currently under active responsible for perceptual priming.1 Direct Comparisons between qualitatively distinct from that supporting conscious Recollection and Perceptual Priming access to memory. These two conditions thus addition to acknowledging that behavioral measures provided a direct comparison between ERPs asso- in memory tests can reflect multiple memory pro. closely resembling responses sures can reflect memory processes whether or not previously associated with face-cued recollection those processes influence behavioral performance. On a the item occurred. 2002) used deep/seman- that supports perceptual priming may occur during tic versus shallow/nonsemantic encoding conditions. performance may be guided by explicit memory. duration. cessed faces was not significantly better than chance. When memory tests are given to healthy individ. or by some combination. or cross-domain priming). and further stimulus accessing a memory for the full episode in which processing was disrupted via backward masking. cued recall). recognition of these minimally pro- may be responsible for some types of priming (i. After each stem was completed. Other-wise.’ In at above-chance levels. 1. and in some of these The logic of this design was thus that ERPs elicited by cases priming may not be preserved in amnesia. the automatic processing and colleagues (Schott et al.92 Neural Substrates of Remembering – Electroencephalographic Studies These behavioral effects are thought to reflect facili.8 from fixation. Three-letter word stems were pre- potentially be reflected in neural measures accompa. ing.

Light colors indicate positive difference potentials in (a) and negative difference potentials in (b)..g. raised the possibility that fron- 3.7 –1. it is vals with different topographies. These Dm for priming that took the form of a relative ERP alterations of behavioral responses are thought to negativity over central and fronto-central locations reflect facilitated processing of stimulus meaning. and Boehm SG (2003) Neural manifestations of memory with and without awareness. Collectively. The ERP difference between recollected faces and new faces is displayed in (a). Paller et al. The finding that simi- lar potentials are intact in amnesic patients (Olichney et al. recognition). Differences are averaged over 100-ms intervals starting at the latency indicated underneath each topographic map. during memory testing. 2003).2 Identification of Neural Correlates tal N400s instead reflect a form of memory that is not of Conceptual Priming disrupted in amnesia. frontal N400 potentials at retrieval tionship between expressions of explicit memory and have been postulated to index the explicit memory expressions of perceptual implicit memory. regardless of whether a behav- included relatively positive potentials at later inter. 2000). which both stimuli are repeated. Neuron 38: 507–516.1 –. It is thus possible that frontal N400 .2 –. 2006) constitute critical recognition memory for meaningful stimuli.0 –1. Neural Substrates of Remembering – Electroencephalographic Studies 93 (a) (b) 0 ms 0 ms 700 100 700 100 Perceptual Recollection priming –1 0 1 2 3 4 5 .3 600 200 600 200 µV µV 500 300 500 300 400 400 Figure 6 ERP correlates of recollection and perceptual priming. these possible that neural activity associated with concep- results (along with those from a follow-up study tual priming occurs incidentally during tests of using fMRI. ioral test of conceptual priming is provided. Trials were categorized as remembered are similar to those of perceptual priming in that they when the correct response was made at both stages can occur in the absence of awareness of remember- and as forgotten if not produced at the completion ing and typically take the form of faster or more stage. Schott et al. Subsequent-memory analyses thus revealed a accurate responses to a specific stimulus. ceptual priming can occur whenever meaningful ing as well as from Dm for recognition. capability termed familiarity..05. Hutson CA. Furthermore. Miller BB. approximately 200–400 ms after word onset (resem.4 –. e. however. The difference between primed but not remembered faces and new faces is displayed in (b). As first steps in characterizing the neurocognitive rela.. Behavioral measures of conceptual priming potentials. with permission from Elsevier.4. such as language comprehension. and they potentially support some of the short-term bling ERP correlates of perceptual priming identified mnemonic operations that are preserved in amnesia. reviewed above. proposed that residual conceptual priming in amne- ing can occur whenever meaningful stimuli are sic patients could be reflected by frontal N400 repeated. Dm for priming was distinct from Because the neural processing that supports con- ERP differences between deep versus shallow encod. Olichney and colleagues (2000) Another form of priming known as conceptual prim. Figure adapted from Paller KA.5 ..

but doing so is critical for under- and Paller. Data are from Voss JL and Paller KA (2006) Fluent conceptual processing and explicit memory for faces are electrophysiologically distinct. Conceptual priming was ma. Conceptual prim.77. vant neural events that precipitate remembering. Therefore. The magnitude of these potentials (quantified in each subject at the electrode exhibiting the greatest frontal N400 conceptual priming effect) was correlated across subjects with the magnitude of conceptual priming indexed behaviorally (right). the hypothesis that frontal conceptual priming that occurs concurrently with N400 potentials are unique neural signatures of explicit memory (Paller et al. electrophysiological recordings were obtained multiple behavioral measures that can allow for while subjects rapidly discriminated celebrity faces valid associations between neuroimaging measures from other faces. Neurosci. biological.94 Neural Substrates of Remembering – Electroencephalographic Studies potentials do not index familiarity but instead reflect fMRI. RT. Electrophysiological responses were obtained during the discrimination test and were 3. p < 0. and activity related to conceptual implicit memory is com. 2006). reaction time.5 0 µV 1 Frontal N400 0.2 3 0. A baseline was provided by performance. tion of neural signals is also possible in studies using ERP techniques have an important role to play in other methods to measure brain activity. might partially (or entirely) reflect the operation of One recent study directly examined this issue implicit memory. J.5 1. this contamina. However. because it is needed to disentangle these two memory functions. . such that this information consisted of faster and more accurate responses to can be used to build an accurate characterization of the subset of faces previously presented with biogra. 2007). This approach further highlights the neces- information along with a subset of celebrity faces. sity of employing experimental manipulations and Later. the brain processes that support human memory phical information.4 1.5 0 10 20 30 40 50 60 70 80 Behavioral conceptual priming (RT. Much work will be needed to by using celebrity faces to elicit neural correlates accurately elucidate the neural substrates of these of conceptual priming and explicit memory (Voss memory processes. standing the neural substrates of familiarity and of nipulated by presenting associated biographical priming. phenomenological facets of memory. a counterbalanced set of celebrity faces that were previously presented without corresponding bio- graphical information. whereas explicit memory was characterizing corresponding neural substrates of related to late positive potentials at posterior locations.05. Evidence for conceptual priming and memory functions. The ERP contrast between famous faces based on whether corresponding conceptual information was primed in an earlier phase of the experiment is shown topographically for the latency interval from 250 to 500 ms (left). such as this endeavor. Given that a monly produced in memory experiments designed to memory cue can unleash such a rapid flood of rele- monitor explicit memory. there is much more to be learned These results attest to the likelihood that neural so as to demystify the cognitive.5 2 0. The frontal N400 effect is indicated. Further work familiarity must be called into question. memory..5 Conceptual priming Frontal N400 amplitude (µV) 1. ms) Figure 7 Neural correlates of conceptual priming include frontal N400 potentials.5 Future Contributions of ERP characterized according to both conceptual priming Studies to Memory Research and ratings of explicit memory for celebrities obtained in the last phase of the experiment. 26: 926–933.6 4 r 2(8) = 0.8 2. Much progress has been made in identifying compo- ing was strongly associated with frontal N400 nent processes of human memory capabilities and potentials (Figure 7). Furthermore. 4.001 250–500 ms 3.

Cogn. Düzel E. Karis D. Brain Res. I: with recollective processing of spoken words. Jr. J. 51: 6–28. Neuroimage 36: 464–471. and Knowlton BJ (2002) The effect of neural dynamics of explicit and implicit forms of testing procedure on remember-know judgments. Schott B.) Handbook of amygdala. Guillem F. Sci. 467–492. Buckner RL. Fell J. and Donchin E (1990) Effects of mnemonic tions of memory research. Helmstaedter C. Brain Res. Neuron 30: 289–298. Whereas our understanding of such spatiotemporal analysis of electromagnetic time-frequency interactions has hitherto derived primarily from be. play in conjunction with other methods of cognitive Diana RA. Cogn. Friedman D and Johnson R. performance. remembered. Natl. Curran T. Bull. Sci. Natl. Neurosci. Heinze HJ. J. and Piatt C (2006b) ERP explorations of dual Heit G. Valid descriptions of the neural bases and USA 94: 5973–5978. 4: 1259–1264. Tepe KL. Hum. and Elger MRI. Heinze HJ. modulatory processes in human memory formation and based recognition of faces. and Knight RT sions. In addition to elucidating differences correlate of a recognition memory search attempt. Microsc. Oxford: Oxford University factors that make implicit and explicit expressions of Press. Neuroimage 18: 185–197. and Ojemann GA (2005) Invasive recordings from the human brain: Clinical insights and The detailed theoretical account of memory that beyond. allow for extremely rapid presentation rates using functional Grunwald T. a deeper understanding of this issue will allow us (2004) Dissociable neural correlates for familiarity and to characterize how these distinct forms of memory recollection during the encoding and retrieval of pictures. Sarfatti S. Tech. and Snodgrass JG (2001) this regard and thus have a continuing role to Remembering the color of objects: An ERP investigation of source memory. tional relationships. Rev. Neurobiol. Cogn. 6: 35–47. memory are essential for understanding their func.and long- Curran T. Woroch B. 1: 81–121. 6: 129–139. Klaver P. Clin. Fabiani M. retrieval. Mangun GR. physiological data are of conscious awareness in memory. Klaver P. Fell J. and Lindenberger U (eds. Hayward D. Spencer DD. rehabilitation are critical in neurological and psy. also needed. CE (1998) Verbal novelty detection within the human Cameron KA. and Halgren E (1988) Neural encoding of processes in recognition memory. Rev. et al. individual words and faces by the human hippocampus and Mecklinger A. Puce A. Neurosci. Nat. Cortex 9: 415–430. 26: 437–458. Neurosci. Proc. Brain Res. 18: 255–272. Psychon. Acad. 30: 61–87. 11: dissociation of familiarity and recollection. Moll CK. Rougier A. and Tulving E havioral evidence obtained from healthy individuals (1997) Event-related brain potential correlates of two states and memory-impaired patients. between conscious and nonconscious memory expres. 1118: 142–154. understanding of ourselves. Grunwald T. Woldorff MG. and Fried I (2001) Human hippocampus proper. et al. (2001) Human memory chiatric diseases that affect memory and during the formation is accompanied by rhinal-hippocampal coupling course of healthy aging. Cereb. and episodic priming. Among the possible practical applica. and McCarthy G (1999) Gonsalves B and Paller KA (2000a) Brain potentials associated Electrophysiological studies of human face perception. Biol. everyday life. Duan L. Psychophysiology 23: 298–308. Gonsalves B and Paller KA (2000b) Neural events that underlie Burock MA. and Hirshman E (2006a) delay intracranial ERP repetition effects dissociate memory Combined pharmacological and electrophysiological systems in the human brain. Mem. retrieval. 9: 139–145. Ranganath C. Explorations of remembering via electrical and hippocampal contributions. recordings of brain activity may thus lead to a better Friedman D (1990) ERPs during continuous recognition memory for words. Yashar S. data of recognition memory. Guo C. and Paller KA (2006) Distinguishing source Curran T and Dien J (2003) Differentiating amodal familiarity memory and item memory: Brain potentials at encoding and from modality-specific memory processes: An ERP study. and Reder LM (2005) Identifying the ERP neuroscience. Psychophysiology 40: 979–988. Engel AK. 24: 674–684. Wilson CL. Neural Substrates of Remembering – Electroencephalographic Studies 95 One theoretical keystone involves identifying Binding and Memory: Perspectives from Cognitive Neuroscience. Cognit. memory unique. and Dale AM remembering something that never happened. Nat. Karis D. can be achieved in this manner will ultimately be Fabiani M. DeBuse C. and Claverie B (1999) Short. Vilberg KL. Li W. (2000) Event-related potential References (ERP) studies of memory encoding and retrieval: A selective review. Fernandez G. recollection. Potentials generated in occipitotemporal cortex by face and 28: 321–330. possibilities for memory strategy manipulation in a Von Restorff paradigm. pp. Smith ME. . Allison T. Much of human experience and decoupling. Habib R. Acad. ERPs have been extremely useful in Cycowicz YM. Nat. and Elger CE (2002) revolves around bringing to mind events from the Human declarative memory formation: Segregating rhinal past. Res. Lehnertz K. Lehnertz K.Neurosci. interact in a variety of situations to drive memory Brain Res. In: Zimmer HD. Friedman D (2004) ERP studies of recogntion memory: Differential effects of familiarity. Rosen BR. Winward L. Fried I. Hippocampus 12: 514–519. (2003) A multivariate. Cereb. (1998) Randomized event-related experimental designs Neurosci. Duarte A. Cortex 11: 322–334. 3: 1316–1321. Yonelinas AP. Eldridge LL. Sci. 1979–1985. Proc. Electroencephalogr. 75: 22–35. Düzel E. non-face stimuli. Halgren E and Smith ME (1987) Cognitive evoked potentials as Curran T and Hancock J (2007) The FN400 indexes familiarity. USA 95: hippocampal neurons predict how well word pairs will be 3193–3197. Nature 333: 773–775. Neuroreport 9: 3735–3739. and Donchin E (1986) P300 and recall in an important for maximizing our memory capabilities in incidental memory paradigm. Neurophysiol. Friedman D. Psychol.

Psychol. repetition effects of high. 26: 792–800. J. Neurophysiol. 7: Sederberg PB. Donner EJ. Cogn. and Donchin E (1984) ‘‘P300’’ and memory: stimuli. and Düzel E (2002) memory. and Raye CL (1998) The role of prefrontal Rugg MD. Brain Kutas M (2006) Absent event-related potential (ERP) word Res. and 49–55. J. hippocampus during the encoding and recognition of visual Karis D. Clin. Van Petten C. Oxford University Press. Becker C. Ranganath C. MA: MIT Press. distinction between conscious recollection and priming. 6: 107–111. 13: Sederberg PB. Gitelman DR. 12: Ranganath C and Paller KA (1999) Frontal brain potentials 33–38. Cambridge. Clin.96 Neural Substrates of Remembering – Electroencephalographic Studies James W (1890) The Principles of Psychology. Walla P. 20: of mental functions. 117: 1319–1330. and D’Esposito M (2000) Left Psychol. Hippocampus 12: 415–420. Paller KA (2000) Neural measures of conscious and Heinze HJ. Exp. 67: 360–371. Holt and Company. Voss JL. and potentials and positron emission tomography. MA: MIT Press. cortex during tests of episodic memory. Cogn. Brain Res. Schloerscheidt AM. J. Sci. Allan K (1998a) Dissociation of the neural correlates of McCarthy G and Wood CC (1985) Scalp distributions of event. 21: 699–710. Yonelinas A. and Düzel E (2006) Neuroanatomical dissociation unconscious memory. Brain 123: 1948–1963. Senkfor AJ and Van Petten C (1998) Who said what? An Neuron 38: 507–516. long-term memory: Latency predicts recognition Paller KA and McCarthy G (2002) Field potentials in the human performance. Luck SJ (2005) An Introduction to the Event-Related Potential Rugg MD (1990) Event-related brain potentials dissociate Technique. Mem. 87: 252–271. conscious recollection via the electrical activity of the brain. Fabiani M. (2006) Oscillatory perceptual detail. Johnson R Jr. In: Rugg MD (ed. 578–592. 2: and Dolan RJ (1998b) Neural correlates of depth of 399–406. J. Salmon DP. Cereb. Electroencephalogr. and Lindsley DB Neurophysiol. 12: 127–141. Cortex 17: 1190–1196. Cortex 16: Ranganath C and Paller KA (2000) Neural correlates of memory 280–290. Psychol. repetition effects in mild Alzheimer’s disease. Iragui VJ. . 6: 93–102. Cogn. Sanquist TF. Brain Res. Neuroreport 8: 1317–1320. Psychol. Syndulko K. and occurrence. 197–237. Psychophysiology 22: 497–507. manifestations of memory with and without awareness. memory for real objects: An ERP investigation with perform. Quayle AH. et al. Walla P. Two different expressions of declarative memory. Individual differences in the von Restorff effect. MacKenzie G and Donaldson DI (2007) Dissociating recollection Cognit. during recognition are modulated by requirements to retrieve Klimesch W. Nature 392: 595–598. Johnson MK. 62: 203–208. et al. (1980) Electrocortical signs of levels of processing: Olichney JM. Cogn. Kutas M. Parrish TB. Trends Cogn. 18: 367–379. Mem. EEG correlates of episodic trace decay. Clin. (2006) 519–531. Rev. New York: Paller KA. implicit and explicit memory. Neurosci. Clin. Trends Cogn. Rohrbaugh JW. anterior prefrontal activation increases with demands to Kutas M and Dale AM (1997) Electrical and magnetic readings recall specific perceptual information. recognition memory for words. Kutas M. Neuroscience. Psychophysiology 17: 568–576. Neurosci. 9: 489–491. Hippocampal and neocortical gamma oscillations predict Paller KA. Richardson-Klavehn A. Neurol. Neurosci. Brain Res. Perceptual priming versus explicit memory: Dissociable Otten LJ. Riggins BR. and Boehm SA (2007) Validating neural Psychol. (2001) Theta Paller KA and Wagner AD (2002) Observing the transformation synchronization during episodic retrieval: Neural correlates of experience into memory. event-related potential investigation of source and item Paller KA and Kutas M (1992) Brain potentials during memory memory. Kounios J. of conscious awareness. Neurosci. Cogn. processing effects on recollection: Evidence from brain Olichney JM. et al. Grabowecky M. Neurophysiol. 24: retrieval provide neurophysiological support for the 1005–1025. Neurosci. J. Learn. Electroencephalogr. Curr. 72: 395–406. Neuron 22: 605–613. Mandler G (1980) Recognizing: The judgment of previous Rugg MD. Pfefferbaum A. and McIsaac HK (1995) Monitoring monitoring. and Kutas M (2000) Word repetition in amnesia. Hutson CA. Heinze HJ. pp. Klimesch W. retrieval and evaluation. and encoding predict subsequent recall. Mesulam MM. index conscious recollection. and Nolde SF (1997) Electroencephalogr. Cambridge. Reber PJ. and Kopell BS (1985) P300 and Psychol. Bozic VS. Henson RN. Nat. Electrophysiological measures of impaired and spared Schott B.and low-frequency words. Sci. Sci. of encoding processes related to priming and explicit Paller KA (2004) Electrical signals of memory and of the memory. 11(6): 243–250. and Paller KA (2002) Neural correlates of successful monitoring event-related potentials. Neurophysiol. Brain Res. Trends Cogn. Neurosci. 23: Yamada S (1999) Brain waves following remembered faces 10809–10814. Madsen JR (2003) Theta and gamma oscillations during Paller KA. Brain Res. Johnson MK. J. Birch CS. Mark RE. Sci. encoding in an incidental learning paradigm. correlates of familiarity. Exp. Miller BB. and Mayes AR (1987) Neural correlates of H. Schulze-Bonhage A. Iragui VJ. Electrophysiological brain activity and memory source Paller KA. Perceptual analysis and recognition memory. Hanslmayr S. Morris SK. J. Johnson MK. Exp. 16: 177–216. Schloerscheidt AM. Doppelmayr M. Ditewig TA. Senkfor AJ. Sci. 22: 9541–9548. Salmon DP. 9: Knowlton BJ and Squire LR (1995) Remembering and knowing: 209–222. Schott BH. New York: Neuroimage 36: 454–463. 50: encoding identified using functional magnetic resonance 218–233. Cognit. 14: Brain activity before an event predicts later recollection. J. Fletcher PC. from familiarity: Electrophysiological evidence that familiarity Rugg MD and Coles MGH (1995) Electrophysiology of Mind: for faces is associated with a posterior old/new effect. Paller KA. Learn. Cogn. Cereb. Neurosci.) Cognitive RC108. Leynes P (2002) The effect of specific test queries on source.. Ranganath C. Madsen JR. Nolde S. Sauseng P. Richardson-Klavehn A. Frith CD. Van Petten C. and Boehm SG (2003) Neural memory formation in humans. Siwiec RM. Behav. and Rugg MD (2006) neural correlates at encoding. and Kutas M (2002) Episodic action Cogn. imaging. Brain Res. 4: 375–391. Brain Cogn. Paller KA. awareness of remembering. 123: 18–23. Akram S. Dir. Howard MW. related potentials: An ambiguity associated with analysis of Rugg MD and Nagy ME (1989) Event-related potentials and variance models. Mem. Event-Related Brain Potentials and Cognition. Kahana MJ.

Neuropsychology 12: drawings in locations: Spatial source memory and event. et al. encoding yields remembering: Insights from event-related Exp. judgments. source. 18: 1654–1662. J. 323–339. Urbach TP and Kutas M (2002) The intractability of scaling scalp Yonelinas AP. Knowlton BJ. Fabiani M. Neurology 63: 1203–1208. action encoding task. Sci. related potentials. (1998) Recollection and familiarity deficits in amnesia: Psychophysiology 39: 791–808. Neural Substrates of Remembering – Electroencephalographic Studies 97 watch. Wilding EL (2006) On the practice of rescaling scalp-recorded Trott CT. Neurosci. and Fried I (2006) on-the-bus phenomenon: When a face seems familiar but is Differences in mnemonic processing by neurons in the not remembered. and Clark RE (2004) The medial temporal 1307–1324. Mem. Soc. (2004) Recognition of new effect index recollection? Int. Convergence of remember-know. J. Dobbins I. Smith ME (1993) Neurophysiological manifestations of Voss JL and Paller KA (2006) Fluent conceptual processing and recollective experience during recognition memory explicit memory for faces are electrophysiologically distinct. Stark CE. and Rugg MD (2006) neural correlates of recognition memory: An Electrophysiological dissociation of the neural correlates electrophysiological study. J. Senkfor AJ. Neurosci. Psychophysiol. Neurosci. study. and Knight RT distributions to infer neuroelectric sources. J. . Staedtgen M. (1999) Episodic priming and memory for temporal source: Wilding EL and Rugg MD (1996) An event-related potential Event-related potentials reveal age-related differences in study of recognition memory with and without retrieval of prefrontal functioning. Hayama HR. Rev. Brain 119: 889–905. Psychol. Learn. Wilding EL (2000) In what way does the parietal ERP old/ Trautner P. Kroll NE. Cogn. Yovel G and Paller KA (2004) The neural basis of the butcher- Viskontas IV. Friedman D. Lazzara M. Psychophysiology 37: 551–564. neuroimaging. Ritter W. Lond. 354: Squire LR. Cogn. R. Otten LJ. J. and Rugg MD (2001) Context effects on the Woodruff CC. Steinmetz PN. 27: 279–306. Neurosci. Koutstaal W. and Schacter DL (1999) When memory components following temporal lobe lesions. Philos. Aging 14: 390–413. Annu. B Biol. Cogn. Trans. 35: famous faces in the medial temporal lobe: An invasive ERP 81–87. Smith ME and Halgren E (1989) Dissociation of recognition Wagner AD. and Newberg WM (2000) Memory for receiver operating characteristic data. Biol. 15: 50–60. and imagine action encoding tasks versus a non. Neuron 31: 497–505. human hippocampus and parahippocampal regions. process dissociation. 26: 926–933. Cogn. 14: 402–419. Psychol. 5: 1–13. of recollection and familiarity. 72: 325–332. Neurosci. Dietl T. Psychol. and Snodgrass JG electrophysiological data. Tsivilis D. J. Neuroimage 21: 789–800. lobe. Brain Res. and Van Petten C. 1100: 125–135.

3. Indeed.06.06 Structural Basis of Episodic Memory L. the definition of episodic memory has been revised 2000).3. from the present to the ory can in fact be solved by means of the semantic past.3. knowledge) memory (Murray. late-devel- It nevertheless remains controversial whether experi- oping. For example. 1972).06. Some studies. Sweden ª 2008 Elsevier Ltd. but go beyond the memory tasks. and probably unique to tual. Eichenbaum. 1998).06. and early-deteriorating past-oriented memory mental protocols in nonhuman species really tap system.3 The Parietal Lobe 103 3.06. 3. 5) (See Chapter 2.26). It makes possible mental time travel many memory tasks that seemingly tax episodic mem- through subjective time. In dubbed ‘episodic retrieval mode. Its operations require.06.4. but if learning takes place over multiple semantic memory system [See Chapter 2.06.1. Morris and Frey.2 The Frontal Lobe 101 3.06.06.06. 1997. according to a strict definition.1 The Temporal Lobe 101 3.06. Umeå University. learning and memory autonoetic awareness.5 Interactions Among Episodic Memory Structures 106 3.4.7 Summary and Future Directions 108 References 109 3.3 Episodic Memory – Brain Structures 101 3.06. if and only if is subserved by a widely distributed network of retrieval can be defined as involving reexperiencing cortical and subcortical brain regions that overlaps through autonoetic awareness previous experiences. (Tulving.1 Introduction Numerous studies of nonhuman species have aimed at examining aspects of episodic memory.06.06.2 Lateral temporal cortex 101 3.4 Relation Between Episodic Memory and Other Systems 105 3. through memory system. with but also extends beyond the networks subser- then one can claim to have taxed episodic memory ving other memory systems.06.g.. A recent definition states that: way toward meeting core aspects of the definition of episodic memory (e.6 Variableness in the Structural Basis of Episodic Memory 108 3. but rather as knowing.1 Introduction 99 3.06.4 Other Structures 104 3. In that case. in particular.3. and In 1972 Endel Tulving introduced the concept of much has been learned from such studies (Squire and episodic memory (Tulving. Episodic memory is a recently evolved.3 The amygdala 101 3. one’s own previous experi- of paired associates are typically seen as episodic ences.4.1 Modality-specific cortical areas 104 3. or conscious recollection. 1996). Clayton and Dickinson. thus allowing one to reexperience.3. sessions.28].1. 2002. 1995. 3.3.1.4 The cerebellum 105 3. A comprehensive discussion of 99 .3.06. have gone a long several times.06. All rights reserved. humans. Nyberg. retrieval (remembering or conscious recollection) is contin- would not be characterized in terms of remembering gent on the establishment of a specific mental set. Umeå.’ Episodic memory short.06.3. Over the years Knowlton.3.1 The hippocampus and medial temporal cortex 101 3.3.3 The basal ganglia 105 3.3. more vulnerable than other memory systems episodic memory or are rather probing semantic (fac- to neuronal dysfunction. p.06.4. retrieval may well be based on semantic Retrieving information from episodic memory rather than episodic memory.2 Thalamus and the mammillary bodies 104 3.2 Episodic Memory Is Not a Unitary Construct: Component Processes 100 3.

encoded information must The crucial role of autonoetic awareness for epi.. which is believed form of encoding. and the other that an appropriate retrieval cue will be on episodic memory retrieval. but how attention-captur. Tulving (1983. To set the stage for ment with psychological theories that memories are this review. 1984) argued that two necessary con- The current definition of episodic memory puts focus ditions must be met for retrieval to occur. one Unitary Construct: Component subprocess was mentioned: retrieval mode. retrieval is critically dependent on encoding/ been described as a neurocognitve set that serves to learning/acquisition of information (henceforth encod. Retrieval mode has more precisely though. 2004. 2000). such as when one meets a person and by a question about what one did during the past attempts to encode her/his name. needed for initial stabilization of memories and related memory decline appears earlier for episodic include restructuring of existing synaptic connections than for semantic memory (Nyberg et al. such as when one is attending a to result from interactions between cues and stored dinner party. 2005). To be stored over time. has different macroscopic and time scales (Frankland and been related to the slow ontological development of Bontempi. Afterward one will be able to remember information (‘ecphory’). retrieval ing or novel an event is seems to be one deciding of episodic memories is most likely the result of factor (Ranganath and Rainer. Nyberg. that been assumed that. reactivation of forms of neuronal dysfunction (Tulving. val likely accounts for why episodic memory is more More recent research has challenged this view by vulnerable than other memory systems to various showing that memory retrieval (i. when possible. an additional element of much of what happened at the party.. and for both suc- memorize. some of which show age-associated regions that support memory. it has changes (Hedden and Gabrieli. the focus mode. Traditionally.e. In his Processes theory on the elements of episodic memory. 2003) may and growth of new ones. and in the subse- How novel events are experienced is determined by quent discussion of brain regions underlying previous experiences. once consolidated.. must be present. several interacting subprocesses. with a special emphasis on condition is that the system must be in the retrieval ‘retrieval mode. The main component process of episodic memory is retrieval. 2002). as the other processes. this mode might be instantiated by intention.06. These findings are in good agree- be engaged by episodic memory. 2000. In an Encoding is a complex construct. con- have to do with its putative reliance on a multitude of solidation refers to a gradual reorganization of brain brain regions. hence encoding and retrieval episodic retrieval. In short. remain stable and be relatively immune to insults. retrieval is induced: remembering or conscious recol- and so on. 2006). of monitoring processes is needed. This is so despite the fact that most people lection. It can also be an vacation. The consolidated memories) can induce changes at the main purpose of this chapter is to provide an empiri.. It is not clear what determines what will cessful and unsuccessful retrieval attempts some form be encoded into memory. is best seen as an umbrella term for a set of subprocesses. a brief discussion of component processes dynamic and subject to change (Schacter et al. 1998). Fast morphological changes are episodic memory (Wheeler et al. hold a segment of one’s personal past in focal atten- ing) and the subsequent consolidation of such tion and treats incoming information as retrieval cues information. strategic do not sit at dinner parties intentionally trying to search processes may be elicited. This takes place at sodic memory. A neurobiological processes will be considered. 2005).2 Episodic Memory Is Not a the definition of episodic memory given above. If information is not retrieved. 2005). goes beyond the scope of this chapter. Functional brain . At the systems level. 2004). Similarly. who said what. 1997).100 Structural Basis of Episodic Memory issues related to the phylogenetic history of episodic model for such integration has been proposed memory and whether or not it is unique to humans (Fernandez and Tendolkar.’ Similarly. In 3. undergo a consolidation process. for events in the past (Lepage et al. a memory will many brain regions seem to be engaged during retrie. of episodic memory will follow next. It can be guided everyday situation. Frankland and cal overview of brain regions that have been found to Bontempi. relative to other memory systems. Retrieval cues may then be self-generated incidental process and occur as a by-product of other or induced by additional questions. That age. specific retrieval operations are closely integrated. The latter type may be the more common actually is retrieved from memory.. in this chapter. neural level (Nadel and Land. One on retrieval of information. Retrieval. If information activities. Needless to say.

Of these. Frontal patients tend more related to semantic memory processes (Murray to be impaired on episodic recognition as well as and Bussey. numerous papers have cessful retrieval of 1-year-old emotional memories tried to determine the functional role of the hippo. As noted earlier.. functional brain imaging poral regions of the ventral visual pathway.06. Interactions Frontal lobe contributions to episodic memory of lateral temporal and medial temporal regions may have been referred to as working-with-memory . It should be stressed that most if not all of the regions that will be discussed are not 3. The amygdala discussed later on in the chapter. (Dolcos et al.. whereas perirhinal cortex seems damage (Wheeler et al. as this underlie episodic memory (Eichenbaum.. 1995) from such studies support a role for frontal regions in and formed associations between visual stimuli both encoding and retrieval of episodic information (Fujimichi et al. Stuss and ‘proper’ by being related to remembering rather than Benson. amnesia.2 The Frontal Lobe role of specific hippocampal subregions as well as nearby areas in the medial-temporal cortex (Squire Lesions to the frontal lobes do not generally lead to and Knowlton. cesses. it is possible to first examine how the regions (Persson et al.3. Of main concern here is that episodic knowing and to recollection of episodes (Brown and memory impairment has been linked to frontal lobe Aggleton.1. 2005).. during retrieval).1. By now. The The amygdala is a key brain region for various forms of issue of relations among memory systems will be emotion. connections with many cortical and subcortical areas. 2004. Following the seminal paper by emotional stimuli (Sharot et al. but frontal regions have direct or indirect some authors refer to multiple temporal lobe systems.3 Episodic Memory – Brain Through such connections the amygdala seems able to Structures emotionally influence various forms of cognition. 2006). dala serves to emotionally flavor recollected episodic tive) memory (See Chapter 3. LaBar and 3. That a deficit is seen on recognition tasks 3. notably fear (LeDoux..1. 1995).06. where it borders ventrolateral tem. but amygdala implicated in episodic memory by discussing the activation has specifically been related to recollecting hippocampus.06. is not part of the core medial temporal memory system but has extensive connections with medial temporal as well as several other cortical and subcortical regions. the hippocam. learning and memory has been recognized for quite pal system is more closely linked to episodic memory some time (See Chapter 3.. and the results by stimulus familiarity (Desimone et al.3. nondeclarative systems intact (Cohen and Squire. 2001). Thus.3 The amygdala uniquely activated during episodic retrieval.06. recall tasks. with technique allows component processes to be studied additional top-down contributions from frontal separately (e.1 The Temporal Lobe Cabeza. see also the section titled brain is activated during encoding and subsequently ‘The frontal lobe’).3. critical for aspects of semantic memory. 2001)..3. The allows separate study of brain systems activated dur- activity in ventral temporal regions is modulated ing the encoding and retrieval stages. Luria. with a more pronounced impairment for recall. 2000). Nyberg et al. 1984). 1999).06. 1980). 2002.06. Eichenbaum. the amyg- hippocampus impair episodic and semantic (declara. 1995. 2004) and to the suc- Scoville and Milner (1957). 2004) and may thus be seen as (Tulving et al. 2004).2 Lateral temporal cortex suggests that the frontal lobes contribute to the Regions of lateral temporal cortex also subserve encoding of information. 1996a). Structural Basis of Episodic Memory 101 imaging studies are critical in this regard. Such enhancement may in part reflect It is natural to begin the review of brain structures encoding and consolidation processes. It has been shown that lesions to the brain regions and neuromodulatory systems. such as the perirhinal versus hippocampal systems and a role for frontal cortex in various forms of (Brown and Aggleton. 3.04) but leave other information (Markowitsch. in interaction with other campus.. while the stronger reduction mnemonic functions.1 The hippocampus and medial is the enhanced memory for emotionally arousing temporal cortex information. In episodic memory one such influence 3.g. 1999).3. including episodic memory (Phelps. 1973. Perirhinal cortex extends from for frontal patients on recall tests indicates that frontal medial to lateral temporal cortex (Murray and regions make a distinct contribution to retrieval pro- Bussey. 1994). 1995).. 1994. Further studies have examined the functional 3.14.

including of retrieval success (Fletcher and Henson. activity in a region of left frontal cortex (right panel) was modulated by retrieval success. and high target density (bars 2–4) relative to verbal encoding (bar 1). p = . mediated by frontal regions. may con. 1998). 2001). y. Rugg. 0) Left frontal area 10/46 (x. notably in the right hemisphere (Lepage et al. (2000) Large scale neurocognitive networks underlying episodic memory.. Cogn. . cannot remember the name of a person.96.102 Structural Basis of Episodic Memory (Moscovitch. A similar pattern is seen for picture retrieval (bars 6–8) relative to picture encoding (bar 5). corresponding frontal activations depend on the level tribute to successful memory encoding. the degree of activity in frontal regions dur. and selection/organization 2004). additional At retrieval. 46.000 (. it may be necessary to verify that in the definition of episodic memory and in the the retrieved information is correct. y. 50. Retrieval mode. The bar graph shows that right frontal cortex is associated with increased activity during verbal retrieval conditions of low. 20) Z = 4. 2001).000 (. Figure 1).. generation. p = .74. frontal activity can reflect a number of search processes may be initiated. 2000).019 corrected) Z = 3. By contrast. failure and success. J. as was mentioned attempt succeeds.. A defining feature of retrieval-mode type of Indeed. And for both these earlier discussion of component processes of episodic outcomes. If the retrieval processes as well. or processes that control and memory. et al. than storage per se (Fletcher and Henson. Several et al. Habib R. Persson J. Neurosci. with permission. 2000. maintenance. 1992). z = –42. whereas other processes and processes. has been linked to several distinct frontal lobe optimize memory encoding and retrieval rather regions. intermediate.327 corrected) 76 60 74 58 Adjusted response Adjusted response 72 56 70 54 68 52 66 50 64 48 62 0 2 4 6 8 0 2 4 6 8 Level Level Figure 1 The left panel shows increased activity in right frontal cortex during episodic memory. such as when one of information (Fletcher and Henson. z = 26. If a retrieval attempt fails. activations is that they should be present in conditions ing incidental encoding predicts subsequent episodic of high as well as low levels of recollection (Lepage memory performance (Wagner et al. as illustrated by the bar graph. there is a need for Right frontopolar cortex (x. 12: 163–173. Adapted from Nyberg L. 2001. This pattern relates right prefrontal cortex to retrieval mode.

and recognition. Fletcher and Henson. Higher levels of cognitive control were associated with increased activity in left frontal cortex (right panel). 3. tations. parietal regions may indeed subserve mnemonic pro- 2001. novelty detection. episodic memory to various forms of attention. and activity when old compared to when new stimuli if explicit is defined in terms of intentional retrieval are presented. Based on similarities in activation patterns The parietal cortex is typically discussed in relation for attention.. the degree of activity seems to be memory are closely related to the concept of cognitive modulated by recollective experience.. include attention to internal representations and the tive control to increased frontal activity (see also active representation of retrieved information. Buckner and Wheeler. By comparing recognition of old versus new pictures (condition 4 vs. information (old/new effects. with permission. success (Cabeza and Nyberg. frontal contributions to episodic Furthermore. ory. 2005). and individuals correctly recognize processes (Schacter et al.3 The Parietal Lobe 2005). it was found that parietal regions are was suggested that parietofrontal activity may reflect often activated both in studies of selective attention integration among multimodal distributed represen- and episodic memory retrieval (Cabeza and Nyberg. 1995). retrieval. Candidate processes stimulus familiarity alone (Persson et al. 2004). This observation suggests that.3. For example. However. 2000). it imaging studies.. 6).. 2005). 2000. 2001). Habib R. . More generally. targets) Figure 2 The left panel outlines the design of the experiment. Cabeza and Nyberg. 1989). and Nyberg L (2002) Decreased activity in inferotemporal cortex during explicit memory: Dissociating priming. thereby relating increased levels of cogni. cesses.. 2004). whether this activation pattern generalizes beyond parietal influences to episodic memory reflect the visual stimulus modality. indicating that 1998. Wagner et al.06. In a review of brain. distracters) 7 Yes/no recognition (items from 5. and conscious perception (Figure 3). control (Wagner et al. that lateral and medial parietal regions show higher cit form of memory (Squire and Knowlton. frontal There are likely anatomical connections between activity is greater when participants have to identify regions of the parietal lobe and the medial-temporal target items among equally familiar distracter lobe system that underlie various mnemonic pro- items than when they can base their response on cesses (Wagner et al. 3) and recognition of old targets with equally familiar distracter items (condition 7 vs. at least in part. activity in parietal regions of the frontal cortex seem to be involved in such regions is frequently modulated by level of retrieval component processes of episodic retrieval (Nolde et al. distracters) 4 Yes/no recognition (items from 2. Adapted from Persson J. the recruitment of the old information and correctly reject the new frontal regions critically contributes to explicit mem. working memory. the demands on cognitive control could be varied (higher in the second than in the first comparison). 2002. Condition 1 Low-level baseline 2 Picture encoding 3 Yes/no recognition (non-studied items. Further research is needed to determine 2000).. targets) 5 Picture encoding 2 6 Yes/no recognition (items from 2 & 4. related account of parietal (and frontal) activations was provided in a study of cross-domain similarities in brain activation patterns (Naghavi and Nyberg. NeuroReport 13: 1–5. Structural Basis of Episodic Memory 103 monitoring the fate of the retrieval attempt. that could account for the old/new parietal effect Figure 2). Specific attentional processes. This notion is further supported by findings Episodic memory is typically classified as an expli. A Fujimichi et al.

perception are subsequently reactivated during epi.104 Structural Basis of Episodic Memory R Episodic Attention memory retrieval Working memory Consciousness Figure 3 Overlapping patterns of frontoparietal activity in brain-imaging studies of attention.e.3.. 1982). cephalon with medial temporal lobe regions (a The hippocampus has been assigned an important disconnection syndrome. and consciousness: Shared demands on integration? Conscious.06. occipital cortex for visual information. 14: 390–425. Nadel and Moscovitch.3.06. with permission of Nyberg. and motor cortex of the frontal lobe for information SPATIAL about activities (Buckner and Wheeler. can cause diencephalic the involvement of modality-specific regions during amnesia (Butters and Stuss. spective by arguing that the mammillary bodies McClelland et al. 2000.4. Inc.1 Modality-specific cortical areas In the discussion of regions of lateral temporal cortex it was mentioned that specific areas have been found to be engaged during episodic retrieval of visual information. the mammillary bodies.4 Other Structures 3. a subsidiary of John Wiley & Sons. 2001. 2002). and visual consciousness. presumably due to close anatomical notably the dorsomedial nucleus of the thalamus and connections (Squire and Knowlton. with permission. Nyberg. 1994. memory.06. 1995). Wiley-Liss. see also Warrington and role in integrating the distributed regions that Weiskrantz. Cogn. It has been stressed that from the perspective of 3. Alvarez and Squire. an remarked that diencephalic amnesia may also result interaction between retrieval cues and stored infor. Adapted from Naghavi HR and Nyberg L (2005) Common fronto-parietal activity in attention. Butters and Stuss episodic memory may relate to ‘ecphory’ (i. 1989).. such as parietal cortex for spatial information. episodic retrieval. 1986. More recently. Figure 4). For example.3. from damage to fiber tracts that connect the dien- mation that results in actual retrieval of information).4. contribute to mnemonic processing through their 1997). Res. 3. lobe amnesia. Tech. As suggested by theoretical models (Damasio. 2002). 1989). The sites of retrieval-related activity overlap with how the brain was activated during initial perception/encoding. projections to temporal regions.. This seems to be a general pattern that extends to other cortical regions and kinds of episo- dic information. Microsc. Vann and jointly define a stored episodic memory (Teyler Aggleton (2004) provided support for the latter per- and DiScenna. regions of Figure 4 Areas of parietal cortex that were activated the parietal cortex that are engaged during spatial during both encoding and retrieval of spatial information. For . sites where overlapping activity is observed during encoding and retrieval may identify distribu- ted sites of memory storage (Nyberg. 1995.2 Thalamus and the mammillary behavioral criteria there is striking similarity bodies between diencephalic amnesia and medial temporal Damage to the medial portions of the diencephalon. Inc. working memory. Thus. Adapted from Persson J and Nyberg L (2000) Conjunction analysis of cortical activations common to encoding and sodic retrieval of spatial information (Persson and retrieval. 51: 39–44.

1999. conscious awareness (Squire et al. 2002). Memory and Other Systems ing to the same functional system as the medial temporal lobe complex (Squire and Knowlton.4 The cerebellum by a network of brain regions that overlaps with the A final region to be mentioned is the cerebellum.06. the diencephalon may be seen as belong. The nature of episodic deficits is memory were observed (Figure 6. These observations of overlap connections of the basal ganglia with most regions of are in line with several other recent findings the frontal lobe. 2006). via the limbic and the associative (Wagner. This relates to the suggestion that the observed during conscious episodic memory. 2002). both groups display similar forgetting rates. The hippocampus. -Parahippocampal 1999).4 Relation Between Episodic Therefore.’ as hippocampal damage leads to part. most research activities have used the dissociation logic to define what is unique for various 3.3. recently. 2006).4.3. show various multivariate analysis of brain-imaging data associated forms of motor deficits. Cabeza and Nyberg. cerebrum in a corticocerebellar network that initiates 2002).. 2002). overlap in activity has begun to attract Patients with basal ganglia disorders. Commonly recruited regions for both episodic damage.. Collectively. 1996). cf. cerebellum has been argued to contribute to cogni. the addition. In a Huntington’s or Parkinson’s disease. Critically. the core defining feature of episodic memory (Tulving. stating that episodic memory is subserved 3.. 2003) and also network (Yin and Knowlton. in keeping with the current definition of episodic memory. 1995). Squire et al. (Tulving. A controversial issue is whether some brain tive processing (Schmahmann. Nyberg et al.4. as well as nonmemory control as well. This overlap in episodic and working memory. 1996). 1995). 2003). ory (Markowitsch. Nyberg et al. in tleneck structure. Instead. and the core cognitive component processes (Nyberg. Increased regions are specifically recruited by episodic mem- activity in the cerebellum has specifically been ory tasks. cortex and right cerebellum (See Chapter 3. 2004).. Furthermore. stem from altered cerebellar activity (Achim massive and enduring impairment of episodic mem- and Lepage. with mnemonic processes. The critical role of the cate that the cerebellum should be considered a part hippocampus is closely related to its importance for of the brain network that subserves episodic memory. 1995. More The basal ganglia are part of the motor system. including right fron- results patterns after lesions to the basal ganglia and tal and medial and lateral parietal cortex (Figure 6.3 The basal ganglia memory systems (Nyberg and Tulving. including episodic memory impairment tasks. 1999). indicating that hippocampal activation in and Figure 5 Schematic view of connections of the of itself is not a definite signature of episodic con- cerebellum with cortical areas that have been associated scious memory (Eichenbaum. networks that subserve other memory systems This region is strongly associated with the coordina. reminiscent of that seen in patients with frontal lobe 2002). as was noted earlier.. Traditionally. Assessment of regional overlap tion of movement. In ported recognition tests (Wheeler et al.06. what may underlie the type of . patients with with measures of episodic memory. similarities between episodic and semantic (Knowlton. 1999. However.06. However. working memory. episodic memory and has been referred to as a ‘bot- deficient episodic memory in schizophrenia may. Structural Basis of Episodic Memory 105 example.. holds and monitors the conscious retrieval of episodic a special role in theories on the structural basis of memories (Andreasen et al. the cerebellum is not across a range of tasks can help to constrain the only connected with motor cortical areas but also functional contribution of brain regions and define with asserted cognitive regions (Figure 5).07). the frontal lobe can be attributable to strong inter. the performance on tasks that are Cerebellum Thalamus Cortex -Parietal not accompanied by awareness has been found to be -Frontal impaired by hippocampal lesions (Chun and Phelps. similarities were noted among measures of section titled ‘The frontal lobe’). such as relatively greater impairment on and semantic memory were located in left frontal more demanding tests of recall than on more sup. semantic mem- basal ganglia deficits can exhibit cognitive deficits ory. such as increased interest (Cabeza and Nyberg. 3. and it network subserving episodic memory extends was proposed that the cerebellum interacts with the beyond those subserving other systems (Tulving. 2004). these findings indi. 2002). 2005).

and the top right panel shows brain regions that were jointly activated by episodic and semantic test. increased firing area 10 in episodic memory (Ramnani and Owen. re. 13: 281–292. lowing section. Brain Res. certain frontal regions have been sug. rl. corresponding to Brodmann’s area 10. but also by other tasks such as complex tests of working memory (MacLeod et al. Forkstam C. of the neuronal ensembles that define the network 2004). VAC. including episodic retrieval.e. random-number generation. as for the hippocampus. recognition) and working (2b. 1998. fact retrieval) memory. through interactions with is interactions among the hippocampus and other other regions. living/non-living categorization. synonym generation. 2-back test) memory. the al. conscious awareness that signifies episodic memory the frontopolar cortex. read low demands (same word repeated). ft. much discussion has concerned the role of anterior frontal cortex. Cogn. 1-back test. Cabeza R. but it still remains a viable hypothesis that (Fuster. read. 2003). Right area 10 is 3. critically contributes an episodic brain regions (Eichenbaum. Memory Structures val. In the review of episodic memory . in par- ticular in the right hemisphere. Petersson KM.. In particular. 1997). 1b. cr. sodic memory (Tulving.. autobiographical memory. The crucial importance of considering gested to have a special status in episodic memory by network interactions is discussed further in the fol- underlying the mental time travel component of epi. posterior and anterior commissures. Adapted from Nyberg L. 1999). 2002).5 Interactions Among Episodic frequently activated during episodic memory retrie.106 Structural Basis of Episodic Memory Sagittal Coronal 30 32 34 36 38 40 42 72 Brain scores 0 0 32 VPC VAC 0 64 –104 68 VPC VAC 0 64 Transverse au sy ft cr 1b rl re Condition Sagittal Coronal 72 31 28 29 30 0 0 Brain scores 32 VPC VAC 0 64 –104 68 VPC VAC 27 0 26 64 cr rn 2b #gn nl ft re Transverse Condition Figure 6 The top left panel shows how a multivariate analysis grouped tests of episodic (au. with permission.. Thus. and Ingvar M (2002) Brain imaging of human memory systems: Between-systems similarities and within-system differences. 2000). i. cued recall) and semantic (sy. In addition to the component to past and ongoing behavior (Koechlin hippocampus. et al. VPC. gn.06. The lower panels show corresponding data for a grouping of tests of episodic (rn. nl. can be viewed as the evidence does not support a selective role of right activation of a network memory. Cabeza and It has been argued that all aspects of memory retriev- Nyberg.

in which various regions have been found to be coactivated in brain. 2006) should to regional interactions is methodological – it is dif. McIntosh.. experimental conditions (Simons and Spiers. 1995. Structural Basis of Episodic Memory 107 brain structures. 2000.09 –20 –. 2003). Krause et al. 16: 3753–3759. Büchel et al.35 –25 22 37 CB –. Neurosci.51 –34 45 . 1994.. studies on episodic memory in which some form of poral regions are functionally separated following connectivity analysis was used is quite low. 1998. 2003). should be However. However. Most likely. Cabeza R.60 24 –.. medial temporal interactions. The findings of these studies substantiate • connections within temporal cortex (1) among different medial temporal regions. . After having identified a set of regions and their connections (anatomical model. 1982). One source of evidence is animal studies that Such examination should include variability in the convincingly have shown that disruption of critical degree to which a network is activated – both with Recognition 10 8 –. lesions in thalamus and the mammillary bodies important contributions have been made (Köhler et al. McClelland et al. Tomita et al.52 –.50 23 –. This is so despite the fact data can be used to quantify interactions and compare that a network perspective has been emphasized for the strength of interregional connections between many years by several authors. Simons and is still known about actual interregional interactions Spiers. 2002). (1996b) Network analysis of positron emission tomography regional cerebral blood flow data: Ensemble inhibition during episodic memory retrieval.33 –. left)..09 –. This includes information about inter- • connections between medial-temporal and (1) parietal cortex.. in particular. interactions. connectivity approaches to data analysis are used imaging studies of episodic memory. relatively little (Büchel and Friston. To date. 1997. connections among several regions pathways has a profound impact on task performance were highlighted. Squire. Computational models have also hippocampus and amygdala. Adapted from Nyberg L. and (3) cerebellum. pave the way for forthcoming contributions. episodic memory retrieval (Simons and Spiers. 2003).59 . 2003). the number of published imaging disconnection syndrome in which frontal and tem... 2003). Prefrontal- ficult to measure and quantify such interactions. including: (Hasegawa et al. interactions among hippo- and (3) diencephalon. ganglia and the frontal cortex (Atallah et al. ple is the suggestion that amnesia might be seen as a Figure 7).. actions among medial-temporal subregions (Norman and O’Reilly. One prominent exam. and (3) among med- provided important information about regional ial and lateral temporal regions. 1999. (Warrington and Weiskrantz. with permission. Although anatomical pathways have been identified One further source of evidence comes from func- for these connections. 2000).55 18 Figure 7 Illustration of connectivity approach to the analysis of functional brain imaging data. (2) modality-specific cortices. Norman and O’Reilly. Gaffan et al. and interactions between the basal (2) basal ganglia. the 1998. With these analytical tools. 2003. brain-imaging during task performance. 1999. J.. (2) between the importance of frontotemporal interactions during memory retrieval. and con- reason for the poor state of knowledge when it comes tinued methods development (Lee et al. structural equation modeling can be used to estimate functional influences among regions of the network (right). several lines of research substantiate the critical to examine further for component processes of critical role of interactions (Simons and Spiers. McIntosh AR. et al. 2004). and although many of these tional brain-imaging studies in humans. campus and modality-specific regions (Alvarez and • connections between frontal regions and cortical and subcortical areas such as (1) temporal cortex.56 20 20 –.

which has not been extensively studied. and basis is quite variable or dynamic. induce variability in the brain structures that are engaged during episodic retrieval. whereas network loop. The idio- may give the impression that a static.or practice-related effects will serve to activity over repeated ignitions (Fuster. such that the degree to which differ. All of these ments at a given time and to variability in network kinds of time. as indicated by the dashed line other systems and brain regions come into play at later in Figure 8 from stage 6 to stage 1.. or may more strongly many sources of variability must be taken into activate the hippocampus and recollective processes account. there is much evidence that the structural the associated memory network (Fuster. Similarly. . such by conscious recollection. 1995. including monitoring and verification are recruited during retrieval in a time-dependent processes mediated by the frontal lobe. such as frontal cortex.. especially prone to show individual differences in However. and as a function of practice (Chein and Schneider. if necessary.g. less supported tests (e. the episodic frontal cortex along with the basal ganglia and memory system and related brain regions may be the cerebellum..06.. This may initiate another memory recruited during the initial stages of learning. 1997). The parietal cortex may represent retrieved on supervisory control functions (Chein and information such that it is accessible for conscious Schneider. examine the brain bases of various forms of variabil- free recall) generally tax some brain regions to a ity (MacDonald et al. Signals from the frontal cortex. such that hippocampal as ‘What did you do last Saturday evening?’ This activity is specifically associated with recollective question would serve as a cue eliciting episodic experience (remembering vs. 2004). 2000).. the hippocampus would underlie recollective experi- and such reductions may reflect attenuated demand ence.. 2002). Markowitsch. Also. 2004).g. and memory (Nyberg et al. In summary.. greater extent than more supported tests (e. reflecting active Another source of variation relates to practice or retrieval.6 Variableness in the Structural Such variability may reflect genetically induced dif- Basis of Episodic Memory ferences in personality and cognitive processes as well as between-person differences in real-life The notion of a structural basis of episodic memory experiences (Fossella and Posner. hippocampal activation has been found to be contin. terms of its neural basis to semantic than episodic We are still in the early stages of exploration. search processes. fixed constella. the amygdala. Alternatively. One source of particularly so for network components that reflect variation comes from the type of retrieval test that memory storage. Feedback projections to reductions are located in frontal and parietal cortices. Yet another source of variation. is individual variability. there is some evidence that partly different brain regions are recruited during retrieval 3. the main goal of the present chapter was autobiographical memory may be more similar in to discuss the structural basis of episodic memory. There is converging evidence from several lobe via subcortical pathways and trigger neuronal studies that network activity is strongly influenced by ensembles in the hippocampus that store sparse practice on a task. 1997). 3. Regions that consistently show practice-related to reactivate memory traces. 2005). basal ganglia. but a network of episodic-retrieval struc- than laboratory tests (Cabeza et al. Depending on the testing conditions. Additional manner depending on the stage of memory consolida. tures can tentatively be outlined (Figure 8. 2006). 2004). knowing. Damasio. These ensembles ent network elements are activated tends to decrease would in turn activate specific neocortical sites. in the case of emotional information. and cerebellum. stages of learning (Poldrack et al. 2005).06. Eldridge retrieval mode and corresponding frontal activity. 2001). recog- nition). The activa- gent on whether or not retrieval is accompanied tion of this network could start with a question. Fuster 1997). representations of past episodes. A related effect is that brain regions processing..108 Structural Basis of Episodic Memory regard to the activation degree of its constituent ele. Future studies are called for that is used. As has been noted. syncratic nature of episodic memory may make it tion of brain regions is recruited during retrieval. 1989. would then reach regions of the temporal experience. Furthermore. are subserved by the tion (Maviel et al. et al.7 Summary and Future of information acquired in the laboratory compared Directions to retrieval of real-life episodes (autobiographical memory).

Figure 8 by the dashed line from stage 1 toward Atallah HE. are the roles of perirhinal cortex and hippocampus? Nat. USA 91: 7041–7045. Mohlberg H.. Cabeza R. Rev. Neurosci. Gilsbach JM. Arrows denote connections and information flow. refine this tentative episodic-retrieval network. Neurobiol. pp. (1999) The cerebellum plays a role in conscious episodic memory form of action might be taken (represented in retrieval. 1092–1105. 12: 1–47. J. 42: 287–296. 2006). Daselaar SM. and Mazziotta JC (eds. In: Zani A and Proverbio AM (eds. Brain Mapp. 2004) can yield unique information Zilles K (2006) White matter fiber tracts of the human brain: Three-dimensional mapping at microscopic resolution. and (Donaldson.) Human Brain Function. Prince SE. Frank MJ. Neurosci. Obviously. . Hum. Amunts K. Cogn. future studies are needed to revise and Learn. In: Frackowiak RSJ. Nat. Büchel C and Friston KJ (1997) Characterising functional integration. Vol. memory networks is modulated by various neuro. and (4) concepts – conceptual analysis is a Butters N and Stuss DT (1989) Diencephalic amnesia. (3) temporal information – electromagnetic pp. Frith CD. Cue -> retrieval mode x t ty-s Frontal cortex e li a 4. 3. Natl. J. Cabeza R and Nyberg L (2000) Imaging cognition II: An empirical review of 275 PET and fMRI studies. New York: activation in schizophrenia: Meta-analysis. 2004) and advanced of remembering. Amsterdam: Elsevier. following: (1) anatomical pathways – an interesting Brown MW and Aggleton JP (2001) Recognition memory: What possibility is that in vivo studies of white matter con. Recollective experience Cu da al 3. the motor cortex). London: Academic Press. Dolan 2001. Schott et al. crucial determinant of progress in defining the struc. transmitters such as dopamine (Bäckman and Farde. Büchel C.. 8: 226–234. cortex. Meyer-Lindenberg et al. 16: 1583–1594. (2004) Brain activity Alvarez P and Squire LR (1994) Memory consolidation and the during episodic retrieval of autobiographical and laboratory medial temporal lobe: A simple network model. Sci. pp. about the timing and temporal profile of retrieval topography and intersubject variability. tural basis of episodic memory (Tulving. Both regions and component processes of retrieval are indicated. Friston KJ. et al. events: An fMRI study using a novel photo paradigm.. 2000). 82: 253–267. some Andreasen NC. RJ. Mem. J. Psychol. 2: 51–61. Scand. Acad. 187: 500–509. In: Boller F and Grafman J (eds. and O’Reilly RC (2004) Hippocampus. 41–68. et al. 127–140. Neurosci. and Friston KJ (1999) The predictive value of et al. J. Post-retrieval processing (monitoring. actions) y) or h Frontal cortex cp Basal ganglia (e o n Cerebellum cti x ra te i n te cor t fic ge ci ar pe 1. Structural Basis of Episodic Memory 109 5. Active representation Parietal cortex 6. 2005. Buckner RL and Wheeler ME (2001) The cognitive neuroscience brain imaging techniques (Rugg. Memory activation Temporal cortex Diencephalon Figure 8 Schematic diagram of the episodic retrieval network. Psychiatry Elsevier Science. Cogn. NeuroImage 29: processes such as retrieval mode (Marklund et al. Neurosci. Hoemke L. Bäckman L and Farde L (2001) Dopamine and cognitive Important issues to take into account include the functioning: Brain imaging findings in Huntington’s disease and normal aging. 107–148. 2007). (2) neuromodulation – the functioning of changes in effective connectivity for human learning. based on the output from the retrieval network. Br. functional magnetic resonance imaging paradigms Bürgel U. Coull JT. Mo ygd Hippocampus Am 2. 2: 1–11.) The Cognitive Achim AM and Lepage M (2005) Episodic memory-related Electrophysiology of Mind and Brain. and basal ganglia: Insights from computational models of complementary learning systems.. Cabeza R and Nyberg L (2003) Seeing the forest through the References trees: The cross-function approach to imaging cognition.) Handbook of Neuropsychology. Proc. Paradiso S. nections can be used to study brain networks (Bürgel Rev. O’Leary DS. Science 283: 1538–1541. 2006).

905–918. Nyberg L. Midbrain dopamine and prefrontal function in humans: Fujimichi R. Knowlton BJ. and Miyashita Y (1998) Callosal Murray EA (1996) What have ablation studies told us about the window between prefrontal cortices: Cognitive interaction to neural substrates of stimulus memory? Semin. Nat. Engel SA (2000) Remembering episodes: A selective role for Markowitsch HJ (1995) Which brain regions are critically the hippocampus during retrieval. LaBar KS and Cabeza R (2006) Cognitive neuroscience of Damasio AR (1989) Time-locked multiregional retroactivation: A emotional memory. 3rd edn. Neuroimage 7: Eichenbaum H (2004) An information processing framework for 41–48. and Bäckman L (2006) Intra- designs: What kind of a state is neuroimaging in? Trends individual variability in behavior: Links to brain structure. Rev. 22: 7288–7296. memory: Representation. Brain Res. R. Rev. Kolachana B.) The Cognitive Neurosciences III. Petersson KM. and attention. and MS (ed. 1: 209–212. 6: 119–130. Psychol. and Parker A (2002) Interaction of inferior 1489–1503. systems-level proposal for the neural substrates of recall and LeDoux JE (1994) Emotion. Ingvar M. Trends there are complementary learning systems in the Cogn. Trends Neurosci. Miller EK. Cogn. Neuroimage 30: MS (ed. Köhler S. In: Squire LR and Schacter DL (eds. Neuropsychology. 7: 217–227. Nyberg L. Cereb. Mottaghy FM. In: Gazzaniga models and dynamic causal modeling. MA: The MIT Press. McNaughton BL. 13: 861–870. Furmanski CS. semantic monitoring and working memory. Neural Netw. Eldridge LL. 50–57. hippocampus. Knowlton BJ (2002) The role of the basal ganglia in learning and Neurosci. Fernández G and Tendolkar I (2006) The rhinal cortex: McClelland JL. Science 281: 814–818. temporal cortex with frontal cortex and basal forebrain: Moscovitch M (1992) Memory and working-with-memory: A Double dissociation in strategy implementation and component process model based on modules and central associative learning. J. Ody C. Rev. and during cache recovery by scrub jays. Müller-Gärtner H-W (2000) Imaging and neural modelling in Cohen NJ and Squire LR (1980) Preserved learning and episodic and working memory processes. Clayton NS and Dickinson AD (1998) Episodic-like memory Krause JB. (2005) and remote memories. MacLeod AK. Science a domain-general control network for learning. 5: of the perirhinal cortex. 1243–1254. 13: retention of pattern-analyzing skill in amnesia: Dissociation 847–859. Acad. pp. and O’Reilly RC (1995) Why ‘gatekeeper’ of the declarative memory system. Curr. Cognition 33: 25–62. 3: 142–151. Nat. Luria AR (1973) The Working Brain: An Introduction to Natl.. MA: The MIT Press. J. Lepage M. Naya Y. Trends Cogn. Bookheimer SY. Science 305: 96–99. Nat. attended experience? Phil. Neurosci. pp. Opin. and Lueschow A (1995) Lee L.) Nadel L and Moscovitch M (1997) Memory consolidation. Science 210: 207–210. and Winocur G (1998) Chun MM and Phelps EA (1999) Memory deficits for implicit Functional interactions between the medial temporal lobes contextual information in amnesic subjects with and posterior neocortex related to episodic memory hippocampal damage. Nat. 102: 419–457. pp. 10: 358–362. Morris RGM and Frey U (1997) Hippocampal synaptic plasticity: pp. York: Guilford Press.110 Structural Basis of Episodic Memory Chein JM and Schenider W (2005) Neuroimaging studies of Koechlin E. Sci. Kohn PD. Neurobiol. Neurosci. Neurosci. Sci. Rev. memory representation by the hippocampus. Neurosci. 1: 41–50. Schmidt D. 2: 844–847. 1149–1152. . Sci. New York: Basic Books. Ihara T. Am. Durkin TP. Soc. Neurosci. Neuropsychology of Memory. Cambridge. Cortex 8: 451–461. Buckner RL. 1255–1266. Nat. Easton A. Nat. Proc. 2: 775–776. Neural Netw. Brain Res. Neurosci. Fukushima T. of knowing how and knowing that. Nature 395: 272–278. 302: 1181–1185. Trans. Rev. pp. Hautzel H. neurotransmission and neuronal activity. lobe memory system in retrieving emotional memories. hippocampus and neocortex: Insights from the successes Fossella J and Posner MI (2004) Genes and the development of and failures of connectionist models of learning and memory. Nadel L and Land C (2000) Memory traces revisited. Nat. 25: 607–623. and Horwitz B (2006) Large-scale neural Multiple memory systems in the visual cortex. Lond. In: Gazzaniga MS (ed. McIntosh AR (2000) Towards a network theory of cognition. Fletcher PC and Henson RNA (2001) Frontal lobes and human Maviel T. Neurosci. 87–97. Friston K. and cognitive control. Cabeza R. systems. McIntosh AR. Nat. retrieval. Cambridge. Nat. MA: The MIT Press. Hasegawa I. 20: 451–459. USA 102: 2626–2631. MA: The MIT Press. Desimone R. 3: involved in the retrieval of old episodic memory? Brain Res.) The Cognitive Neurosciences. and Raichle Eichenbaum H (2000) A cortical-hippocampal system for ME (1998) Right anterior prefrontal cortex activation during declarative memory. Donaldson DI (2004) Parsing brain activity with fMRI and mixed MacDonald SW. 475–486. Taylor JG. memory and the 29: 474–480. Proc. Ghaffar O. 8: retrieve long-term memory. 13–22. Cortex 43: 22–37. memory and the brain. and Miyashita Y (2004) Associative Interaction and modulation by COMT genotype. Sci. memory. neural networks underlying cognitive processes. 7: 54–64. and Cabeza R (2005) Remembering one cortex and episodic memory retrieval mode. and Bontempi B (2004) Sites memory: Insights from functional neuroimaging. New retrograde amnesia and the hippocampal complex. Trends Neurosci. 679–690. activation. In: Gazzaniga Marklund P. Neurosci. B 352: Gaffan D. Role in spatial learning or the automatic recording of Fuster JM (1997) Network memory. Hedden T and Gabrieli JDE (2004) Insights into the ageing mind: Murray EA and Bussey TJ (1999) Perceptual-mnemonic functions A view from cognitive neuroiscience. Neurosci. Sci. 270: recognition. 143–153. year later: Role of the amygdala and the medial temporal Acad.) The Cognitive Neurosciences III. Chelazzi L.) The Cognitive Neurosciences III. Menzaghi F. LaBar KS. Frankland PW and Bontempi B (2005) The organization of recent Meyer-Lindenberger A. Eichenbaum H (1999) Conscious awareness. memory. 4: 257–266. and Tulving E (2000) Prefrontal Dolcos F. Fransson P. Rev. In: Gazzaniga MS (ed. Moscovitch M. Brain 124: of neocortical reorganization critical for remote spatial 849–881. Nyberg L (2007) Sustained and transient neural modulations in Cambridge. 27: 442–444. Miezin FM. et al. and working memory. Rev. prefrontal cortex related to declarative long-term memory. Natl. Petersen SE. Neurosci. Cambridge. and Kouneiher F (2003) The architecture of practice-related change: fMRI and meta-analytic evidence of cognitive control in the human prefrontal cortex. 8: 594–596. Neurosci. Cogn. 21: 117–127. Neurosci. USA 97: 506–511.

J. 546–550. J. 727–737. 26: 1407–1417. J. Wagner AD (1999) Working memory contributions to human Ramnani N and Owen AM (2004) Anterior prefrontal cortex: learning and remembering. Rev. 95: 3–28. Neurosci. MA: The MIT Press. Acad. In: Gazzaniga MS (ed. (2006) The complementary-learning-systems approach.) 1376–1380. Schacter DL. Aging 18: brain systems. and recognition.) Organization of Memory. and Nyberg L (2002) Decreased activity in Tulving E (1984) Relations among components and processes inferotemporal cortex during explicit memory: Dissociating of memory. 100: 147–154.) The Cognitive detecting and remembering novel events. and Ingvar M in executive control of memory retrieval. 1. 825–837. Hillsdale. In: Tulving E Brain Res. London: Psychology Press. priming. Behav. 8: 163–183. neuroscience of constructive memory. et al. Petersson KM. memory? Curr. Bull. USA 91: 2016–2020. Nyberg L and Tulving E (1996) Classifying human long-term pp. 47–65. Rev. Insights into function from anatomy and neuroimaging. Kahn I.) The Cognitive Neurosciences III. Neurol. . Neurosciences III. Nat. Teyler TJ and DiScenna P (1986) The hippocampal memory Nyberg L. Rev. MA: The MIT Press. FIM (eds. Opin. Behav. 2: 399–406. and Raye CL (1998) The role of 49: 289–318. Rev. and factor model of declarative memory. Fenker DB. Sharot T. Nat. Nat. Vol. pp. bilateral hippocampal lesions. Nature 414: memory systems in one? Nat. Cambridge.) Trends Cogn. (eds. and consciousness: Erlbaum. (2003) Press. In: Squire LR and Schacter DL (eds. 3: 134–147. Nyberg L. Habib R. memory: Evidence from converging dissociations. In: Gazzaniga MS (ed. Hum. criterion.) The Cognitive 149–160. et al. Neuron 22: 19–22. Neurosci. Proc. 4: Psychology. Brain Sci. Oxford: Cogn. 5: 35–44. semantic memory. Trends Schmahmann JD (1996) From movement to thought: Anatomic Cogn. pp. Phelps EA (2004) The human amygdala and awareness: Tulving E. substrates of the cerebellar contribution to cognitive Norman KA and O’Reilly RC (2003) Modeling hippocampal and processing. 18: 424–428. Neurol. Rönnlund M. et al. dopaminergic midbrain participates in human episodic 110: 611–646. Miyashita Y (1999) Top-down signal from prefrontal cortex Nyberg L. Moscovitch M. 13: 281–292. Neurosurg. New York: Guilford Press. systems similarities and within-system differences. In: Jing Q et al. 33–43. Nyberg L (2005) Any novelty in hippocampal formation and Psychiatry 20: 11–21. pp. Nolde SF. J. and Bayley PJ (2004) Medial temporal lobe encoding and retrieval: The HERA model. pp. 16: 3753–3759. Shannon BJ. Microsc. Forkstam C. NeuroReport 13: Tulving E (2000) Concepts of Memory. hippocampus. 53: 1–25. and Cogn. Clark RE. 14: Schacter DL. Neurosci. Nyberg L (2002) Where encoding and retrieval meet in the brain. Bunge SA. prefrontal cortex during tests of episodic memory. 709–725. neurocognitive networks underlying episodic memory. Persson J. Nakahara K. Persson J. 163–176. Neurosci. Selective adult age differences in an age-invariant multi. Progress in Psychological Science around the World: Simons JS and Spiers HJ (2003) Prefrontal and medial temporal Proceedings of the 28th International Congress of interactions in long-term memory. et al. and Donaldson W (eds.) Oxford Handbook of Memory. Habib R. In: Lewandowsky S. Cambridge. Poldrack RA. Structural Basis of Episodic Memory 111 Naghavi HR and Nyberg L (2005) Common fronto-parietal Implicit Memory: Theoretical Issues. Natl. Psychol. and priming: Contributions from Ranganath C and Rainer G (2004) Neural mechanisms for prefrontal cortex. Clarendon. Psychol. 4: 193–202. Bowers J. Sci. novelty detection. MA: The MIT Nyberg L. Nat. Cabeza R. Science 281: Cambridge. In: Gazzaniga MS (ed. et al. activations common to encoding and retrieval. Schacter DL. Psychol. Craik FIM. Rev. Kapur S. Obhayashi M. MA: The MIT Nyberg L. (2001) Vann SD and Aggleton JP (2004) The mammillary bodies: Two Interactive memory systems in the human brain. Tulving E (1983) Elements of Episodic Memory. Neurosciences. Rotte M. Sci. 691–708. Eur. and Buckner RL (2005) awareness and implicit memory: The retrieval intentionality Parietal lobe contributions to episodic memory retrieval. and Koutsaal W (1998) The cognitive 390–425. New Persson J and Nyberg L (2000) Conjunction analysis of cortical York: Oxford University Press. Wagner AD. Psychol. 193–203. (2000) Large scale indexing theory. and Kirsner K (eds. Neurosci. 5: 184–194. Tech. Sci. 381–403. 637–648. Tulving E (2002) Episodic memory: From mind to brain. Dunn J. neocortical contributions to recognition memory: A Schott BH. NJ: activity in attention. In: Gazzaniga (1994) Hemispheric encoding/retrieval asymmetry in MS (ed. Bull. Cabeza R. New York: Academic. and Tulving E (1996a) PET studies of Squire LR. Rev. pp. Shared demands on integration? Conscious. Psychol. Cambridge. Nature 401: (2002) Brain imaging of human memory systems: Between. 4: 174–198. Tulving E (1972) Episodic and semantic memory. Rugg MD (2004) Retrieval processing in human memory: Wagner AD. In: Tulving E and Craik 1–5. 7: 257–268. Annu. Hasegawa I. and Booker J (1989) Intention Wagner AD. In: Gazzaniga MS memories: Remembering and forgetting of verbal (ed. (1996b) Network Press. Neurosciences III. and Phelps EA (2004) How emotion Nyberg L (2006) Imaging cognition: Recent developments and a enhances the feeling of remembering. 699–703. J. et al. Paré-Blageov EJ. MA: The MIT Neurosci. pp. Clark J. memory retrieval. Norman KA. Brain Mapp. 1005–1015. (1998) Building Electrophysiological and fMRI evidence. Psychonom. and Houle S Interactions between emotion and cognition. Johnson MK. Neurosci. Cabeza R. episodic memory: Positron emission tomography findings. 12: 163–173. Delgado MR. Cambridge. memory. 1188–1191.) The Cognitive Rev. memory formation: Evidence from genetic imaging. Squire LR and Knowlton BJ (1995) Memory. pp. Neurosci. Cogn. pp. Maitland SB. experiences as predicted by brain activity. McIntosh AR. function and memory. Press. Tomita H. Rev. Cogn.) Neuropsychology of Scoville WB and Milner B (1957) Loss of recent memory after Memory. pp. 7: tentative hierarchical cognitive model. Res. 51: 39–44. Annu. Seidenbecher CI. analysis of positron emission tomography regional cerebral Stuss DT and Benson DF (1984) Neuropsychological studies of blood flow data: Ensemble inhibition during episodic the frontal lobes. 9: 445–453.) The Cognitive Neurosciences III. Psychol. Neurosci. and Badre D (2004) Cognitive control.

episodic memory: The frontal lobes and autonoetic Wheeler MA. and Tulving E (1997) Toward a theory of disconnection syndrome? Neuropsychologia 20: 233–248. habit formation.112 Structural Basis of Episodic Memory Warrington EA and Weiskrantz L (1982) Amnesia: A Wheeler MA. 1: 525–536. Int. Psychol. damage produces episodic memory impairment. J. Stuss DT. 121: 331–354. Soc. and Tulving E (1995) Frontal lobe consciousness. Yin HH and Knowlton BJ (2006) The role of the basal ganglia in Neuropsychol. Rev. Stuss DT. Bull. Neurosci. 7: 464–476. Nat. .

3. 2007). memory has undergone considerable evolution events.1 Reason #1 to think that property regions are involved in conceptual-level processing: Activity in category regions transcends stimulus features 124 3.07.4 Two Case Studies in Category Representation: Animate Entities and Tools 121 3. semantic idea of semantic memory as a distinct memory memories lack information about the context of 113 .3 Object Categories in the Brain 121 3. National Institute of Mental Health.3 Reason #3 to think that property regions are involved in conceptual-level processing: Retrieving information from memory depends on reactivating property regions engaged while learning that information 125 3. semantic memory since that original formulation (for a brief history covers a vast cognitive terrain.4. This chapter tobiographical memory.4. are recollect) past experiences (Tulving.07. system that allows us to remember (consciously mantic memories. 3.5.07. 2002) and per- represented in the brain. Martin and W.4.5. Bethesda. ideas haps may also be critical for imagining and/or about the neural systems underpinning conceptual simulating future events (Hassabis et al.3.4.07.07. (Tulving.3. in ogy and neuropsychology dating back at least to the contrast. In recent times.3 Cortical Lesions and the Breakdown of Semantic Memory 115 3.07. USA Published by Elsevier Ltd. ideas.07.07. MD. K. Episodic memory is the focuses on our current understanding of how se.5 Summary 127 References 127 Published by Elsevier Ltd.3.6 Learning about Objects by Building Property Circuits 126 3.5. to details of the properties of semantic memory in relation to public events and mathematical equations.07.4.1 Introduction system began in 1972 with Endel Tulving’s distinc- tion between semantic and episodic memory Semantic memory refers to a major division of long.4 The Organization of Conceptual Knowledge: Neuroimaging Evidence 118 3. 3. especially object concepts.07.07. Although the notion of episodic term memory that includes knowledge of facts. In current formulations.1 Introduction 113 3. Semantic memories. to the episodic memory. 1972).1 Models of category-specific disorders 117 3. however. episodic information that allows us to identify objects and memory can be thought of as synonymous with au- understand the meaning of words.07. the experience. ranging from informa.4. and concepts.3. As we discuss later.07.3.2 Reason #2 to think that property regions are involved in conceptual-level processing: Activations in property areas occur as property inferences 124 3.1 Neuroimaging of Semantic Memory 118 3..07 Structural Basis of Semantic Memory A. 2002).1 Object Concepts 115 3. 3.2 Object Concepts as Sensorimotor Property Circuits 119 3. Simmons. knowledge have a long history in behavioral neurol.3. Thus.07.07.4.2 Semantic Dementia and the General Disorders of Semantic Memory 116 3.5 Category-Related Activations in Property Regions Are the Bases of Conceptual Representations of Objects 123 3.07. 2007. are devoid of information about personal late nineteenth century.07.07.07. it remains helpful to describe tion about historical and scientific facts.2 Semantic Memory and the Medial Temporal Lobe Memory System 114 3.2 Functional neuroanatomy of category-specific disorders 118 3.07.4. see Tulving.4. Schacter and Addis. Unlike episodic memories.3 Category-Specific Disorders of Semantic Memory 117 3.

and newspaper. 1997). 2002). These cases of developmental amnesia have substrates might be at least partially independent disproportionately better semantic than episodic (Nielsen. In contrast. and that you had a slight head. Simons For example. cases of developmental delay resulting et al. amnesia came plished by the surrounding neocortical structures in two types. As Nielsen noted. thereby noting that their respective neural 1997)... isolation. Nielsen further noted that there were different showing that carefully selected patients with damage varieties of categorical amnesias. 1958). Recent nal experiences. studies of patients with con. 1998). indeed. 1998). Hamann and Squire. others have argued that perhaps first proposed by the American neurologist acquisition of semantic memories can be accom- J. see Squire and were dependent on a distinct neural substrate was Zola. and personal dimensions like how we felt Medial Temporal Lobe Memory System at the time the event was experienced. entorhinal. acquire information about the world. The other type. toast. 1995. First. standard measures of vocabulary and general knowl- ing both semantic and episodic memories. Nielsen’s claim (Hodges and Graham. it is commonly assumed results in deficient acquisition of new information that episodic memory emerged as an add-on or about vocabulary and famous individuals (e. before discussing those from hippocampal damage. However. Nielsen (1958). 2003). rather than arising as an independent perirhinal cortices). including amnesias limited to the hippocampus are impaired in learning for animate objects and amnesias for inanimate semantic information about public events (Manns objects (Nielsen. One type.. acquisition of semantic standing the words you are now reading. that you Studies of patients with impaired episodic memory read the newspaper. and system. Knowing have established three broadly agreed-on facts about and. including the hippocampal region (CA sodic memory. These findings . Indeed. although retrieval of semantic memory However. resulting from damage to the medial temporal lobes ache is dependent on episodic memory. 2000. conscious mediation. like cereal. edge (Vargha-Khadem et al. and perform in the average range on structures play a critical role in acquiring and retriev. like episodic memory.. which he termed categorical amnesia. especially mammals Verfaellie et al. the semantic memories requires medial temporal lobe organization of semantic memory can also be structures. were able to acquire normal lan- damage to the medial temporal lobes. although failing to pro- patients. Neely. 1988.g. Finally. 2002). there is disagreement concerning the revealed via implicit tasks such as semantic priming role of the hippocampal region. as well as under. we first discuss studies of semantic memory vide accurate descriptions of their daily activities in patients with profound amnesias resulting from (episodic memory). despite broad agreement that acquiring often requires explicit..114 Structural Basis of Semantic Memory learning. nal information about day-to-day occurrences.. this claim comes from studies of individuals who Nielsen also maintained that the temporal (episodic) have sustained damage to the hippocampus at birth and categorical (semantic) amnesias could occur in or during early childhood (Vargha-Khadem et al.g. keep up with their have provided evidence that medial temporal lobe schoolwork. is memories is dependent on medial temporal lobe dependent on semantic memory. embellishment to semantic memory (Tulving. memories. semantic memory is considered to be fields. which he termed temporal alone. One position holds (e.. see Mishkin et al. they (Manns et al. 1991). In relation to epi. In fact. participation of the hippocampus is not neces- amnesia. that the hippocampus is necessary for acquiring The idea that our semantic and episodic memories semantic information (for discussion. patient SS) and public events and birds. was defined by a loss of memory for perso- sary (for discussion.. Damage to these structures evolutionary development. Although many animals. 1946. was defined by a loss of acquired studies seem to favor the hippocampal position by facts. 2001.07. and subiculum) and surround- both a phylogenically and an ontologically older ing neocortex (parahippocampal. Remembering that you had cereal and toast for breakfast.. being able to visually recognize objects the functional neuroanatomy of semantic memory.M. Gabrieli et al. structures. including situational properties like time 3. suggesting that the hippocampus may ceptual deficits have provided some support for not be necessary for acquiring semantic information. presaging a distinction et al. 2003). and the extent of this deficit is are assumed to lack the neural machinery to con. roughly equivalent to the deficit for acquiring perso- sciously recollect detailed episodes of their past.. One potentially important caveat to that is prominently highlighted later in this chapter.2 Semantic Memory and the and place. 1958). dentate gyrus. These studies guage and social skills.

An object concept refers to the representation (i. monly referred to as the ‘basic level. Although discussion of Breakdown of Semantic Memory this important issue is outside the purview of this chapter. 2005. impaired knowledge and ‘living things’ and has subordinate categories about objects and their associated properties acquired such as ‘poodle’ and ‘collie. because these patients have either no or. for an extensive amnesia onset remains intact as assessed by both review of cognitive studies of concepts).e. To address this issue. Most impor. sal of the role of the hippocampus in memory more commonly. public event knowledge is temporally the studies of amnesic patients.. more category sorting (e. Schmolck et al. The reason for this is that these tasks regard to the extent of medial temporal lobe damage allow memory performance to be assessed for events and its behavioral consequences. Evidence in favor of the claim of a time.. It is the level used nearly . Brooks. for memory consolidation (Moscovitch et al. 1992).. investigators status of information acquired prior versus after the have turned to patients with relatively focal lesions amnesia onset... limited damage to regions outside retrieval). a critique of prolonged consolidation and a reapprai. object property verification. that belongs to the superordinate categories ‘animal’ tantly for our present concerns.. Manns et al. Rosch. these studies have graded. that is used to pound nails. 2005). with the broadest knowl- object naming. The length of this temporally graded amnesia. a consistent with the claim that the hippocampus has a ‘hammer’ means that we know that this is an object time-limited role in retrieving semantic memories. Structural Basis of Semantic Memory 115 pose a clear challenge to the standard hippocampal 3. For example. can be surpris- ingly long and probably varies as a function of type of 3. but see Moscovitch et al.3 Cortical Lesions and the model of declarative memory. 2002. 1999. which is. 2003). and object edge represented at the superordinate level.1 Object Concepts information tested and testing method. ated by a test of recall but less than 5 years when 2002). it is certain that a reconciliation of this Studies of semantic memory in amnesia have con- issue will depend on detailed and direct comparison centrated largely on measures of public event of adult onset and developmental amnesias with knowledge.. to cortex outside this region (e.07.1995. 2002) and show specific knowledge at an intermediary level com- normal semantic priming (Cave and Squire.g.3. explicit and implicit tasks. 2004). ‘dog’ is a basic-level category types are stored in the cerebral cortex. presumably whether the memory impairment shows a temporal related to a prolonged consolidation process (Squire gradient – a critical issue for evaluating theories of and Alvarez. 1978). so we do not have to Conceptual information about the meaning of objects rediscover this property each time the object is and words known to be acquired decades prior to encountered (see Murphy. 1976. the poral lobe damage but rather to the extent of damage basic level has a privileged status (Rosch et al. for example. For example.g.’ and the most The third major finding established by studies of specific information at the subordinate level (Rosch. That is. However.. amnesic patients is that semantic memories of all 1978).07. the medial temporal lobes. Eleanor Rosch and colleagues in the 1970s. 2003). with increasing accuracy for events further focused predominantly on measures designed to probe in time from the onset of the amnesia (Kapur and knowledge of object concepts. These measures also allow perfor- amnesic patients is that the medial temporal lobe mance to be evaluated for events that occurred at structures have a time-limited role in the retrieval different times prior to amnesia onset to determine of semantic memories. The primary function of a concept is to allow evaluated by a recognition test (Manns et al.’ As established by prior to amnesia onset is not related to medial tem. Patients with damage to A major feature of object concepts is that they are the hippocampal region are unimpaired on tests of hierarchically organized. in turn. the temporal gradient for information stored in memory) of an object category news events lasted from 10 to 15 years when evalu- (a class of objects in the external world) (Murphy. In contrast to example. in the Manns et al. however. Such studies have revealed that. known to have occurred either prior to or after The second major finding established by studies of amnesia onset. identifying an object as.. they are not informative limited role for the hippocampal region in retrieving about how semantic information is organized in the semantic memories comes from studies assessing the cerebral cortex. for compromising different cortical areas.. study. Levy et al. us to quickly draw inferences about an object’s prop- Studies of object and word knowledge are also erties.

The initial neuropathological and 1992). left- (Warrington. As discussed patients can sort objects into superordinate categories. However. the other to functional characteristics dementia (SD) (Snowden et al.g. or a poodle). no difficulty indicating which are documented the neurobiological reality of this hier.. gory fluency). 1996).. These disorders are considered general ‘goat’) or a superordinate term (‘animal’) (Warrington. of the disorder. Many of Recent studies have expanded our understanding these patients suffer from a progressive neurological of SD in two important ways: one related to location of disorder of unknown etiology referred to as semantic neuropathology.e. to verify category membership (i. including the poster- the name of another basic level object from the ior portion of the fusiform gyrus. Martin and Fedio. the basic level is the easiest Broad levels of knowledge are often preserved.. when con- General Disorders of Semantic Memory fronted with a picture of a specific basic level object Several neurological conditions can result in a rela.g.. recent (e.g. certain domains of (motor-based properties related to the object’s custom. General disorders of semantic memory are imaging studies of SD indicated prominent atrophy also prominent in patients with Alzheimer’s disease of the temporal lobes.3. Gorno-Tempini et al. impaired generation of the names of objects SD as is atrophy in the most anterior regions of the within a superordinate category (i. Hodges et al. advances in neuroimaging that allow for direct and The defining characteristics of semantic dementia..2 Semantic Dementia and the object property knowledge. they are not category specific. (Mummery et al. ‘dog’ rather than agnosias. 1983). rather. In fact.. It is also the level at which we are fastest sented in a single modality like vision but. semantic cate. including the temporal impairment than SD patients) (Martin and Fedio. the impairment is not limited to stimuli pre- ‘poodle’). 2005). next. 2004. Martin and Fedio. Specifically. Thus. the collies and poodles have similar shapes and patterns semantic deficit reveals a hierarchical structure. Patterson sided atrophy seems to impair information about all and Hodges.07. For example. Hart and Gordon. and the like archical scheme and the central role of the basic level (Warrington. In agreement with category members share the most properties (e.. It is also the level at which subordinate tactile) or format (words. 1990). for example. knowledge may be preserved. which are foods. boundaries. Finally. These def. and an inability to retrieve information The other major advance in our understanding about object properties – including sensory-based of SD is that it is not as global a conceptual disorder information (shape..116 Structural Basis of Semantic Memory exclusively to name objects (e. pictures).. studies of patients with cortical damage have having. 2000. 1989. 1983. these imagine an elephant but not an animal). especially to the anterolateral (who typically have a greater episodic memory sector of the left temporal lobe. the inferior and middle temporal gyri. are relatively isolated deficits on measures shown that the atrophy extends more posteriorally designed to probe knowledge of objects and their along the temporal lobe than previously appreciated associated properties (Warrington. has been reported to same category. like ‘camel. 1995). auditory. and the pattern of ary use – but may include other kinds of information impaired and preserved knowledge appears to be not directly related to sensory or motor properties) related to the locus of pathology. animals. Rather.. Their for representing objects in the human brain. in the sense that they cut across multiple category 1975. icits include impaired object naming (with errors Williams et al. primary difficulty manifests as a problem distinguish- ing among the basic level objects as revealed by impaired performance on measures of naming and 3. temporal lobes (Williams et al. studies of the psychological nature of concepts. which are tools.’ these patients often produce the name of tively global or general disorder of conceptual another object from the same conceptual category (e. 1975). 2005).. 1975. we are faster to extends to all tasks probing object knowledge regard- verify that a picture is a dog than that it is an animal less of stimulus presentation modality (visual. polar cortex. detailed comparison of brain morphology in SD initially described by Elizabeth Warrington in the patients relative to healthy control subjects have mid-1970s. 1983) and can also occur following left hemisphere and the most anterior extent of the fusiform gyrus stroke. prominently involving the left temporal lobe (Hodges and Patterson.g. color) and functional information as initially thought.e. knowledge. of movement). 1983).. In contrast to modality-specific object categories except person-specific knowledge . 1975. the amount of atrophy in typically consisting of semantic errors – retrieving ventral occipitotemporal cortex. or retrieving a superordinate category be as strongly related to the semantic impairment in name). level at which to form a mental image (you can easily whereas specific information is impaired. Martin and Fedio.

taste) and motor proper- of Semantic Memory ties associated with the object’s use (Martin et al... most turn is associated with involvement of the right ante. the mediated properties such as where an object is typi- modern era of the study of category-specific disorders cally found (for discussion of sensory/functional began in the early 1980s with the seminal reports of models. motion. to explain these disorders. flowers).07. 1985. neurologists during the late nineteenth and early twentieth centuries. which in posited by Warrington for her initial cases.. conspecifics. quickly and efficiently. property-based organization of conceptual knowl- ematical concepts (Cappelletti et al. and other behavioral (Dehaene et al..3. On this including common tools – than for animals and other account. In this Warrington and colleagues (Warrington and Shallice. except that the impairment is largely ical for defining and for distinguishing among limited to members of a single superordinate object category members. 2005). our evolutionary history provides the major living things (for extensive review of the clinical constraint on the organization of conceptual knowl- literature.g. 2003).. 2003). members of the superordinate category (e. a domain edge.. animals is assumed to be heavily dependent on know- patients with category-specific disorders have diffi.3.g. 2003). Caramazza and Mahon. 1999). 1998.07. smell.1 Models of category-specific neural machinery for solving. This is because visual appear- members of this superordinate category relative to ance is assumed to be a critical property for defining members of other superordinate categories (e. cat. For example. Likely Two major theoretical positions have been advanced candidate domains offered are animals. see Forde and Humphreys. color. 2004). ing about subtle differences in their visual forms. property-based view.3. 2003). damage to regions that category. thereby suggesting a loss or that the lesion should affect knowledge of all object degradation of information that uniquely distinguishes categories with this characteristic. about a particular property or set of properties crit- istics as SD. A culty distinguishing among basic level objects (e. critical prediction of sensory-/motor-based models is between dog. than for a variety The alternative to these property-based theories is of inanimate object categories. 1958).3 Category-Specific Disorders form. 2003). Although less com. uals on the relationship between and among object Crutch and Warrington. the known patterns of category-specific disorders. Category.. four. and in some models other functional/verbally knowledge disorders date back over 100 years. a patient with a category. In a similar fashion. store information about object form. edge in the brain. furniture. 1987). not only animals. horse). property-based models (Cree and McRae. animals and because the distinction between different tools.g. disorders computationally complex survival problems.. most common have been reports of patients thus providing additional evidence in support of with relatively greater knowledge deficits for ani. 3. will produce culty naming and retrieving information about a disorder for animals.. an idea that was prominent in the writings of strongly associated with left posterior parietal cortex Karl Wernicke. Also deficits are a direct consequence of an object relatively spared is knowledge of number and math.. However. Structural Basis of Semantic Memory 117 (i. the theory proposes that selection pressures have resulted in dedicated 3.e. and form- specific disorder for ‘animals’ will have greater diffi. Warrington and McCarthy. category-specific semantic dis- 1984. related properties like color and texture. Following the explanation plant life. see Capitani et al. the domain-specific view championed most recently mon. Cognitive studies with normal individ- knowing about fruits and vegetables (Hart et al. orders occur when a lesion disrupts information specific disorders have the same functional character. Specifically.g. mate entities – especially animals. damage to regions that store legged animals) (for recent collection of papers on information about how an object is used should pro- these patients see Martin and Caramazza. and possibly tools (for a detailed discussion . current investigators assume that category-specific rior temporal lobes (Thompson et al. 2003). duce a category-specific disorder for tools and all A variety of category-specific disorders have been other categories of objects defined by how they are reported such as relatively circumscribed deficits for manipulated. consistent features and attributes show broad consistency with with Nielsen’s clinical observations (Nielsen 1946.. other patients show the opposite dissociation: a by Alfonso Caramazza and colleagues (Caramazza and greater impairment for inanimate manmade objects – Shelton. Although case reports of relatively circumscribed 2000). Sigmund Freud. Similar to patients with SD. information about famous people). The central idea is that object knowledge is organized in the brain by sensory (e. Thus.

the general tack taken sive..4.. 2002. Property-based Although the particular methods. Does this object’s name contain the letter b?). however. including the fusiform gyrus vant to real-world interactions with an object is stored (Vandenbulcke et al. 3.g. whereby the information most rele- tral temporal cortex. (e. 2006). Nevertheless. Martin.. it is certainly possible that concepts in most studies has been to compare brain activity are organized by domains of knowledge. For example. In contrast.g.. Thompson-Schill et al. As a result. concepts are represented in the brain as distributed ischemic lesions to the more posterior regions of ven. 2005).2 Functional neuroanatomy of made by these studies are further strengthened when category-specific disorders subsequent research observes activity in the same There is considerable variability in the location of brain regions using stimuli that are conceptually lesions associated with category-specific disorders for related (e. 2000). 2001.1 Neuroimaging of Semantic tion among competing alternatives (Badre et al. wide agreement that this region’s primary role is to tional magnetic resonance imaging (fMRI) to explore control and manipulate information stored elsewhere the functional neuroanatomy of semantic memory.07.. in the same sensorimotor regions active when that specific knowledge disorders for tools and their asso. some physically different (e. exhibit word retrieval difficulties while retaining con- ceptual knowledge for those same words (Baldo and Shimamura.07.. this etiology is herpes simplex encephalitis. . 1997.g. recent functional are consistent with neuropsychological findings with neuroimaging studies of the intact human brain have patients who. using a variety of concepts and object cate- 1997). it is important to stress the ventral and lateral regions of the temporal lobes. subsequent to left inferior frontal lesions. seeing photographs vs. In contrast. 1998. experimental para- and category-based accounts are not mutually exclu. of concept property information stored in other brain these cases have been relatively uninformative for regions (Bookheimer. ciated with equally difficult nonconceptual processing tent with this view. using the same stimuli. category.g. Category-specific knowledge disorders for gories. contrary.. 1998). 2003). In particular. ing sounds vs. processing consistently identify three brain regions – tices of the left hemisphere (Tranel et al. the left ventrolateral prefrontal cortex (VLPFC) and Gainotti. guiding retrieval and postretrieval selection patient to another (Capitani et al. reading names of artifacts).3. does not mean that objects’ physical properties are tion with a predilection for attacking anteromedial and unimportant to their neural representations. information was acquired. However.118 Structural Basis of Semantic Memory of these models. there is used positron emission tomography (PET) and func.. 1998).. Much of the functional object a man-made artifact?) versus the activity asso- neuroimaging evidence to be discussed later is consis. and stimuli vary widely. 2000). see Caramazza. 2005. digms. judging whether objects are artifacts) but animate and inanimate entities. begun to shed some light on this thorny issue.4 The Organization of Information about this region’s role in semantic Conceptual Knowledge: Neuroimaging memory has been augmented by recent functional Evidence neuroimaging studies that find distinct mechanisms within the VLPFC for information retrieval and selec- 3. Such category-specific knowledge disorders for animals findings demonstrate that the regions are responding are disproportionately associated with damage to the to the stimuli’s conceptual content rather than their temporal lobes (Gainotti. The neuroanatomical claims 3. large functional neuroimaging literature demonstrates specific knowledge disorders are often large and show that the VLPFC serves as a control center for semantic considerable variability in their location from one memory. These functional neuroimaging findings ories in cerebral cortex. Thompson- questions concerning the organization of object mem. Schill.. Memory Although claims for dissociable retrieval and selection For nearly two decades cognitive neuroscientists have subregions in VLPFC remain controversial. (Gold et al.. it has been well documented that object animals also have been reported following focal. a viral condi.07. ciated actions have been most commonly associated Studies comparing conceptual to nonconceptual with focal damage to lateral frontal and parietal cor. The most common physical characteristics. hear- general tendencies can be observed. On the inferior (ventral) temporal cortices (Adams et al. 2006). 2003). implemented when subjects engage in tasks requiring the encoding in the brain by large-scale property-based systems or retrieval of conceptual information (e. A that the lesions in patients presenting with category. property circuits. As we will see. 2003).3. Is this (Mahon and Caramazza.

they observed that the most about object concepts. In particular. Color properties > motor properties Early functional neuroimaging research using PET imaging supported these lesion study findings.0001).10 within the color verification within the color perception ROI to a pencil) elicited activity in premotor cortex as perception ROI 0. strong evidence for these accounts combined green and red patches). ... a significant body of cated whether color or motor property words could be research demonstrates that various property regions true of a concept word. The functional overlays ities as well. however. with sound. Wiggs et al.01 that retrieving object property information activated with a cluster size of at least 108 mm3).. Martin and colleagues (Martin et al. saying ‘yellow’ in response to an achromatic picture of a pencil or to 0. Green patches indicate regions where activity sory. The y-axis indicates percent signal change relative demonstrated a direct overlap in the neural bases of to signal baseline. between knowledge retrieval and sensorimotor sented in the human brain come from how information is organized within these regions. 2006). Next. Structural Basis of Semantic Memory 119 3. saying ‘write’ in response 0. these studies find color-responsive region in the perception task.g. the average BOLD has come from fMRI studies demonstrating direct response to color property words in the property verification overlaps in the neural bases of knowledge. also see Chao and Martin.2 Object Concepts as Sensorimotor brain regions underlying color perception. 2007). In contrast.. Using a property production task in which subjects were required to generate a word describing a specific property of a visually presented object. touch. located that the posterior ventral and lateral temporal lobes are particularly important for storing information in the left fusiform gyrus. Beauchamp MS.00 well as a region of the left posterior middle temporal gyrus (pMTG) just anterior to primary visual Figure 1 Overlap in perceptual and conceptual color motion-selective cortex MT/V5.30 % signal change gyrus just anterior to regions activated when subjects Motor passively viewed color stimuli. in Property Circuits separate scanning runs. and taste properties represent Talairach-normalized group data from the random activating the corresponding auditory. On top. where color perception produced a greater response than grayscale perception (in other words. color perception and color knowledge retrieval. they first mapped the perceiving and knowing about color. Martin A. somatosen. effects analysis. Using the color perception task located in or near perceptual cortex store information as a functional localizer.006).. 1995. with error bars representing  1 standard error of the subject means. Blue patches indicate regions where activity was greater for verifying color properties Functional neuroimaging findings demonstrating than motor properties in the knowledge retrieval task (p < 0. perception.4. Neuropsycholgia. 2001. As described earlier. (in press) (p ¼ 0. within the recently. The red patch in the left regions near perceptual and motor cortex are highly fusiform gyrus indicates the region of overlap between the two suggestive of the sensorimotor hypotheses generated tasks. and an attention-demanding task requiring fine-grain dis. 2003. the figure depicts sagittal and coronal sections from the N27 template brain warped to Talairach have now been observed for other property modal- space (template available in AFNI). they presented subjects with a verbal property verification task in which they indi- In addition to the VLPFC. McRae K. Goldberg et al. Simmons et al. Similar effects processing. We will see knowledge (Figure 1).. knowledge retrieval relative to retrieving motor Thompson-Schill. Using Ramjee V. color perception task (p < 0. was also activated for color about object concepts (Martin and Chao.40 its written name) elicited activity in the fusiform 0. producing 0. For example.g. 2001. cases of category-specific Color wheels > grayscale wheels Overlap of color perception and deficits point toward a central role for property infor- color knowledge retrieval mation in the organization of semantic memory. 1999) demonstrated that producing color-associate words (e.20 property Color property verification action word associates (e. and gustatory cortical regions (Kellenbach was greater for processing color than grayscale wheels in the et al. 1999. Martin K.07. The inset bar graph demonstrates that within the left in the literature on category-specific disorders. Barsalou LW (in press) A common neural substrate for criminations among color hues. More fusiform ROI. Evidence for direct overlaps that important clues about how knowledge is repre. Adapted from Simmons WK. Martin. task was greater than the response for motor property words and action.

how objects look).. a low-level picture repetition detection task was respectively) (Hauk et al. the most salient properties of appetizing For example. Simmons actions performed with a particular body part (e. When necessary. The green squares in the insula/operculum at Z ¼ 20 and Z ¼ 9 represent peak activations observed when participants taste sucrose. and thus it is a likely candidate for being the center of flavor representation (de Araujo et al. and how demonstrated that simply reading words referring to rewarding they are to eat. pick) activated the corresponding motor appetizing foods (e.120 Structural Basis of Semantic Memory property systems are not limited to color properties. et al. the red circles in the insula/operculum at Z ¼ 9 and in the OFC at Z ¼ 18 and Z ¼ 6 indicate the activation peaks observed in the present study when participants viewed food pictures. Pecher. enough to elicit bilateral activity in ventral occipito- Together.. cookies. then object categories should have predict. (2005) demonstrated that viewing images of lick. (b) Locations of peak left orbitofrontal cortex (OFC) activations across various tasks. The aqua squares in the insula/operculum at Z ¼ 5 and in the OFC at Z ¼ 18 represent peak activations when participants tasted glucose (Francis et al. and hand representations. Poster presented at the Annual Meeting of the Cognitive Neuroscience Society. This appears to be the case. The pink squares in the insula/operculum at Z ¼ 10 and in the OFC at Z ¼ 6 indicate the peak activations observed when participants taste umami (de Araujo et al. Cortex 15: 1602–1608. resides in the the sensorimotor account of knowledge representation. whereas the green square in the lateral OFC at Z ¼ 10 is the peak activation in the area observed to respond to the combination of gustatory and olfactory stimuli. left orbitofrontal cortex (OFC) (secondary gustatory able neural representations based on their multimodal cortex) activated in prior fMRI studies when subjects property profiles... The blue diamond in the insula/operculum at Z ¼ 13 indicates an area of common activation when participants tasted either glucose or salt (O’Doherty et al. April 2003). The blue circle in the OFC at Z ¼ 10 represents peak activation observed when participants verify the taste properties of concepts using strictly linguistic stimuli (Simmons. New York. Martin A. Yellow diamonds in the inferior medial OFC represent peak activations observed when participants receive abstract rewards (O’Doherty et al. NY. how they taste. If this right insula/operculum (primary gustatory cortex) and is correct.g. Adapted from Simmons WK. pizza) while performing cortex (e. 2004).. kick. and Barsalou LW (2005) Pictures of appetizing foods activate gustatory cortices for taste and reward. Consistent with color or the actions associated with it. same sensorimotor regions that are active when that the researchers also observed activity in regions of the information is experienced in the external world. such as its information (e.g. Cereb. coordinates reported in other studies were converted from Talairach to MNI space. temporal regions tuned to represent object form edge about a particular object property. foot. and Barsalou. (a) Z = 20 Z = 13 Z = 10 Z=9 Z=5 Z = –9 (b) Z = –30 Z = –24 Z = –20 Z = –18 Z = –10 Z = –6 Figure 2 (a) Locations of peak right hemisphere insula/operculum activations reported in taste perception studies.... For received tastants orally in the scanner (Figure 2). these findings support claims that knowl. 1999). 2003b). face. Finally. Hamann.g. example. Zeelenberg. Using fMRI. Pulvermuller and colleagues have foods are how they look. 2003a).g... 2001a). 2001b). .

. 1984. rather tral occipitotemporal cortex (Haxby et al. and motor properties.07. as well as animals and tools.07. and some of the best evidence to this effect comes from the study of two broad classes of knowledge: animate entities and tools. Epstein and Kanwisher. 2007 for reviews). Clearly. For now. throughout the brain.4 Two Case Studies in Category bases of object concept knowledge have presented Representation: Animate Entities and Tools subjects with photographs of exemplars from various object classes.g. Rather. that the ventral occipitotem. Reddy and Kanwisher. and in a later study.. see Martin and Chao. fusiform gyrus. Chao et al. 2000). 1999. for review). (Figure 3). including medial regions in the Martin.e. at least some aspect of object knowledge is maintained 2002. much research interest. that naming pictures or reading words denoting ani- Perhaps the most well-known category-responsive mal concepts activated lateral regions in the fusiform brain region is the fusiform face area (FFA). how- in a modality-specific. perceptual format (Barsalou.4. representation of concepts from any particular cate- Findings to this effect provide strong evidence that gory (Haxby et al. which gyrus (located along the ventral surface of the tem- responds reliably and selectively to face stimuli poral lobes and including the FFA). stream stretching along the ventral surfaces of the . through which the information was acquired (Fodor. In addition. knowledge of animate entities and small. respectively and how they move but also their function-associated (Chao et al. 2001. it is unsur. (e. the property regions that compose a 2003). Spiridon and Kanwisher. (1999) demonstrated responsive to some categories relative to others. each activating underlying not only what these objects look like lateral and medial fusiform cortex. In contrast. comparisons between classes of while perceiving animate entities’ most salient prop- objects demonstrate that local regions within the erties. This stands in stark contrast to accounts concept representation within a single brain region. tools) activates a distributed neural circuit 1998). 1997. however. In contrast. it remains an open question as to concept’s neural representation overlap with the how central these nonpeak areas are in the cognitive brain regions mediating that property’s perception.. fMRI pattern analysis techniques. Structural Basis of Semantic Memory 121 These findings demonstrate that information about have demonstrated that they are also associated with object-associated properties is stored and represented distinct neural signatures across large swaths of ven- across numerous property regions in the brain. Kintsch. posterior extent of the superior temporal sulcus 2006. objects (i. Spiridon and Kanwisher. these authors demonstrated which reliably activate a region in parahippocampal that performing these tasks with manipulable artifacts cortex (Aguirre et al. the pMTG. As we will see. distinct categories The fusiform region activated by animal and tool are associated with activation peaks in particular stimuli is part of a larger object-form processing regions. to this debate about the distributedness of 1999).. human functional neuroimaging evidence leaves lit- dal. 2006). 1998). For example. 3. than in a single unitary semantic memory storehouse. Cox and Savoy. 2001.. located laterally along the temporal lobe categories include environmental scenes (places). 1998. This does studies have shown that tasks involving animate not imply. Grill-Spector and Malach. as well as the (Kanwisher et al. The topographic relations among these category. however. Other frequently studied object (pSTS). ever. 2001.4. these categories invariably activate this region. namely. 2004). Given that a large body of monkey Motivated by category-specific deficits for animals neurophysiology and human neuroimaging evidence and tools reported in the neuropsychological litera- indicates that the occipitotemporal cortex plays a ture. focused on defining the neural substrate underlying 2003..3 Object Categories in the Brain The vast majority of studies examining the neural 3. posterior parietal and ventral premotor regions responsive ventral temporal regions are a topic of (Chao and Martin. 2002. Conceptual knowledge is unequivocally distributed 1975. propositional. and linguistic formats that bear tle room for debate as to whether conceptual arbitrary relationships to the perceptual experiences information is distributed across brain regions. what they look like and how they ventral occipitotemporal cortex are particularly move. see Kanwisher and Yovel. many functional neuroimaging studies have central role in object perception (Grill-Spector. Pylyshyn.. people and animals) are associated with poral cortex is an undifferentiated object-processing activity in the distributed neural circuit engaged system. manipul- prising that in these studies the different object able artifacts such as tools. describing human knowledge solely in terms of amo.

sound) predicts the conglomera- moving stimuli of humans (photographs and video tion of sensorimotor regions underlying that object clips of people performing actions such as jumping. medial object concepts from their preferred categories – region of the fusiform gyrus). fully articulated motion vectors that char- stream the form features of visual inputs are processed acterize biological motion (Oram and Perrett. pMTG). Beauchamp et al. Grill-Spector and Malach. that activity in ventral temporal cortex differentiates rior to the much-studied visual motion area V5/MT along category boundaries. lateral regions responded more strongly to depictions of humans. (b) Lateral view of the left hemisphere showing relative location of regions assumed to form features that are present in both static and represent biological motion typical of animate entities (3.. strate that an object concept’s property profile (e. The pSTS responded associated with using tools. clearest evidence for the specific functions played by As reviewed earlier. As expected.. Beauchamp et al. lateral region of the fusiform gyrus. light of this. presumably due to differ- (Watson et al. In a subsequent study. the two able artifacts also recruit posterior parietal and ventral classes of stimuli activated distinct regions in the premotor regions supporting the representation of .. 2004).) regions of the left hemisphere assumed to represent information about the motor movements strong category selectivity. poral regions were more responsive to dynamic than typically centered on the intraparietal sulcus) and ventral static stimuli. 2005). ent visual form characteristics for animate objects and Although the location of these category-responsive manipulable artifacts. and sitting) and manipulable objects (e. motion. in the same way animal and tool stimuli are located immediately ante. with more ante.122 Structural Basis of Semantic Memory ventral and lateral temporal lobes (Figure 4).g. (2003) observed that the lateral and medial fusiform responded much more to videos of humans and tools. with the pSTS and pMTG exhibiting premotor (6. 2003). concept’s storage and representation in the brain. Grossman and Blake. 1982). 2002. subjects were shown static and form. 2007) evidence demon- comes from the work of Beauchamp and colleagues. moving in characteristic ways) form and motion properties. we can infer that the lateral and medial form-related properties like color and texture of animate fusiform regions are not modulated by motion but. and scissors. In walking. Taking these two sets of findings location of regions assumed to represent visual form and together. relative to human actions. entities (1. the about what objects look like (Riesenhuber and pMTG responded more strongly to the rigid... in a hierarchically organized manner. lateral tem- shown are the relative locations of the posterior parietal (5. taste. however. these two regions in conceptual processing per se 2003) and imaging (see Martin. (2002) observed that in the fusiform gyrus. In culated motion vectors characterizing dynamic contrast.. (a) Ventral view of the right hemisphere showing relative their motion vectors. 2001. unarti- Poggio. 1998. Importantly. including. Within this so-called ventral visual to flexible. 1993). than they did to point-light displays of Figure 3 Schematic illustration of regions exhibiting humans and tools that lacked the form and color category-related activity for animate entities such as animals features of the video stimuli. the fusiform face area) and tools. the lateral temporal regions activated by depictions of tool motions. dynamic depictions of an object. This finding is consistent with monkey neurophysiology and human fMRI studies occipital and temporal lobes (Ungerleider and demonstrating that this region is particularly tuned Mishkin. respond to the form features characterizing not limited to. its In a series of studies. respectively. 1994. the properties of the two categories. so activity in lateral temporal regions relative to form and motion processing areas cortex similarly differentiates the distinctive motion was suggestive as to their functional significance. more strongly to dynamic depictions of human actions than to tool motion. and medial regions responded more to manipulable objects. both regions responded equally to their preferred stimuli. it should come as no surprise that in photographs and video clips of tools such as hammers. pSTS) and rigid motion vectors typical of tools (4. Thus. In contrast. but rather. Puce et al. behavioral (Cree and McRae. Also Unlike the ventral temporal cortex. (2.g. Pelphrey rior regions representing higher-order information et al. addition to the temporal regions representing their saws. tasks involving manipul- (Beauchamp et al. regardless of whether those stimuli were static or dynamic. but which maintained and people (red) and manipulable artifacts such as tools (blue). 1999.

4.3 0.1 0.4 0. Lee KE. in Property Regions Are the Bases of ogy evidence demonstrating that neurons in the ventral Conceptual Representations of Objects premotor and parietal cortices respond when monkeys Processing various object categories elicits activity in grasp objects.1 0.2 0.1 0.0 –0. and Martin A (2002) Parallel visual motion processing streams for manipulable objects and human movements.1 0. and pSTS regions that are more responsive for identifying people than tools (yellow). Dashed lines indicate 1 SEM.6 0. . Rizzolatti and Fadiga. where as ventral areas depicted in (b) exhibit only category effects.4 0. 2000). as well as when they merely see objects sensorimotor property regions.5 0.4 0. But how do we know they have previously manipulated (Jeannerod et al. Haxby JV.2 0.3 0. Adapted from Beauchamp MS.6 0.0 0.6 % MR signal change 0.2 0.1 pMTG Medial fusiform 0.0 –0. Structural Basis of Semantic Memory 123 (a) (b) pSTS Lateral fusiform 0.1 –0. that the activity in property regions represents 1995.5 Category-Related Activations This finding is consistent with monkey neurophysiol.07.2 0. 3. Neuron 34: 149–159.5 0.1 Figure 4 (a) Lateral view of the left hemisphere showing MTG regions that are more responsive when subjects identify static and moving images of tools than people (blue).. object-associated actions (Chao and Martin.3 0.1 –0.3 0. and the lateral fusiform region that is more active for identifying people (yellow).5 0. The lateral cortical areas in (a) exhibit category and motion effects. 1998). Below each brain are group-averaged bold response functions depicting activity for static and moving images in each region shown in (a) and (b).4 0.5 % MR signal change 0. Vertical gray bars indicate stimulus presentation periods. (b) Coronal section illustrating medial fusiform regions that are more responsive when subjects identify static and moving images of tools than people (blue).0 0.

is not. motor property inferences are also observed in ity has even been observed when participants view response to tool photographs. appetizing foods.. 2002. 2006.. In addition. entities. 1944). differential responses are observed in processing comes from studies in which property the lateral fusiform to animate entities in response to inferences manifest as activations in property areas. 2005. 2002.2 Reason #2 to think that conceptual-level processing: Activity in property regions are involved in category regions transcends stimulus conceptual-level processing: Activations features in property areas occur as property For both animate and manipulable artifact categories.. stimuli depicting point-light displays of human 1999.. with subjects exhibiting abstract representations of social situations depicted activations in premotor cortex. 2001. 2005). Chao et al. this region is responsive to representations of animate Devlin et al. the spoken names The ability to make inferences about an entity’s of tools (Noppeney et al. Okada et al. Whatmough et al. pictures and written names of animals (Chao et al. 3... 2005. 2002. Weisberg. often when that informa- mechanical interactions such as a bowling ball tion is not present in the immediate stimulus. findings are important because they demonstrate that Kellenbach et al.07. 2002). Wheatley et al.. the lateral Martin. Medial fusiform activity has even been categories. Beauchamp et al.5.5.4. inferences substantial evidence demonstrates that activity in the Further evidence that property regions for animate categories’ property regions is not stimulus specific. 2005. 2006). 2006). 2002). Mechelli et al. 2003. Kable et al. properties is at the very core of what most cognitive plays depicting tools in motion (Beauchamp et al. 2005). Upon viewing pictures of appetiz- 2003). 2006). Devlin (Simmons et al... even though they are in interactions among simple geometric shapes not physically manipulating the tools (Chao and (Heider and Simmel. 2003. These et al. present in the stimulus. 1999. Tranel et al. Mechelli et al. 2003). Rogers et al.. 2005. activations were observed in insula/ Similarly.07. when a response occurs within a 1999.124 Structural Basis of Semantic Memory conceptual information? Three findings in particular knocking down pins or billiards (Martin and from the literature on animate and manipulable arti. Mechelli et al.4. 2005. Noppeney et al. For example. Given . appears to be related to high-level conceptual representations. 2000. 2002. In addition to the motion property inferences. Phillips et al. vations described earlier when subjects viewed 2000. rather. operculum and OFC regions known to represent form to manipulable objects are observed in response the tastes and taste rewards of foods. 2003... et al. static pictures of tools or reading tool names activates Beauchamp et al... 2002. 2005). human voices (von Kriegstein perceiving pictures of animals or people or reading et al. has been stripped from the stimuli. 2006). 2005. scientists call conceptual knowledge. Kan et al. stimulus features but. in fact.. even when the stimuli fusiform activity is also observed in response to presented are static photographs (Chao and Martin. ing foods.. Devlin property area even though that particular property et al. 1999. mocking and bluffing (Castelli et al. 2006). Similarly. Peelen et al. Creem-Regehr and Lee. lateral fusiform activ. 2003). and manipulable objects are involved in conceptual For example.. social interactions such as hide-and-seek (Schultz These findings are in the same vein as the obser- et al.. For example. how- Perhaps most significantly. and sharing (Martin and Weisberg. Creem-Regehr and fusiform gyrus responds to animations suggesting Lee.. 2002. 2000. Beauchamp like stick figures (Peelen and Downing... 2002. logical motion (Chao and Martin. activation in these cate- fact object concepts strengthen the case that property gory-related property areas is not due to particular regions are involved in conceptual-level processing. and point-light dis.. Chao et al. in other words. 2005. Across various 2003).. Clearly. b.. even after most form and color information 2005a. Price et al. human subjects frequently exhibit activa- observed when participants view simple geometric tions in property regions that correspond to salient shapes that move and interact in ways that suggest object concept information.. ever. 2006) and the region of the pMTG known to represent nonbio- degraded and abstract visual stimuli such as human. lateral tive to biological motion.. 2000). differential responses in the medial fusi. and when simply imagining faces animal names activates the region of the pSTS sensi- (O’Craven and Kanwisher.1 Reason #1 to think that property regions are involved in 3. perceiving bodies in motion (Grossman and Blake. even though to both pictures and written names of tools (Chao subjects were not receiving any gustatory stimulation et al... 2005....

A recent finding using fMRI brain state classification Note that these ventral and lateral temporal activa. USA 101: 17516–17521. Sci. Adapted from Gil-da-Costa R.5 Normalized rCBF 55 R L coss screams non-bio (c) (e) 60 Normalized rCBF R L 57 Subject #1 Subject #2 Subject #3 coss screams non-bio Figure 5 Activation of visual cortical areas in response to species-specific vocalizations in the rhesus macaque. namely. activation occurs across a expect that the ability to infer properties would be distributed network of property regions to represent preserved across primate species. conceptual-level processing: Retrieving logical sounds were attended by activity in auditory information from memory depends on cortex. . and recent evi. The coronal slices in (d) and (e) illustrate regions in each monkey that were more responsive to conspecific vocalizations (coos and screams) than to nonbiological sounds. and in the STS (Figure 5). and likely survival value. Natl.4. Structural Basis of Semantic Memory 125 its great utility. when simply responding to features present in experimen- monkeys process information about animate entities. provides yet more evidence that property regions are tions in visual form and motion property regions involved in conceptual processing per se. rather than occurred to auditory stimuli. the conspecific calls also elicited activation in reactivating property regions engaged area TE/TEO. (2005) demonstrated that (a) E D (b) (d) 57. (2004) presented both species-specific calls and nonbiological sounds 3. Polyn et al.07.5. As with humans. Gil-da-Costa et al. (a) Lateral view of a rhesus monkey brain demonstrating the approximate locations of the coronal sections in (d) and (e). even when those prop- In a study demonstrating evolutionary continuity erties are not immediately present in the stimulus. respectively. coos and screams exhibited reliably greater activation than nonbiological sounds. the presumed monkey homologue of while learning that information human fusiform gyrus. Lopes M. Proc. et al. what they dence demonstrates that it is.3 Reason #3 to think that to awake rhesus macaques undergoing PET imaging. property regions are involved in Although both the species-specific calls and nonbio. (2004) Toward an evolutionary perspective on conceptual representation: Species-specific calls activate visual and affective processing systems in the macaque. Braun A. look like and how they move. tal stimuli. we might in this case other monkeys. in the neural mechanisms for representing concep- tual information. those entities’ salient features. Acad. Both (b) and (c) show the mean ( SEM) normalized rCBF for the activations in TE/TEO and MT/MST/STS. In both regions.

James and Gauthier found that its representation in memory. subjects were then given extensive training interacting James and Gauthier’s findings are important for at with the objects. ventral information was unnecessary for successfully per- temporal activity after training was largely restricted forming the task and not present in the stimuli. After scanning. For example. least two reasons. storage. it illustrates that this process can occur region previously implicated in representing the visual even when experience is verbally mediated. an auditory functional localizer) and viewing gree- strates that this is indeed the case. along with the findings of form a specific tool-like function. and Gauthier (2003).4. displays of human forms in motion elicits activity in and common manipulable objects while undergoing lateral fusiform and pSTS. left pMTG and was positively correlated with a subject’s behavioral parietal cortex activity best predicted manipulable performance. places.07. Prior to scanning. object recall. new activations finding is important precisely because so much of . object from the training set elicited activation in cant changes in the objects’ neural representation. namely. the amount of activity in both regions best predictor of famous face recall. the subjects were not only better at indicat- were trained on data collected during encoding were ing when a human form was present in a noisy visual able to detect distinct patterns of category-related display. each of which was designed to per. and manipulable objects immediately prior to (nonbiological motion) and left intraparietal sulcus the free recall of information about each of these and premotor cortex (physical manipulation). and learned associations. and re. who demonstrated that prop- vating the pattern of activity in property regions that erty circuits can develop even through verbal occurred during learning. suggests that conceptual viewing greebles associated with auditory properties property circuits develop out of experience with produced activity in auditory cortex (as defined by objects. sub- Building Property Circuits jects underwent fMRI while performing a visual The close correspondence between brain regions matching task that did not require retrieval of the underlying perception and action with an object. learning. simply seeing a particular using the objects in a tool-like manner led to signifi. Grossman et al. Comparing the data from the two property circuits. lateral fusiform activity served as the Interestingly. In both studies. to the medial aspect of the fusiform gyrus. this finding further estab.6 Learning about Objects by properties (e. the displays). This shape or form of tools.126 Structural Basis of Semantic Memory machine-learning algorithms are capable of detecting emerged after training in other regions observed in activity across the property regions underlying faces.g. the Weisberg et al. Brain state classifiers that training. (2004) fMRI. subjects trained subjects to perceive human forms in point- were instructed to recall the items they had learned light displays embedded within visual noise. pSTS activity in response to detecting those forms. First. viewing point-light labels and photographs of famous people. trained that a particular family of greebles were associated with an auditory property (e. After during the training phase... studies of tool knowledge. A small but growing body of literature demon. By demonstrating that property circuits with learning comes from James retrieving an item from memory depends on reacti. but they also exhibited greater fusiform and activity that occurred several seconds prior to recall. they illustrate how experience subjects were once again scanned while performing the with category exemplars leads to the development of visual matching task. During the classifier training phase of Learning effects have also been observed for ani- the study.g. For example. As described earlier. Importantly. while still undergoing fMRI. entities called greebles. Similarly. subjects learned associations between mate entities. (2007) asked subjects to perform a activity in the biological motion-sensitive region of simple visual matching task on photographs of novel the pSTS (as localized by moving point-light objects while undergoing fMRI. either premotor (for tools) or auditory and motion- Whereas the novel objects elicited only diffuse ventral sensitive cortex (for greebles). categories. even though that temporal activation in the first scan session. whereas other types of greebles had action 3. the same Second. and parahippocampal activity best Yet further evidence for the development of predicted recall of places. the left pMTG places. subjects learned verbally lishes the centrality of property-specific information presented facts about families of novel animate-like systems in the memory encoding. (2007). hops or jumps). subjects were trieval of conceptual information. After training. roars or squeaks). bles associated with action properties produced Weisberg et al. Later. After training. which can later activate as property imaging sessions revealed that physical experience inferences.

Brain 125: 1839–1849. and Caramazza A (2003) activations constitute property inferences about the What are the facts of semantic category-specific deficits? A critical review of the clinical evidence. it appears that information Barsalou LW (1999) Perceptions of perceptual symbols. Nat. three property types may form a core set of informa. Chao LL. New York: McGraw-Hill. Sci. Neurosci. J. Chao LL and Martin A (1999) Cortical representation of properties. Cereb. 991–1001. The neuropsychological and functional neuroimag. regardless of whether their respective properties distinction. properties such as form. knowledge representation. imaging (fMRI) ‘‘brain reading’’: Detecting and classifying . Exp. rather visual motion processing streams for manipulable objects than being represented in a single brain region. Lee KE. and Cappelletti M. Trends Cogn. the extant findings suggest that for 20: 213–261. Haxby JV. motion. Caramazza A and Shelton JR (1998) Domain-specific knowledge systems in the brain: The animate-inanimate tified. Insler RZ. the object. Neuropsychol.5 Summary Badre D. regions activate automatically when an object is iden. the development of property circuits as subjects Caramazza A (1998) The interpretation of semantic category- specific deficits: What do they reveal about the organization gained experience with an object. J. that future research using a wider mind: A functional imaging study of perception and array of object categories will reveal important roles interpretation of complex intentional movement patterns. Mahon B. and Ropper AH (1997) Principles of Neurology. ing findings presented here tell us much about the Baldo JV and Shimamura AP (1998) Letter and category fluency in patients with frontal lobe lesions. or functional use of. Mem. the neural bases underlying perceptual and perception. 3. naming. however. and Wagner AD (2005) Dissociable controlled retrieval and generalized selection mechanisms in ventrolateral prefrontal cortex. Pare-Blagoev EJ. for color and motor about objects.. Frith C. and Martin A (2002) Parallel across a discrete network of cortical regions. Asperger tion that is necessary and sufficient for representing syndrome and brain mechanisms for the attribution of mental stats to animated shapes. Laiacona M. 22: 637–660. Brain Sci. and Frith U (2002) Autism. Finally. Poldrack RA. 6th edn. Structural Basis of Semantic Memory 127 our knowledge is acquired verbally. Neuron 34: 149–159. Kopelman MD. J. Chao LL and Martin A (2000) Representation of manipulable man-made objects in the dorsal stream. 15: most commonly engaged during perceptual experi. Behav. which claim that concep. 22: 771–793. Neuropsychology 12: organization of conceptual knowledge in the brain. Frith U. 25: 151–188. Adams RD. and Martin A (2003) fMRI addition. Annu. Neuropsychol. As such. Neuron 21: 373–383. Neurosci. There is good reason to Castelli F. 10: 1–34. and Martin A (2002) Experience- tual property information is stored in the perceptual dependent modulation of category-related cortical activity. Happe F. of conceptual knowledge in the brain? Introduction. Cogn. and Frith C (2000) Movement and believe. rather than References through direct sensorimotor experience with objects and their properties. and D’Esposito M (1998) An area within human ventral cortex sensitive to ‘‘building’’ stimuli: Evidence and implications. Beauchamp MS. J. Cogn. Haxby JV. At present. overlap. Zarahan E. these dementia. any given object. In and human movements. 18: 509–520. Morton J. 2: 913–919. a close physical proximity exists between the Chao LL. One reason for suspecting that these property Neurocase 4: 265–272. Neuroimage 12: 314–325. Learn. regions are not strictly perceptual but. Cogn. Neurosci. the particular neural circuit for a given responses to video and point light displays of moving object concept includes property regions that are humans and manipulable objects. Surveying these studies. As such. Cortex 12: 545–551. Cogn. and motor systems active when that property informa. Weisberg J. Castelli F. as approaches to understanding the cortical organization of demonstrated by training studies that documented semantic processing. taste properties Cave CB and Squire LR (1992) Intact and long-lasting repetition may be particularly important for food concepts). Indeed. conceptual property representation may partially Neurosci. for other property types as well (e. object. Neuroimage 12: port so-called embodied cognition accounts of 478–484. priming in amnesia. Cox DD and Savoy RL (2003) Functional magnetic resonance tion is learned. Aguirre GK. Victor M. Neurosci. object concepts in memory. Bookheimer S (2002) Functional MRI of language: New ence with. rather. 7: 354–361. about any particular object concept is distributed Beauchamp MS. these Capitani E. Rev.07. Taken together. Cogn. Lee KE. Neuron 47: 907–918. support Caramazza A and Mahon BZ (2003) The organization of conceptual knowledge: The evidence from category-specific conceptual representations is that some property semantic deficits.g. are immediately present in the stimulus. and knowledge of color. 259–268. 11: 25–35. Cogn. Happe F. Psychol. these findings strongly sup. Haxby JV. and Butterworth B function-associated motor actions are particularly (2005) Dissociations in numerical abilities revealed by progressive cognitive decline in a patient with semantic likely to be retrieved automatically. and Martin A (1999) Attribute-based neural neural systems underlying perception and knowledge substrates in temporal cortex for perceiving and knowing representation for objects.

R. 26: 6523–6532. 13: 159–166. Cogn. J. and evolutionary perspective on conceptual representation: Chatterjee A (2005) Conceptual representations of action in Species-specific calls activate visual and affective the lateral temporal cortex. 1027–1044. MA: MIT aphasia and semantic dementia. Neuron 35: Cree GS and McRae K (2003) Analyzing the factors underlying 1167–1175. Cambridge. Nature 392: 598–601. McGlone F. Jones SJ. Cereb. Hodges JR and Patterson K (1996) Nonfluent progressive Fodor JA (1975) The Language of Thought. 439–440. Sci. 26: 4917–4921. Cohen NJ. Cambridge. Vision Res. Heider F and Simmel M (1944) An experimental study of Epstein R and Kanwisher N (1998) A cortical representation of apparent behavior. and Matthews PM (2005) Category. and Schill SL (2006) Fractionating the left frontal response to Andersen AH (2006) Dissociation of automatic and strategic tools: Dissociable effects of motor experience and lexical lexical-semantics: Functional magnetic resonance imaging competition. Biol. and Corkin S (1988) The impaired James TW and Gauthier I (2003) Auditory and action semantic learning of semantic knowledge following bilateral medial features activate sensory-specific perceptual brain regions. Cortex 36: 539–559. Cogn. Soc. Forde EME and Humphreys GW (1999) Category-specific Hodges JR. Soc. 1475–1482. Grossman ED. J. evidence for differing roles of multiple frontotemporal Kanwisher N. 361: 2109–2128. Hippocampus 9: 247–254. is task specific. Progressive fluent aphasia with and influential theories. chisel. Goldberg RF. UK: Cambridge University Press. Gorno-Tempini ML. 27: 649–677. louder than functions: The importance of manipulability Annu. 55: 335–346. and Sakata H (1995) the nature of the cognitive defect underlying category. Philos. and Patterson K (2001) Large. Balota DA. Acad. of faces and objects in ventral temporal cortex. Int. Brain 115: 1783–1806. Exp. Am. Kringelbach ML. Gabrieli JD. 27: 226–231. Rolls ET. Ishai A. Cogn. USA 104: 1726–1731. 18: 314–320. J. the local visual environment. and Chun MM (1997) The fusiform regions. Francis S. and Funnell E (1992) recognition impairments: A review of important case studies Semantic dementia. Lond. J. Dronkers NF. J. Ann. J. Neuropsychologia 43: 69–74. 109: 457–469. Psychol. Grill-Spector K (2003) The neural basis of object perception. Sci. Piazza M. Common and dissociable activation patterns associated with Kanwisher N and Yovel G (2006) The fusiform face area: A controlled semantic and phonological processing: Evidence cortical region specialized for the perception of faces. 13: 1792–1796. Rushworth MFS. Perfetti CA. Johnsrude I. et al. Brain Res. Psychol. (1999) The representation of Hodges JR and Graham KS (2001) Episodic memory: Insights pleasant touch in the brain and its relationship with taste and from semantic dementia. Hart J and Gordon B (1990) Delineation of single-word semantic de Araujo IET. and Thompson- Gold BT. Braun A. B Biol. Neuron 41: 301–307. Neurosci. Aphasiology 13: 169–193.128 Structural Basis of Semantic Memory distributed patterns of fMRI activity in human visual cortex. Van Scoyoc A. colourful Cognition and anatomy in three variants of primary or noisy? Attribute. Brendt RS. Trans. J. Arbib MA. Nature 316: channels account of fine-grain category specificity. Kellenbach ML. Curr. Proc. 41: Kintsch W. B Biol. et al. J. Cogn. the structure and computation of the meaning of chipmunk. Wilson A. Trans. Press. and Caramazza A (1985) Category-specific of fruit and vegetable knowledge: A multiple processing naming deficit following cerebral infarction. and Kim CY (2004) Learning to see cherry. correlation. representation of the pleasantness of flavor. Neuropsychol. Lopes M. Neurosci. Kan IP. Cogn. Bowtell R. Neurosci. J. Opin. and the experiences. and Cohen L (2003) Three premotor cortex. Kringelbach ML. 16: 1669–1679. parietal circuits for number processing. temporal lobe atrophy. and Schneider W (2006) Perceptual Kapur N and Brooks DJ (1999) Temporally-specific retrograde knowledge retrieval activates sensory brain regions. Thompson-Schill SL. Cortex 15: 1438–1450. Cogn. representation of action words in human motor and Dehaene S. Gobbini MI. Neurosci. Rolls ET. Crutch SJ and Warrington EK (2003) The selective impairment Hart J. and cello (and many other such biological motion: Brain activity parallels behavior. Vann SD. Kumaran D. Sci. 90: 313–319. Grasping objects: The cortical mechanisms of visuomotor specific disorders: A review. Rizzolatti G. Haxby JV. Balota DA. Gainotti G (2000) What the locus of brain lesion tells us about Jeannerod M. Smith CD. Lond. 1855–1870. Pinel P. 18: 267–277. 356: 1423–1434. 20: 487–506. Chatterjee A. and action in tool representation. Neurosci. and Neuropsychol. 15: Grossman ED and Blake R (2001) Brain activity evoked by 20–46. with anatomical Representation of umami taste in the human brain. Neurobiol. Hassabis D. Neurol. J. Kirchhoff BA. Eur. Curr. USA 101: 17516–17521. Behav. Gen. 22: declarative knowledge in amnesia. Philos. Soc. Science 293: related activation for written words in the posterior fusiform 2425–2430. Neurosci. Affect. Neuropsychol. 1: 207–221. (2004) Toward an Kable JW. Trends Neurosci.and modality-specific activations during progressive aphasia. Furey ML. amnesia in two cases of discrete hippocampal pathology. Schouten JL. Neurosci. visual perception of biological motion. 17: processing systems in the macaque. Ann. retrieval of perceptual attribute knowledge. concrete nouns). Natl. olfactory areas. Blake R. Neurosci. transformation. . Rankin KP. in the human Hauk O. Neurosci. 57: 243–249. Brain Cogn. Behav. Patterson K. 17: 4302–4311. 7: 157–177. et al. Kable JW. Kan IP. and Buckner RL (2005) for face perception. cheese. Natl. Proc. and Patterson K (2003) Actions speak Grill-Spector K and Malach R (2004) The human visual cortex. Acad. Neurol. Rev. Neurophysiol. R. 20: 355–372. (1998) Comprehension: A Paradigm for Cognition. Brett M. face area: A module in human extrastriate cortex specialized Gold BT. Gil-da-Costa R. Sci. from fMRI adaptation. inverted and imagined biological motion. and Maguire EA (2007) de Araujo IET. and Patients with hippocampal amnesia cannot imagine new Phillips N (2003b) Taste-olfactory convergence. Rolls ET. Cogn. Neurosci. (2004) Kellenbach ML. J. Powell DK. 132: 153–201. Neuroreport 10: 453–459. and Pulvermuller F (2004) Somatotopic brain. and Hobden P (2003a) comprehension deficits in aphasia. Grossman ED and Blake R (2002) Brain areas active during Neuroimage 19: 261–270. Oxbury S. temporal-lobe resection. McDermott J. Brett M. 18: 2059–2068. Pietrini P (2001) Distributed and overlapping representations Devlin JT. J. Neurosci. 2: 511–524. Creem-Regehr SH and Lee JN (2005) Neural representations of Hamann SB and Squire LR (1995) On the acquisition of new graspable objects: Are tools special? Cogn.

and Devlin JT (2003) How is Cambridge. Lond. Polyn SM. 264–336. categorical differences in the human brain areas commonly Manns JR. 153–186. Penny WD. Pylyshyn Z. et al.) Handbook of Functional temporal polysensory (STPa) neurons to ‘‘biological motion’’ NeuroImaging of Cognition. (2001b) Representation of pleasant and aversive taste in the Natl. fusiform face area in social cognition: Implications for the Nielsen JM (1958) Memory and Amnesia. Hillsdale. Schacter DL and Addis DR (2007) Constructive memory: The pp. Mervis CB. CA: San pathobiology of autism. and Price C RSJ. Novartis Found. Mishkin M. and places activates corresponding stimulus-specific brain Simmons WK. Neurosci. and Gadian DG (1998) Amnesia Reddy L and Kanwisher N (2006) Coding of visual objects in and the organization of the hippocampal system. Processes in Reading: Visual Word Recognition. in the human orbitofrontal cortex. Cogn. Lucas Press. Curr. Martin A (2001) Functional neuroimaging of semantic memory. Peelen MV. and hand movements. J.. Sartori G. M. Bayley PJ. and Price CJ (2006) Semantic mouth movements. Structural Basis of Semantic Memory 129 Levy DA. Psychol. and Squire LR (2003) Semantic used for naming visually presented objects.) Basic 8: 382–339. NJ: Erlbaum. Cortex 15: 1602–1608. Science 270: 102–105. et al. J. (2003) The role of the cerebral localization. Annu. Grelotti DJ. New York: Psychology body in the fusiform gyrus. theories. MA: MIT Press. 2nd ed. Philos. Sci. 20: 561–574. 47: 36–45. the fusiform gyrus related to category-specificity? Cogn. Nakai T. Gray WD. USA 101: 6710–6715. pp. Phillips J. 167–186. et al. Frackowiak Rogers TT. 2: 433–450. Aphasia: Their value in Schultz RT. Hopkins RO. MA: MIT Press. Neurophysiol. Proc. Gore JC. 27–48. Martin A. (1984) Computation and Cognition. Ann. lateral temporal lobe. Oram MW and Perrett DI (1994) Responses of anterior superior In: Cabeza R and Kingstone A (eds. and other patients with responses. Sci. and McCarthy Martin A and Fedio P (1983) Word production and G (2005) Functional anatomy of biological motion perception comprehension in Alzheimer’s disease: The breakdown of in posterior temporal cortex: An fMRI study of eye. Psychol. 296: 33–36. Ashburner J. J. 207: 35–66. pp. O’Craven KM and Kanwisher N (2000) Mental imagery of faces Hippocampus 12: 520–533. 358: 415–427.) Cognition and Categorization. (2000) Naming of animals and about the organization and representation of conceptual tools: A functional magnetic resonance imaging study of knowledge. Cortex 16: 437–445. MA: stimuli. Johnson DM. 2: 1019–1025. 11: that respond to biological motion. Neurosci. Ungerleider LG. Cogn. Cogn. (2005) object recognition in cortex. B Biol. and Squire LR (2004) The anatomy of O’Doherty J. category-specific deficits? Brain 125: 2067–2080. J. and Barsalou W (in press) A common neural . Beauchamp MS. MA: Lloyd BB (eds. lobe atrophy and semantic memory. and Friston KJ (2006) Two Schmolck H. Neurophysiol. mouth semantic knowledge. Ramjee V. In: Gazzaniga MS (ed. Klin A. 17: 434–445. Brain Lang. Mahon BZ and Caramazza A (2003) Constraining questions Okada T. Allison T. Morris JP. Noppeney U. Neurosci. Bowtell R. Gilboa A. 1023–1036. Patterson K. Cereb. NJ: Erlbaum. premotor cortex (area F5). Neurobiol. Rolls ET. McRae K. Martin A. 58: 25–45. Wiggs CL. Pelphrey KA.. bilateral medial and lateral temporal lobe lesions. Cohen JD. Neurol. Neurobiol. Neurosci. 12: 1013–1023. In: Baddeley AD. Orlandi P. Cortex 15: 1866–1876. and Barsalou LW (2005) Pictures of regions. Noppeney U. and Haxby JV (2000) Category. and Watts FN (eds. J. Rolls ET. Los Angeles. and Hodges JR (2000) A voxel-based morphometry (2005) Fusiform activation to animals is driven by the study of semantic dementia: Relationship between temporal process.) brain. Lalonde FM. Nielsen JM (1946) Agnosia. New York: Martin A and Caramazza A (eds.) memory search. In: Rosch E and Murphy GL (2002) The Big Book of Concepts. Apraxia. 6: 99–116.. 85: 1315–1321. Symp. Vargha-Khadem F. Hillsdale. Neuroimage 30: 992–1002. Trans. and Norman KA (2005) specificity and the brain: The sensory-motor model of Category-specific cortical activity precedes retrieval during semantic representations of objects. Conceptual Knowledge in the Brain: Neuropsychological Peelen MV and Downing PE (2005) Selectivity for the human and Neuroimaging Perspectives. and Squire LR (2002) distinct neural mechanisms for category-selective Semantic knowledge in patient H. J. and Price CJ (2002) (1995) Discrete cortical regions associated with knowledge Can segregation within the semantic system account for of color and knowledge of action. Neuron 49: 815–822. Mechelli A. Curr. Kringelbach ML. 18: 2188–2199. Michelich CR. understanding social and mechanical concepts. Natu VS. Puce A. Anat. Patterson K and Hodges JR (1995) Disorders of semantic Martin A (2007) The representation of object concepts in the memory. Opin. Allison T. Neuron 38: 127–133. memory and the human hippocampus. Wilson BA. Neuropsychol. and Ungerleider LG Phillips JA. J. Cogn. Press. and Boyes-Braem recognition: A selective review of current findings and P (1976) Basic objects in natural categories. Wiggett AJ. Humphreys GW. Patterson K. and McGlone F semantic knowledge: Medial vs. MIT Press. Neurosci. Hocking J. and Downing PE (2006) Patterns of Martin A and Chao LL (2001) Semantic memory and the brain: fMRI activity dissociate overlapping functional brain areas Structure and processes. Rev. Cogn. Handbook of Memory Disorders. 19: 124–141. Mummery CJ. Cereb. Neely JH (1991) Semantic priming effects in visual word Rosch E. Rosch E (1978) Principles of categorization. human brain. Cereb. Francis S. Nature 445: 7123–7127. Cognit. Price CJ. Martin A. Price CJ. Hornak J. Martin A and Weisberg J (2003) Neural foundations for Neuropsychol. New York: Hafner Publishing Company. Functional neuroanatomy of remote episodic. ghosts of past and future. 194–201. Neurosci. Noppeny U. pp. Cambridge. and McCarthy G (1998) Neuropsychol. relevance explains category effects in medial fusiform gyri. Riesenhuber M and Poggio T (1999) Hierarchical models of Moscovitch M. Nat. R. Haxby JV. The New Cognitive Neurosciences. Rosenbaum RS. Lett. Soc. In: Besner D and Humphreys GW (eds. and Andrews reward. the ventral visual stream. Bentin S. Price CJ. Cambridge. Cambridge. Opin. C (2001a) Abstract reward and punishment representations Simmons WK. Kensinger EA. Neurosci. Science 310: 1963–1966. Tanaka S. Acad.) (2003) The Organization of John Wiley and Sons. 93: 603–608. Hippocampus 8: 212–216. semantic and Rizzolatti G and Fadiga L (1998) Grasping objects and grasping spatial memory: A unified account based on multiple trace action meanings: The dual role of monkey rostroventral theory. pp. Corkin S. Nat. 16: 408–414. Temporal cortex activation in humans viewing eye and Mechelli A. MIT Press. 218: 81–95. 20: 575–587. 4: 95–102. Martin A. not the stimulus. appetizing foods activate gustatory cortices for taste and O’Doherty J.

Weisberg J. London: Academic Press. and Martin A (1999) Neural correlates of Tulving E (1972) Episodic and semantic memory. Tranel D. Vandenbulcke M. Proc. Galton CJ. Wiggs CL. (2006) Knowledge of visual attributes in the right hemisphere. In: Ingle DJ. Patterson K. Brain 107: 829–854. 17: 367–376. Cogn. Goodale MA. location. von Kriegstein K. Damasio H. 40: 174–186. Verfaellie M. Neurosci. Williams G. and Martin A (2005) (2005b) Effects of noun-verb homonymy on the neural Automatic priming of semantically related words reduces correlates of naming concrete entities and actions. Kapur N. Graham KS (2002) Recollection-based memory in MA: MIT Press. D’Esposito M. (1993) Area V5 of and Knight RT (1998) Verb generation in patients with focal the human brain: evidence from a combined study using frontal lesions: A neuropsychological test of neuroimaging positron emission tomography and magnetic resonance findings.) Analysis of Visual Behavior. Swick D. Frackowiak RSJ. impairments. 2: 167–182. Goulding PJ. Connelly A. Behav. Myers R. Brain 125: 2523–2536. USA 95: 15855–15860. Martin C. frontotemporal dementia: Implications for theories of long. 5: 169–177. 381–403. Tranel D. and Hichwa R Wheatley T. Koseff P. Hodges JR. Psychol. Curr. Cortex 3: 79–94. knowledge – Further fractionations and an attempted Thompson-Schill SL (2003) Neuroimaging studies of semantic integration. Warrington EK (1975) The selective impairment of semantic and Hodges JR (2004) Dissociating person-specific from memory. Peeters R. general semantic knowledge: Roles of the left and right Warrington EK and McCarthy RA (1987) Categories of temporal lobes. Lang. and Alexander MP (2000) Acquisition of Squire LR and Alvarez P (1995) Retrograde amnesia and novel semantic information in amnesia: Effects of lesion memory consolidation: A neurobiological perspective. Nestor PJ. . Grabowski TJ. semantic (2005) Interaction of face and voice areas during speaker memory. and Damasio H (2005a) Naming Whatmough C. systems. Q. In: Tulving E semantic and episodic memory retrieval. Science 277: 376–380. 27: 635–657. and Hodges JR (2005) Neural Tulving E (2002) Episodic memory: From mind to brain. Neurosci. Graham KS. J. correlates of semantic and behavioural deficits in Rev.’’ Neuropsychologia Warrington EK and Shallice T (1984) Category specific semantic 41: 280–292. Vargha-Khadem F. Cambridge. 9: 964–970. Cereb. Paesschen W. Neurpsychologia 38: 484–492. Neuropsychologia and Donaldson W (eds. Lyon J. and (eds. Kleinschmidt A. Exp. Miller BL. Sterzer P. Van Neurol. and Mishkin M (1997) Differential effects of Spiridon M and Kanwisher N (2002) How distributed is visual early hippocampal pathology on episodic and semantic category information in human occipito-temporal cortex? An memory. fMRI study. 17: 1293–1305. J. pp. Farah MJ. frontotemporal dementia. Watkins KE. et al. van Turrennout M. Cereb. Neuropsychologia Cogn. Psychol. object structure during picture naming. Murtha S. 513–521. Hippocampus 8: 205–211. Acad. Chertkow H. Snowden JS. Sci. 17: 1871–1885. memory: Inferring ‘‘how’’ from ‘‘where. Cortex 17: 1319–1327. imaging. Neurosci. Weisberg J. 53: 1–25. and Neary D (1989) Semantic Nat. and Mansfield RJW Simons JS. Tranel D. Neuropsychologia 42: 359–370. and amnesia. Beauchamp MS. dementia: A form of circumscribed cerebral atrophy.) Organization of Memory. Watson JDG. and Giraud AL Squire LR and Zola SM (1998) Episodic memory. Neuroimage 24: 1042–1051. J. and Hanratty K (2002) the same entities from visual or from auditory stimulation Dissociable brain regions process object meaning and engages similar regions of left inferotemporal cortices. 92: 288–299. and Vandenberghe R term memory. J. Ungerleider LG and Mishkin M (1982) Two cortical visual Neuropsychologia. Fannes K. pp. Neuron 35: 1157–1165. 37: 103–118. Verfaellie M. Opin. Thompson SA. and Damasio AR (1997) A neural basis for Weisberg J. Kan IP. Natl. Annu. Neurobiol. Williams GB. Brain activity in the fusiform gyrus. Neuropsychologia 35: system for learning about object function. Brain 110: 1273–1296. Damasio H. recognition. Thompson-Schill SL.130 Structural Basis of Semantic Memory substrate for perceiving and knowing about color. Cogn. 549–586. Grabowski TJ. Neurosci. Gadian DG. and Martin A (2007) A neural the retrieval of conceptual knowledge.

The recollection process is proposed to involve belongs to. (or knowing). This ability to familiarity process was followed by a secondary accurately recognize a person or an item requires that slower process necessary for retrieving specific an individual encode the person or stimulus into detailed information. 2001).08. In the preceding example. He termed these compo- person or object: ‘‘That is my dog’s vet’’ or ‘‘That is nents recollection (or remembering) and familiarity my coat. 1972. we cannot recollect their name or sented stimuli and stimuli that were not previously recall where you last met this person. over the last 50 years. Within the framework of recognition memory. 3. person is familiar. but I don’t know who it tered. Boston.08. An everyday example cesses as being important for recognition (Atkinson is our ability to recognize a person we have met once and Juola.’’ the retrieval of the prior episodic experience. USA ª 2008 Elsevier Ltd. you subject must be able to then subsequently discrimi. 1974). 2001). in rately identify a particular item or stimulus as having the 1970s. ‘‘That person can recognize that you simply know the individual. In the example in signal is then followed by a slower secondary search which a subject recognizes a person as someone process in which you are able to retrieve detailed whom they have met before. 3. In the recognition example memory at the time of encoding. person’s name and their relationship to us. In keeping with the example given iar.5 Functional Imaging Studies of Recognition Memory 136 3. After a delay.08 The Neurobiological Basis of Recognition Memory C.08. E. Recognition memory refers to the capacity to accu. Boston University.3 Lesion Studies of Recognition Memory 134 3. presented (termed lures). Tulving. All rights reserved. Endel Tulving suggested that long-term we encounter novel and familiar stimuli all the time. but in research the last time you saw that individual. We can all recall episodes in which this re- be able to accurately identify the person or select the trieval attempt is successful. and then at a later time would need to them. E. when you meet someone in the street. may initially recognize that person as being someone nate between a novel or new stimulus and a who is familiar to you. memory was described as being a you remember the contextual details associated with solitary process (Tulving. but I can’t think of their name.2 Experimental Tests of Recognition Memory 132 3. you or object was encountered before. In other vidual had not previously met. This initial fast familiarity previously encountered stimulus. Eichenbaum and Cohen. recognition memory could be could be divided into Sometimes we are able to recognize with certainty a two separate components. and where you last saw an earlier time.6 Computational Modeling of Recognition Memory 138 References 139 3.’’ or or you can clearly remember the individual encoun- ‘‘I have seen that hat before. Stern and M. Until the 1960s.’’ Other times the face or object looks famil.08. MA. They proposed that an initial fast as someone that we have seen before.1 Introduction to the Concept of moved toward describing memory as being com- Recognition Memory prised of multiple memory systems (Tulving. and we remember the person from a group of other subjects that the indi.1 Introduction to the Concept of Recognition Memory 131 3. recognition is often tested by we still have our initial reaction of knowing that the having subjects distinguish between previously pre.08. memory researchers have familiarity is thought to represent a process that 131 .08. Atkinson and Juola described two pro- been previously encountered. In research on cases.4 Electrophysiological Studies of Recognition Memory 135 3. 2001. and while memory function. need to have accurately encoded the person’s face at their relationship to you. that is. our retrieval attempt is unsuccessful. the subject would first information about who that person is. In contrast.08. In 1983. the given above. looks familiar. Hasselmo. their name. but one is less certain about where the person above. if you pass someone on the street.

A person may there. of Recognition Memory ble multiple-choice exams. stimuli are during the test period. The subject must generate a presented during the sample period. and measurements of stimulus. cing less demand on memory retrieval systems by simply requiring that the subject distinguish pre- viously learned information from new information. First. In another com- nize the correct one. as they will stimuli. This recognition paradigm is known are provided with multiple choices. where the subject must retrieve the learned In one classic recognition memory task. the experi- during aging and in diseases such as Alzheimer’s menter then presents the subject with a list of disease. it proves very important the two processes that combine to support recognition . whereas recognition tests can role for medial temporal lobe structures in the encod- assess whether information was encoded while pla. another useful distinction that followed by sections describing lesion studies of memory researchers make is the distinction between recognition memory in humans and animals. However. after which recog- involve the retrieval or recall of a previously pre. nition is tested by presenting both a previously viewed sented stimulus without the stimulus being presented sample stimulus and a novel lure at the same time during the test period. an over- In addition to dissociating the processes of recol. Both recogni. free recall are presented during a sample period. Usually answers are pro. often respond no to all stimuli. menter first presents subjects with a series of stimuli to erate the answer. (lures). the humans. the experi- information from long-term memory in order to gen. prior event. This method is known as a vided verbally. The method is not known as a paired associates test. and computational studies of recognition. animals is described in greater detail. Both free recall and cued recall delay period of different durations. The simple provides a cue. such as be learned. response bias and can more sensitively assess recogni- and the subject must generate the previously paired tion memory impairments. as a simple yes–no recognition test. In a free-recall task. and retrieved from memory. but if they each stimulus.’’ that assists the subject in recalling the with patients with memory disorders. where the answer is present but must be identified from false lures. ing of stimuli for subsequent recognition. recall performance declines earlier and faster previously presented stimuli mixed with new ones than recognition performance.2 Experimental Tests One analogy is that recognition memory tests resem. during the test choice recognition memory test is less sensitive to period only a single stimulus in a pair is presented. stimulus pairs are often used because of this response bias. imaging studies of recognition memory in stored. not previously presented. electro- the processes of recognition and recall.08. lection and familiarity. process of recall places a much greater demand on The human and animal studies indicate an important the retrieval system. followed by a and cued recall. When memory declines. In a cued recall test. such as ‘‘name the animals that were yes–no recognition test cannot be used effectively on the list. they can recog. Recognition memory has been studied extensively in whereas a recall test resembles a fill-in-the-blank human subjects using a number of different paradigms. view of recognition memory tests is provided. stimuli in humans. you recognize the person as In the following sections. that is. The forced presented during the sample period. The subject simply responds yes or no to fore be unable to recall someone’s name. After a variable delay period. a single stimulus or set of stimuli Recall can be divided into two types. work on the neu- familiar but cannot place where or when you met roscience of recognition memory in humans and in them. exam. In one specific type of cued recall task. the experimenter forced-choice recognition memory test.132 The Neurobiological Basis of Recognition Memory does not rely on any explicit remembering of the in neuroscience studies of memory function in details of the spatial or temporal context of the animals. Because recognition can be tested More recent studies have elaborated on the simple without requiring the subject to generate the pre. 3. There are a number performed for stimuli that can be generated by the of variations of these two basic behavioral designs that subject and therefore usually require the use of verbal have been created and are described here. and during the response indicating which stimuli (target stimuli) were test period the subject is requested to generate as previously presented and which stimuli (lures) were many stimuli as possible. Free recall and cued recall can only be response bias can be generated. physiological studies of recognition memory in tion and recall require that stimuli be encoded. animals. mon recognition test. yes or no response to more accurately characterize viously presented stimulus.

tasks requires recognition memory. and recollection is an all or include a series of visual stimuli presented to mon- nothing response. This work The DMS or DNMS task designs have been used uses signal detection theory-based receiver-operator. the sample stimulus of recollection-based recognition. recollection. in which the subject must respond on the new (for the lures). The Neurobiological Basis of Recognition Memory 133 memory. but the process In contrast. There have been studies that keys. By modifying the task. In experimental tests of these processes. low-confidence yes (familiarity). or a set of odor stimuli presented to rats. the animal encounters the sample stimulus and. Some pairs of stimuli. of recognition memory. however. Others have used 5. structures support familiarity. Alternatively. delay period. They may also be asked to retrieve single sample stimulus followed by a delay period. the experimenter can suggest that recollection and familiarity are part of distinguish between recollective and familiarity a unified single process. variant of this task known as a delayed non-match- fidence in remembering an item or event. the recollection process is assumed either short-term active maintenance of the stimulus to provide sufficient confidence for subjects to (working memory) or long-term synaptic encoding of respond with certainty in the highest yes category the stimulus. the studies remain controversial. The DMS task involves presentation of a lus (familiarity). The subject must select use an arbitrary scale to provide information about the stimulus that matches the sample stimulus. supported by the same neu- responses. 3-point scale. which could variable phenomenon. Both recollec. Alternatively. the animals are then tested with a port recollection and familiarity-based recognition. DMS task. each of which includes one sample . and familiarity. ory task. in which the animal encounters However. is presented with a pair of stimuli and recognition memory includes a graded mechanism must respond to either the match (target) or non- of familiarity recognition and a high-threshold process match (lure). Performance of these ity but not as much as the previously presented items. which can be extended to longer intertrial confidence no (reject lure) response (numerical value intervals. when recollection is triggered for an of recognition could be interpreted as involving individual item. Studies of behavioral confidence rankings have the N-back task is considered to be a working mem- been used to support a two-process theory of recogni. In the dual-process model. information about contextual information related to followed by a pair of test stimuli. depending on the length of the delay. the subject (human or behavioral and imaging studies have used a simple animal) must select the nonmatching novel stimulus. Since it typically uses very short delays. other data digm. In a the strength of their memory or their level of con. 1). must generate a response. during the sample period versus separate regions a subject may be requested to identify whether mediating recognition during the testing period.or 6-point basis of whether the current stimulus matches the scales ranging from a high-confidence yes (identify stimulus presented n trials previously (an N-back target) response (numerical value 5 or 6) to a high. in another twist on the matches the presentation stimulus and one does not yes–no recognition test. Similar controversy surrounds efforts to make sometimes referred to as the remember versus know distinctions between regions supporting encoding distinction. researchers support the idea that the hippocampus tion and familiarity would generate a yes response in is required for recollection. This work used the dis. whereas parahippocampal a yes–no or forced-choice recognition memory para. task). where one stimulus the stimulus. and of stimuli. forced-choice design. The distinction between a strong feeling roanatomical structures within the medial temporal of recollection and a weaker sense of familiarity is lobe. subjects may be requested to match the sample stimulus. after tribution of proportions of responses to argue that a delay period. Another common recognition memory task is ving the stimulus (recollection) or whether they only known as the delayed match-to-sample task. (allowing the responses to cross a high-threshold Other paradigms used in animals include presen- criterion). familiarity is a tation of a longer list of sample stimuli. or have a general feeling of familiarity about the stimu. they have a distinct recollection of the event of obser. tion memory in work by Yonelinas (2001). most extensively in studies of the neural basis of characteristics (ROC) curves to analyze the nature recognition memory in animals. Some to-sample or DNMS task. corresponding to high-confidence yes A further variation involves continuous presentation (recollection). In these paradigms. After a suggest distinct medial temporal lobe regions sup. until the target stimulus appears and the animal whereas new items (lures) give rise to some familiar. The Yonelinas model could be followed sequentially by single lure stimuli assumes that previously presented items give rise to a (during which the animal withholds its response) range of confidence ratings based on their familiarity.

1992. the rat would return to the back of the results continue to be controversial. 2003). Patient HM showed a striking impairment of recognition tasks have also been shown with inacti- episodic memory encoding following bilateral vation or lesions of the perirhinal cortex in rats (Otto removal of his medial temporal lobes in an attempt and Eichenbaum. lesions (Manns et al. Hippocampal amnesics also for human studies of recollection and familiarity have show moderate impairment on tests of recognition been modified for odor recognition to examine these memory (Manns et al. perirhinal cortex. recognition based recognition associated with medial temporal memory is commonly tested using the non-match-to. individual trial unique stimuli in a range of different sented with a sample stimulus. Alvarez and a novel lure. responses from conservative to liberal. in further work by Milner and Corkin (Corkin. when presented with an identical sample stimulus Zola-Morgan and Squire. the odor (old). it will spend more time exploring et al. in which a rat was exposed to a set and parahippocampal gyrus (Rempel-Clower et al.. 1992).5. 2003). including Murray. Selective hippo- the novel lure.. 1995). other hippocam. odor recognition. On the other hand. role of the perirhinal cortex beyond its interactions pal amnesic patients show strong impairments of free with the hippocampal formation. Research on the neuroanatomi.. but the impairment is much more severe if lesions include damage to surrounding cortical areas such as the 3. following the case of important for recognition memory function that is patient HM.134 The Neurobiological Basis of Recognition Memory stimulus and one lure stimulus. consistent with the notion cal structures critical for memory function focused that entorhinal cortex and perirhinal cortex are on the medial temporal lobe. first described by Scoville and Milner distinct from short-term active maintenance for (Scoville and Milner. long-term memory. campal lesions alone cause impairments in DNMS tasks (Zola-Morgan and Squire.. In primates. These studies show that object recogni- patient HM’s impairment focused on his almost tion is not impaired with lesions of the fornix or the total loss of free recall and severe impairment of orbitofrontal cortex. Alvarez et al. Initial descriptions of et al. response used in human visual recognition memory tude of damage to the parahippocampal structures was altered to a delayed nonmatching paradigm for including the entorhinal cortex. to capitalize on the natural ten. The yes–no impairment appears to be associated with the magni... In rats.. In these studies.. The rat can be pre. Ennaceur et al.. 1994).and familiarity. This was studies by Manns and Squire have demonstrated achieved by changing the relative cost and benefit impairments of both recollection. If the odor did match a previous sample within the medial temporal lobes.3 Lesion Studies of Recognition perirhinal cortex and parahippocampal gyrus Memory (Meunier et al.08.. 1996. lesions of medial temporal lobe struc- dency of rats to explore novel stimuli more tures cause impairments on the recognition of extensively than familiar ones.. of sample odors and then subsequently had to 1996). The tasks designed recall and cued recall. 1957) and studied extensively working memory. Entorhinal and perirhinal cortex ablations There is extensive lesion evidence supporting the idea also impair performance in DMS tasks (Gaffan and that the medial temporal lobe structures. 1994. Researchers experimental chamber to dig in a separate cup for including Yonelinas have argued that hippocampal food reward. 2005). of a positive recognition response by changing the . Similarly.g. 1993. 1985. 2004). and subsequently studies on delayed matching function (Gaffan.or odor- 1984). Impairments of performance in object. 1. 1992.. 2000). the rat could dig in the cup for a and familiarity are subserved by different regions food reward. again suggesting an important cued recall performance. 1 min. and the amount of processes in rats (Fortin et al. sample paradigm. 1993.. 1989. 1974. Bussey to control epileptic seizures. 1993). 1996. However. Leonard et al. 1985). The impairments in these studies the hippocampus and neighboring parahippocampal appear at longer delays (e. 10 min) but not regions. approach used in humans was applied to rats by based recognition than familiarity-based recognition manipulating the selectivity criterion for recognition (Yonelinas et al. 2001. the signal detection lesions cause a greater impairment of recollection. respond to individual test odors presented in cups Lesion studies in humans have also addressed the filled with sand. If the odor did not match a previous question of whether the processes of recollection sample odor (new). Zola-Morgan et al. are critical for encoding information into short delays (0. or 3 s) (Gaffan and Murray.

Scopolamine neural activity for a previously presented target stim- also impairs recognition memory in humans (Sherman ulus in comparison to nonpresented lures. Lesions of the hippocam. during short. with occasional repeat pre- demonstrated that infusion of scopolamine into peri. The Neurobiological Basis of Recognition Memory 135 amount of reward given for an old response and by stimuli was also impaired by selective lesions of ento- increasing the height of the cup containing the test rhinal cholinergic innervation in monkeys (Turchi odor. Increased responses to matching stimuli were . Miller demon- vation of cortical structures. 1993. unit recording nists has been shown to impair performance on studies analyze the generation of action potentials by recognition memory tasks at longer delays in individual neurons during performance of such tasks. Fahy effect has been tested with localized injection of cho.08. 1993. 2003. and not for distractor stimuli (Miller et al. Pharmacological manipulations can also affect most studies of electrophysiological activity asso- recognition memory function. rats show a threshold response et al. Recent experiments used strated that repetition suppression occurs even for the selective agent IgG 192 saporin to lesion the repeated distractor stimuli presented during a delay cholinergic innervation of the entorhinal cortex in period. formance.. 2005). 1986. as shown by recordings of single- choice behavior in an eight-arm radial maze in rats neuron spiking activity during performance of a when there is a delay between the individual choices serial recognition task by monkeys (Brown et al. Schon et al... Riches et al. Systemic injections of scopolamine also impair arm not more activity. In both monkeys and rats.or long-term recognition memory Systemic injections of muscarinic cholinergic antago. whereas infusion neurons recorded from parahippocampal cortices into adjacent structures did not cause impaired per. In this paradigm. 1991...... In those experiments. 1976) and atropine (Penetar presented target stimulus during the test period and McDonough. 1996. Similarly. Wilson et al. DMS function for novel visual 1996). and 1996). entorhinal cortex of monkeys during repeated pre- uli for subsequent recognition have also been induced sentation of stimuli in a DNMS task (Miller et al. the prefrontal cortex.. Injections of scopolamine with the response to a nonpresented lure or by com- appear to impair the encoding of new stimuli for paring the response to the target stimulus with the recognition but not retrieval of stimuli learned before response to the same stimulus during the sample scopolamine injection (Aigner and Mishkin. Tang et al. The locus of the cholinergic 1987. Microdialysis shows a 41% increase in 1991). most studies of memory function in animals the response properties to resemble those proposed use tasks that test recognition memory.4 Electrophysiological Studies based memory in humans. 1993). period... 2003. (Bolhuis et al. whereas enhancement of the target test rats (McGaughy et al. 1990. causing As noted... A number of studies have shown changes in Aigner et al. 2005). feasible to have animals perform recall tasks. of Recognition Memory pal formation in rats appear to selectively remove this recollection-based response property.. 2005) but has a stronger effect studies suggest that recognition memory for an item on free recall and cued recall (Atri et al. Wilson et al. These selective lesions stimulus relative to the sample stimulus (match caused a significant and selective impairment of enhancement) only occurs for the target stimulus DNMS function for novel odors (McGaughy et al. Riches et al. 1997). demonstrating a signifi- cant number of hits even when the false alarms are near zero. 1990. Thus. 1987. (Brown et al. Tang et al... while sparing performance at 0-s delay. The activity associated with recognition can be eval- this occurs with both the antagonists scopolamine uated by comparing the response to a previously (Bartus and Johnson. for a conservative criterion. 3. The familiar stimuli rhinal cortex before encoding impairs performance cause a reduced level of spiking activity in single on a recognition memory task. sentations of familiar stimuli. indicating a role for ciated with memory function have been conducted cellular neuromodulatory effects in this process. Suzuki et al. (1997) series of novel stimuli. 1988). inferotemporal cortex.. 1991. 1983). involves a reduction of neurophysiological activity. 1997). Many et al. et al.. monkeys. with manipulations that remove the cholinergic inner. 2004). This resembles the threshold response proposed to result from the use of recollection.. paradigms. A similar reduction of neuronal spiking activ- acetylcholine levels in perirhinal cortex during encod. monkeys see a linergic antagonists in monkeys. Selective impairments of the encoding of stim. 2005). since it is not to indicate use of familiarity-based recognition alone. 1991. ity to familiar stimuli was observed in recordings in ing in this visual recognition task (Tang and Aigner.

match enhancement or match verbal stimuli learned prior to scanning. 1982).. However. 1995. 2003). Persistent activity has also been positron emission tomography (PET) primarily iden- shown for entorhinal neurons in the rat during the tified changes in activity in cortical structures outside delay period of a continuous DNMS task using odors of the medial temporal lobe. (Miller and Desimone. 1994) of Recognition Memory and in parahippocampal regions including the ento- rhinal cortex (Suzuki et al. Neurons in the entorhinal cortex prefrontal cortex (for review. early imaging studies using (Miller et al. 2000.08. network simulations to examining the recognition period.. 1997). 1993). 1996. trasting blocks of stimuli that were novel with ing activity during delay periods occurs due to blocks of repeating stimuli provided clear evidence attractor dynamics caused by excitatory recurrent that the hippocampus and neighboring para- synapses (Amit. Fuster... 1998). Lisman et al. Bogacz and Brown. and this effect has been proposed to nance of information during the delay period. but the longer-term match sup. In arise from cholinergic activation of long-term support of active maintenance. 2003). involvement in either persistent spiking for active pression effects have been modeled as dependent maintenance or synaptic depression for repetition upon synaptic long-term depression (Sohal and suppression. though it did cause a behavioral impairment the delay period until the test stimulus is presented. 1992. 1997)..e. on the prefrontal cortex (Fuster. In the rat. because scopol- recording studies during DMS tasks (Fuster. 2000) have depression in that study. These mechanisms may coexist in single Hasselmo. The tasks used in these earlier studies associated with match or mismatch between sample focused on priming and on recognition memory for and test (i. to the increase in cerebral blood flow measures seen 1998). 2000). it became possible to examine fMRI activity .. gests that cholinergic modulation may also play a tenance of the stimulus for performance of the role in persistent activity or match enhancement recognition task. 1999).. Much of this research has focused mechanisms important for recognition memory.. 1996). 1982. 1995. a number of unit depression (Warburton et al. most models of persistent spiking activity later. the perirhinal The process of recognition could be supported cortex shows reduced levels of activity in response to either by synaptic modification or by active mainte. items with the goal of being able to identify them However. Stern and demonstrate how the patterns of unit firing during colleagues (Stern et al. scopolamine did demonstrated that some neurons activated by the not block repetition suppression in a study in mon- sample stimulus will maintain their activity during keys. most notably within the (Young et al. Using func- suppression). but delay activity has also been shown for neurons in the inferotemporal cortex 3. see Schacter and of the rat also show increases or decreases of activity Wagner. 1997. Miller and Desimone. hippocampal regions were necessary for encoding Amit and Brunel. Computational tasks in human subjects. 1988. neurons of perirhinal cortex. Miller et al. For example. whereas the medial temporal delay hippocampus and medial temporal lobes for encod- activity is more likely to be terminated by a distractor ing and recognition.. 1973. Lansner and Fransen. With the devel- type of persistent spiking activity could contribute opment of event-related fMRI methods (Rosen et al. modeling indicates that the delay activity and match The role of cholinergic receptors in recognition enhancement over short periods could be due to memory function described above could be due to intrinsic currents. This latter result sug- This delay period activity could reflect active main. Hasselmo and Stern. 1973. This simple block-designed paradigm con- have focused on the hypothesis that persistent spik. The prefrontal Despite considerable evidence from human and ani- cortex activity shows greater resistance to the effect mal lesion studies regarding the importance of the of distractors. 2002. 2006). This complex visual stimuli into memory.5 Functional Imaging Studies (Fuster and Jervey.. amine blocked the induction of synaptic long-term Fuster and Jervey. as opposed population..136 The Neurobiological Basis of Recognition Memory shown in the anterior thalamus in studies of a serial during the delay period of DMS tasks or two-back recognition task (Rolls et al. repeated stimuli. Fuster. 1996) identified robust acti- DMS and DNMS tasks could arise from cholinergic vation of the hippocampus and parahippocampal activation of intrinsic persistent spiking mechanisms structures when subjects viewed sets of novel visual (Fransén et al. 1982. The persistent activity could arise tional magnetic resonance imaging (fMRI) and a from intrinsic mechanisms within individual neurons task design that focused on examining functional or from excitatory feedback between neurons in a changes during the encoding period.

In neuroimaging Henson et al. cellular level. memory tasks typically cause extensive activity in atomical differences between the hippocampus and prefrontal cortex and parietal cortex. Current research is ongoing demonstrated a subsequent memory effect. When parahippocampal gyrus is sufficient and necessary for familiar stimuli are used in a delayed matching task. This study demonstrated a cor- um as opposed to the dual-process model (Wixted relation between reduced activity (as measured using and Stretch. answer this question. research. including areas dividual stimuli during a subsequent recognition within the ventrolateral prefrontal cortex and parie- memory task. contextual details. Two of the earliest event-related tal lobe (Schacter et al. whereas activity in the medial temporal lobe during a DMS hippocampal activity is necessary for successful task with novel stimuli correlated with subsequent source memory.. et al. 2001). It should be noted that at a address the hypothesis that recollection and familiar... Gabrieli In addition to the long-term recognition studies et al. Kirchhoff et al. Ranganath it does not cause significant medial temporal lobe et al. only a short time previously. 2005). (2004) demonstrated that delay-related ciated with successful item recognition. Stern et al. The Neurobiological Basis of Recognition Memory 137 during individual trials as opposed to blocks of stim. Proponents of this account and D’Esposito. memory function (Miller and Desimone. 2003. (Stern et al. whereas encoding related activity within the for medial temporal lobe structures in active main- hippocampus and posterior parahippocampal cortex tenance and encoding of novel stimuli (Ranganath predicted recollection.. Researchers designed studies to focus on how the lobe regions. 1997). the issue remains controversial. another area of research has exam- ined subsequent memory effects for individual words ined the recognition of stimuli that were observed and pictures (Brewer et al. mation from memory or is attributable to control bered with high confidence.. (2003) conducted an event-related fMRI study activity. supporting the idea that the by Gonsalves et al.. (2005) provides crucial data to strength of recognition varies along a single continu. 1997. functional imaging studies have working memory N-back tasks as well as DMS extended these subsequent memory studies to tasks with short delays.. described above. 1998. whereas the presented two or three stimuli previously. Schon et al. 1994. 2000. 1997. suggest that the parahippocampal regions are asso. outside the medial temporal lobe. Since the first et al.. as there these previous studies were not able to answer the is also research supporting the idea that the hippo.. 2001. 2001). the N-back task and other working (see Brown and Aggleton. these tasks most likely require active ity are dependent on different regions within the maintenance rather than modification of synapses. 1998. but when novel stimuli are used. However. which increases parahippocampal regions. these studies have primarily employed More recently. 2001. 2004)... Recently. question of whether the reduction in spiking activity campus and parahippocampal regions support both correlated with the parametric graded strength of recollection and familiarity (Manns et al. which includes information about recognition memory for the stimuli. This could reflect a greater role tion. neuroimaging researchers when the memory load is increased by increasing the have examined the hypothesis that the hippocampus value of N to require matching with a stimulus is critical for the process of recollection. recognition memory performance. 2003.. 1998) Yonelinas et al.. for review) and an. 2006). Wagner et al. It should be noted that while Primate neurophysiological studies have suggested there are several studies supporting the idea that that repetition suppression is involved in recognition separate neuroanatomical regions support recollec. 1999. At medial temporal cortex. 2004). Wagner et al. Brown and Aggleton. In addition to activity within medial temporal uli.. these event-related studies have exam. but not for stimuli mechanisms driving recollection attempts (Kahn that were subsequently forgotten. 2001).. A recent article Squire et al. Kirchhoff et al.. 2000). Based on animal studies a systems level. supporting familiarity-based recognition... 1998. 1996.. Daselaar et al. it causes a in which activity at the time of encoding within the significant increase in medial temporal lobe activity rhinal cortex supported familiarity-based recogni. block-designed studies (Stern et al. studies (Brewer et al. fMRI and MEG) and higher confidence recognition . 2004). in which to determine whether this prefrontal and parietal increases in fMRI activity in medial temporal lobe activity is related to the successful retrieval of infor- structures were noted for stimuli that were remem. functional neuroimaging studies have event-related fMRI activity during presentation of provided data that support the idea that encoding and single stimuli during the encoding period would subsequent recognition require widespread activity correlate with recognition performance for those in. Suzuki tion and familiarity (Ranganath et al.

ing a memory task showed less activity during The persistent spiking of single cortical neurons repeated presentation of a single stimulus versus stemming from intrinsic properties can persist for substantial activity during sequential presentation of many minutes. This study provides a crucial requiring excitatory synaptic feedback between neu- extension of previous neuroimaging studies showing rons (Fransén et al. In addition tasks. van Turennout et al. Kirchhoff et al. Lansner and Fransen. Even greater activity was and retaining the information until the test period..138 The Neurobiological Basis of Recognition Memory of stimuli. 2006). lation to maintain activity for the full delay period ing the stimulus (K-hit). In this case. synapses are strengthened dependent upon presyn- mance on recognition tasks. in which activity during encoding and subsequent perfor.08. 2003). stimuli (Weis et al. allowing the popu- was less confident and merely had a feeling of know. in which persistent spiking vides a mechanism for recognition memory with activity of a population of neurons during the delay higher capacity than Hebbian long-term potentiation period is proposed to form the basis for retaining (Bogacz and Brown. The earliest demonstration of for novel stimuli for which a representation has not changes in medial temporal lobe fMRI activity dur. 2002. some hippocampal regions is associated with the encoding representation of stimuli based on changes in synap- of information into long-term memory (Brewer et al. 2000). 2000). The separate . the test period if they were previously active during rent graded levels of performance on recognition the sample period (match enhancement). it is not for both visual and verbal stimuli and that greater possible to maintain persistent spiking representing activity for novel stimuli in hippocampal and para. all previously viewed stimuli. while stim. 1992. Lisman et al. These excit- (recollected–R-hit) showed the least activity in the atory recurrent connections can cause a population perirhinal and parahippocampal cortices. Most models of active maintenance seen as reflecting familiarity processes. The above processes information about the stimulus presented during the focus on responses to the item alone and could be sample stimulus. Computational models have demon- 2003). 1998). in which a single neuron correlated with stronger recognition of the stimuli as can maintain persistent spiking activity without measured behaviorally. Sohal and Hasselmo. Stimuli tex (Amit. Hasselmo and Wyble. In contrast. recognition can also be modeled as the capacity to retrieve a context given a specific item (Hasselmo and Wyble. previously been formed (Hasselmo and Stern. within cortical structures such as the prefrontal cor- tion memory paradigm. This mechanism would be particularly appropriate uli becoming familiar.. Though these and other studies have shown strated how presentation of a sample stimulus could correlations between increased medial temporal lobe induce Hebbian synaptic potentiation. This synaptic mod- et al. to nonselective match enhancement. and studies have of several minutes.. Alternately. activated by the sample stimulus to continue to uli showed somewhat more activity when the subject reactivate its component neurons. These data clearly demonstrated recent models of active maintenance focus on intrin- that the reduction of activity for familiar stimuli is sic single-cell mechanisms. allowing this mechanism to contrib- different novel stimuli being encoded for subsequent ute directly to performance of DMS tasks with delays recognition (Stern et al.. and others have shown aptic activity coupled with postsynaptic activity reduced activity associated with recognition of old (Hasselmo and Schnell. which pro- of active maintenance. Amit given the highest confidence recognition rating and Brunel. appropriate. other rejections (CR). 1988. 1997. 1997). 2004). Many have proposed that repetition suppression might modeling studies have focused on the mechanisms arise from Hebbian long-term depression... 2006). seen for the incorrect responses (miss) and correct when a comparison can be made. researchers been modeled on a number of different levels. a reduction in activation associated with novel stim. 3. 1994. described earlier. 2000. the study by Gonsalves 1997.. Hasselmo and Stern.6 Computational Modeling In contrast. This study builds on extensive cognitive focus on the role of excitatory recurrent connections research using the remember versus know recogni. tic connections between neurons would be more 1998. 1996). (2005) evaluated correlations between reduced ification results in greater activity of neurons during activity during recognition testing and the concur. demonstrated that the repetition reduction occurs For recognition over longer periods. match suppression or repetition suppres- of Recognition Memory sion can be obtained by competitive long-term depression of synapses for nonactivated neurons The processes involved in recognition memory have (Sohal and Hasselmo.

cholinergic receptors impairs new learning and increases Comp. Norman and O’Reilly. Conscious Recollection: Memory Systems of the Brain. Neurophysiol. I.M.H. 2: 361–375. Desmond JE. and Cabeza R (2006) Triple involve a test item reactivating a full representation of dissociation in the medial temporal lobes: Recollection. Nature 431: 188–191. Neurosci. Norman KA. well visual experience will be remembered. Zola-Morgan S. 45: 81–87. Duck J. Behav. New York: Oxford University Press. Amit DJ (1988) Modeling Brain Function: The World of MA: MIT Press. 409: 158–162. are the roles of the perirhinal cortex and hippocampus? Nat. Brain Res. 1997. Memory and Thinking. and novelty. Brain Res. Desmond JE. Behav. Neurosci. Neave N. 2: 51–61. Brain Res. 14: 3898–3914. Fortin NJ. hippocampus. Sherman S. J. familiarity. University Press. recollection may Daselaar SM. In: Suppes P (ed. Atkinson RC and Juola JF (1974) Search and decision Gabrieli JD. Cambridge: Cambridge Fuster JM (2000) Memory networks in the prefrontal cortex. 89: 1–34. J. and Aggleton JP (1996) Neurotoxic complex neuroanatomical circuits required for reacti. Norman KA. 106: 30–38. and Riches IP (1987) Neuronal evidence of scopolamine on human memory function. 22: monkeys. pp. being recognized. Rugg MD.) Separate neural bases of two fundamental memory Contemporary Developments in Mathematical Psychology. Neurosci. Amit DJ and Brunel N (1997) Model of global spontaneous Fuster JM and Jervey JP (1982) Neuronal firing in the activity and local structured activity during delay periods in inferotemporal cortex of the monkey in a visual memory task. and Wagner AD models of familiarity discrimination in the perirhinal cortex. which could involve Corkin S (1984) Lasting consequences of bilateral medial temporal lobectomy: Clinical course and experimental an increase or decrease in activity of the neurons findings in H. J. transection or neurotoxic lesions of the perirhinal and cesses in different structures (Norman and O’Reilly. contextual features associated with a specific episode Eichenbaum H and Cohen NJ (2001) From Conditioning to (Hasselmo and Wyble. 55: 61–67. Strijkstra AM. 10: Brown MW and Aggleton JP (2001) Recognition memory: What 487–493. This recollection process could involve more Ennaceur A. et al. J. Trends Cogn. the cerebral cortex. Neurol. and Brown MW (1993) Neuronal activity related to visual recognition memory: Long-term memory and the encoding of recency and familiarity information in the primate anterior and medial inferior temporal and rhinal References cortex. Kahn I. Alonso AA. Multimodal imaging of medial temporal cortical contributions Bolhuis JJ.versus familiarity. Behav. proactive interference in a word paired-associate memory Gaffan D and Murray EA (1992) Monkeys (Macaca fascicularis) task. Behav. Brain Res. Behav. Neurosci. Wilson FA. Muir JL. 118: 223–236. 36: 61–78. Fleck MS. (1999) Recollection and familiarity in recognition memory: An . 2003). Sci. 5637–5641. Learning. of the cholinergic suppression of synaptic transmission Brewer JB. with rhinal cortex ablations succeed in object discrimination Bartus RT and Johnson HR (1976) Short term memory in the learning despite 24-hr intervals and fail at matching to rhesus monkey: Disruption from the anticholinergic sample despite double sample presentations. Hasselmo ME and Wyble BP (1997) Free recall and recognition Rev. J. USA 91: transient memory. Neurosci. 111: 187–202. Walker DL. Bussey TJ. Science 276: vol. and Mishkin M (1991) Comparison of the the role of the muscarinic-activated calcium-sensitive effects of scopolamine administered before and after nonspecific cation current INCM in entorhinal neuronal acquisition in a test of visual recognition memory in activity during delayed matching tasks. 4: 249–259. 43: 403–409. Shallice T. Behav. 80: 9–25. Neurosci. and Gabrieli JD in rat hippocampal region CA1: Computational modeling (1998) Making memories: Brain activity that predicts how and brain slice physiology. J. and Kramers RJ (1988) Effects of to recognition. The Neurobiological Basis of Recognition Memory 139 theoretical processes of recollection. processes in the human medial temporal lobe. 96: 1902–1911. In contrast to familiarity. Semin. Fuster JM (1995) Memory in the Cerebral Cortex. Attractor Neural Networks. and Squire LR (1994) The animal Fuster JM (1973) Unit activity in prefrontal cortex during model of human amnesia: Long-term memory impaired and delayed-response performance: Neuronal correlates of short-term memory intact. 5: 39–40. and Hasselmo ME (2002) Simulations of Aigner TG. Gaffan D (1974) Recognition impaired and association intact in Atri A. Alvarez P. Psychol. scopolamine. Biochem. and Aggleton JP (2000) Distinct patterns of behavioural impairments resulting from fornix based recognition have been modeled as different pro. hippocampus in certain processes underlying recognition Henson RN. Riches IP. 1081–1097. postrhinal cortices in the rat. Freeman. Pharmacol. Josephs O. and Dolan RJ memory. Curran T. (2004) Blockade of central the memory of monkeys after transection of the fornix. Proc. Natl. Brain that inferomedial temporal cortex is more important than Res. Physiol. lesions of the perirhinal cortex do not mimic the behavioural vation of the temporal or spatial context of the item effects of fornix transection in the rat. Brewer JB. Cambridge. Wright SP. and Eichenbaum H (2004) Recollection- Aigner TG and Mishkin M (1986) The effects of physostigmine like memory retrieval in rats is dependent on the and scopolamine on recognition memory in monkeys. Neurophysiol. scopolamine on performance of rats in a delayed-response Hasselmo ME and Schnell E (1994) Laminar selectivity radial maze task. Prog. Fahy FL. Neurosci. 86: 1100–1109. Exp. Behav. Glover GH. Zhao Z. Cereb. Sci. 243–293. in a network model of the hippocampus: Simulating effects Brown MW. Brain Res. Acad. Neural Biol. Neuron 47: 751–761. 96: 457–472. Behav. Science 281: Hasselmo ME and Stern CE (2006) Mechanisms underlying 1185–1187. San 264–266. Bogacz R and Brown MW (2003) Comparison of computational Gonsalves BD. 2003). representing the sample stimulus. Behav. and Glover GH (1997) processes in recognition memory. (2005) Memory strength and repetition suppression: Hippocampus 13: 494–524. 122: 309–316. Cortex 7: 237–252. working memory for novel information. Physiol. Fransén E. Francisco: W.

assemblies comprised of cortical neurons. damage limited to the hippocampal formation. New York: monkeys. J. Amaral DG. complementary-learning-systems approach. 75: 1190–1205. Neurosci. Schacter DL and Wagner AD (1999) Medial temporal lobe receptor channels in working memory. (1996) The hippocampal task. Acad. Sci. and Desimone R (1997) Object and place Norman KA and O’Reilly RC (2003) Modeling hippocampal and memory in the macaque entorhinal cortex. and Wilson FA (1982) Neuronal 105–119. Hasselmo ME. Lond. Soc. Riches IP. Eichenbaum H. 381–403. Suzuki WA. Eichenbaum H. Science 263: 520–522. 13: 5418–5432. related to long-term encoding in the parahippocampal gyrus Meunier M. Mishkin M. 24: 4172–4180. Tricarico MD. Three cases of enduring memory impairment after bilateral Kahn I. Kitchener EG. parahippocampal gyrus and inferior temporal cortex of the Lansner A and Fransen E (1992) Modeling Hebbian cell primate. and retrieval. Network 3: Rolls ET. Wilson FA. 356: 1505–1515. 24: 11088–11097. and Murray EA (1996) Pychiatry 20: 11–21. present. Buckner RL. J. Koutstaal W. Neurosci. Neurosci. Psychol. rats impairs encoding of novel but not familiar stimuli in a Schon K. J. responses related to visual recognition. and Hasselmo ME (2001) working and recognition memory in inferior temporal cortex. Neurosci. and Wagner AD (2004) Functional. Stark CE. and Murray EA (1993) during delayed matching in humans. Kirchhoff BA. Sci. Cholinergic deafferentation of the entorhinal cortex in rats Schon K. Network 11: 169–190. Bachevalier J. Sci. . memory tasks with novel and familiar stimuli. and Desimone R (1991) A neural mechanism for Stern CE. and Zola-Morgan S (1995) Rosen BR. Miller EK. R. and Rosen BR (2003) Recognition memory and the human hippocampus. Neuropsychologia 41: 378–389. Atri A. and Stern CE (2000) visual stimulation and memory on neurons of the Prefrontal-temporal circuitry for episodic encoding and hippocampal formation and the neighboring subsequent memory. Sci. and Desimone R (1996) Neural from functional magnetic resonance imaging. USA term memory. (1997) Late onset of anterior prefrontal activity during true Neuron 37: 171–180. (2005) Cholinergic deafferentation of the entorhinal cortex in J. Tricarico MD. and Squire LR Schacter DL. J. and false recognition: an event-related fMRI study. Maril A. Corkin S. Johnson MK. Effects of rhinal cortex lesions combined with Sherman SJ. information. Medial temporal and prefrontal contributions to working Science 254: 1377–1379. Meunier M. Erickson CA. Sherman SJ. Neurosci. and Hasselmo M (2003) Neuroimage 6: 259–269. LoPresti ML. Abstr. ablations of the entorhinal and perirhinal cortex in rhesus Scoville WB and Milner B (1957) Loss of recent memory after monkeys. Neurosurg. 117: 526–539. and Howard MW hippocampectomy on visual recognition memory in rhesus (2003) Scopolamine impairs human recognition memory: monkeys. Pharmacol. Squire LR. Fellous JM. Lin L. Turchi J. and D’Esposito M (2003) Prefrontal cholinergic deafferentation of the rhinal cortex on visual activity associated with working memory and episodic long. Proc. B Biol. 11: 1763–1779. Nat. Perrett DI. Neurosci. Lisman JE. J. 15: 5637–5659. mechanisms of visual working memory in prefrontal cortex of Acad. 25: Effects on visual recognition of combined and separate 9112–9123. and Mishkin M (2005) Effects of Ranganath C. Academic Press. Stern CE. 13: 1460–1478. and Brown MW (1991) The effects of Kirchhoff BA. 20: 6173–6180. Neurosci. Leonard BW. neuroanatomic correlates of recollection: Implications for 16: 5233–5255. J. models of recognition memory. Natl. Lin L. Miller EK and Desimone R (1994) Parallel neuronal mechanisms Squire LR. Atri A. In: Donaldson drugs on delayed match to sample performance of Rhesus W (ed. Behav. Sohal VS and Hasselmo ME (2000) A model for experience- term memory but not inferior temporal neurons. 25: Stern CE (2005) Scopolamine reduces persistent activity 10273–10281. and Dale AM (1998) Event-related Transient memory impairment in monkeys with bilateral functional MRI: past. Neurosci. Biochem. Wagner AD. lesions of the entorhinal cortex. Gonzalez RG. Natl. Neuron 31: 865–873. Davachi L.) Organization of Memory. and Desimone R (1993) Activity of neurons in 11: 337–346. Rempel-Clower NL. Saunders RC. Hasselmo ME. J. Rev. Brain 105: 611–646. Squire LR. anterior inferior temporal cortex during a short-term memory Stern CE. Neurosci. Neurosci. Annu.140 The Neurobiological Basis of Recognition Memory event-related functional magnetic resonance imaging study. Trans. Ranganath C and D’Esposito M (2001) Medial temporal lobe Tulving E (2001) Episodic memory and common sense: How far activity associated with active maintenance of novel apart? Philos. recognition memory in monkeys. 1: activations in fMRI and PET studies of episodic encoding 273–275. Neurosci. Neurosci. and Amaral DG (1996) J. Miller EK and Desimone R (1993) Scopolamine affects short. and Aigner TG (1997) Effects of muscarinic in an odor-guided delayed-nonmatching-to-sample task. blockade in perirhinal cortex during visual recognition. Sci. pp. Neurosci. Zola SM. J. bilateral hippocampal lesions. 102: 2158–2161. Caan AW. Miller EK. A functional magnetic resonance imaging study of McGaughy J. J. and Hasselmo ME subsequent memory after a delayed match-to-sample task. Behav. the macaque. USA 95: 773–780. formation participates in novel picture encoding: evidence Miller EK. McGaughy J. and Clark RE (2004) The medial temporal for short-term memory. Acad. 29: the absence of stimulus input enhances long-term encoding: 424. Hadfield W. Dale AM. Hopkins RO. Proc. and delayed nonmatch-to-sample task. Natl. Lopresti ML. Reed JM. Jindal M. Proc. Proc. Buckner RL. Rev. Bachevalier J. and Wang XJ (1998) A role for NMDA. Penetar DM and McDonough JH (1983) Effects of cholinergic Tulving E (1972) Episodic and semantic memory. Soc. Manns JR. Neurophysiol. Koene RA. Hasselmo ME. Acad. USA 94: 12667–12669. lobe. J. Neuroreport dependent changes in the responses of inferotemporal 4: 81–84. Tang Y and Aigner TG (1996) Release of cerebral acetylcholine 110: 611–646. Neurosci. Data and modeling. Neurol. Mishkin M. 19: 963–967. 19: 3962–3972. Hippocampus Miller EK. 106: 762–775. 27: 279–306. Behav. USA 93: 8660–8665. Neurosci. Neurosci. and future. Natl. 16: 5154–5167. Neuroreport 7: neocortical contributions to recognition memory: A 2231–2235. Otto T and Eichenbaum H (1992) Complementary roles of the Neuroreport 7: 2231–2235. and Stern impairs encoding of novel but not familiar stimuli in a delayed CE (2004) Persistence of parahippocampal representation in non-match to sample task (DNMS). J. Hippocampus 9: 7–24. increases during visually mediated behavior in monkeys. J. et al. orbital prefrontal cortex and the perirhinal-entorhinal cortices Tang Y. neurons. Neurosci.

11: 616–641. Neurosci. Lesions of perirhinal and parahippocampal cortex that spare Wixted JT and Stretch V (2004) In defense of the signal the amygdala and hippocampal formation produce severe detection interpretation of remember/know judgments. Psychon. and Martin A (2003) Yonelinas AP. Otto T. Young BJ. behavioural responses during the performance of memory Zola-Morgan S. . occipital-temporal contributions to associative and item Wagner AD. (2003) Cholinergic familiarity in recognition memory. Bielamowicz L. Specht K. 9: 4255–4270. Squire LR. 13: 251–265. Cereb. Riches IP. Neuroreport 12: memories: Remembering and forgetting of verbal 359–363. Bull. et al. Separating the brain regions involved in recollection and Warburton EC. and Rugg MD (2005) 1188–1191. and Rempel NL (1993) Yonelinas AP (2001) Receiver-operating characteristics in Damage to the perirhinal cortex exacerbates memory recognition memory: Evidence for a dual-process model. J. The Neurobiological Basis of Recognition Memory 141 van Turennout M. J. 40: 7–28. Zola-Morgan S and Squire LR (1985) Medial temporal lesions in Wilson FA. Otten LJ. Shaw KN. Klaver P. 17: 5183–5195. Mem. Exp. between contextual retrieval and item recognition. and Eichenbaum H (1997) Memory Weis S. experiences as predicted by brain activity. Neurosci. Squire LR. Psychol. 25: neurotransmission is essential for perirhinal cortical plasticity 3002–3008. et al. Cho K. (1998) Building recognition memory: An fMRI study. Zola-Morgan S. J. J. 99: 22–34. Hopfinger JB. J. Amaral DG. Neurosci. Neurosci. Rev. (2004) Process dissociation representation within the parahippocampal region. 20: 1341–1354. Neurosci. Learn. impairment following lesions to the hippocampal formation. parahippocampal and time and practice. Buonocore MH. memory impairment. Schacter DL. Behav. and Brown MW (1990) Hippocampus and monkeys impair memory on a variety of tasks sensitive to medial temporal cortex: Neuronal activity related to human amnesia. Rotte M. Brain Res. Fox GD. Hum. and Modulation of neural activity during object naming: Effects of Baynes K (2001) Hippocampal. Neuron 38: 987–996. and Suzuki WA (1989) tasks by primates. Behav. et al. Science 281: Yonelinas AP. Neuroreport 15: 2729–2733. Kroll NE. Cortex 13: 381–391. and recognition memory. Clower RP. Koder T.

1 Transverse Patterning 154 3.09. and their recent use to study both human from more recent neuroimaging studies in both patients amnesic patients and normal human subjects will be with memory disorders and normal subjects. C.1 Why Animal Models? 145 3. this chapter will Memory: Earlier Studies focus on research centered on the medial temporal lobe structures and their participation in memory pro.4 Performance of Human Amnesics on Animal Tests of Spatial Memory 158 3. Emory University.09 Animal Models of Amnesia M.09.09.3.09. diencephalic structures).09.5 Conjunctive/Relational Memory 153 3.09.2 Morris Water Maze 157 3.09. 3.4.09.09. Atlanta.4 Recognition Memory 148 3.5.3 Performance of Human Amnesics on Animal Tests of Recognition 153 3.3. After a brief review of the early work in humans brain dates back at least several hundred years when leading to the description of the temporal lobe amnesia Descartes proposed that experience left traces in the 143 .1 Radial Arm Maze 157 3.4.3 Role of the Temporal Cortex 155 3.3 Animal Models 145 3. GA. but also discussed.2 The Neural Substrates of stood (e.09.2 What Is Being Modeled? 147 3.09.6.9 Summary 161 References 162 3.09.6 Spatial Memory 157 3.09. USA ª 2008 Elsevier Ltd. We Our understanding of the nature of the human amnesic then describe several classes of behavioral paradigm syndrome has expanded enormously in the last 60 and their analogs that have been used in both rats and years.09.09.3 Spatial Tasks in Monkeys 158 3.09.5.7 Episodic Memory 159 3.1 Introduction syndrome.8 What We Have Learned from Animal Models 160 3.2 Visual Paired Comparison/Spontaneous Recognition 151 3. All rights reserved.g.5. 3.6.09. Alvarado and J. we will briefly review some of the from a growing number of animal models of different conclusions derived from the use of animal models of species. Because the field is so broad and covers a number of brain systems whose contribution to memory is certain.09.2 Transitive Inference 155 3.09.1 Introduction 143 3.2 The Neural Substrates of Memory: Earlier Studies 143 3.09.6. 3.1 Delayed Nonmatch to Sample 148 3.6. The concept that memory can be localized in the cesses. Finally. Bachevalier.4 Performance of Human Amnesics on Animal Tests of Conjunctive/Relational Memory 156 3.5.09.09. processes mediated by each region. This increased knowledge stems not only from monkeys to more specifically determine the memory the detailed description of memory disorders in processes mediated by different medial temporal lobe patients with more circumscribed brain lesions and structures.4. but less well under..09.09. we will discuss the importance of develop- ing animal models for the study of this syndrome. These animal models have been used to refine amnesia and how they have contributed to our under- our understanding of the specific brain regions involved standing of the neural substrates of both human and in human amnesic syndromes and the critical memory animal memory.

PH.’s left medial temporal lobe (MTL). What was Figure 1 Multiplanar views of 18 averaged T1-weighted magnetic resonance imaging volumes showing preserved structures in H.’ localized memory traces had been laid to rest. H. transaxial view. surface rendering showing locations of transaxial and coronal planes. a specific locus where memory resided could cortex to pinpoint the critical tissue. The ability of knife cuts or ablations these stimulations deep in the temporal lobe. Abbreviations: CS. Thus. 1958). The asterisk marks the intersection of the three viewing planes. M. or ‘engrams. (Lashley. M. In 1950. ‘In search stimulation. Neurosci. although childhood memories were equipotentiality. which equivalent.? Nat. Rev. Wilder Penfield. Limited regions may be essential memory and the brain was that of patient H. Top left. and a somewhat milder which states memory is distributed throughout the retrograde amnesia for events closer to the time of brain (no centralized loci). bottom left. Reprinted from Corkin S (2002) What’s new with the amnesic patient H. and mild memory impairments were observed in some patients It is not possible to demonstrate the isolated locali- with unilateral temporal lobe resections. However. he dis- to affect memory was directly related to the amount covered that the patient would suddenly describe of tissue damaged. 1662/1972). he would stimulate areas of efforts. for learning or retention of a particular activity. coronal view. At resided. M. top right. 1957). he published the results of detailed memories depending upon the location of the 30 years of research in his landmark paper. with permission from Macmillan Magazines Ltd. These two principles However. 1950: 478) subjacent temporal cortex (See Figure 1). that persists to this day. collateral sulcus. In spite of his removal of epileptic foci. it wasn’t until the beginning of the twen. and (2) the principle of his operation. Furthermore. explained why only large lesions produced deficits tieth century that systematic searches for these traces and why recovery of function was possible from dis- were made. hippocampus. of this surgery was a profound anterograde amnesia ples of brain function: (1) principle of mass action. The engram is represented throughout included parts of the amygdala. quite different results were dis- then systematically disconnecting cortical areas covered in the brain mapping work of neurologist using knife cuts. bottom right. In preparing epileptic patients for to assess their effects on performance. . the zation of a memory trace anywhere within the landmark case that to this day drives the study of nervous system. which proposes that if one area of the intact (Scoville and Milner. or ablating areas of cortical tissue. it appeared that the theory of where in the brain memory traces. hippocampus. tions of the damaged region. parahippocampal gyrus. and the region. 3(2):153–160. just caudal to the left MTL resection. stimulating the same locus of the engram.’ in which he concluded that: elicited the same memory (Penfield.144 Animal Models of Amnesia brain which could then be retrieved through the brain is damaged. seen best in the transaxial view. In a decades-long search to discover crete lesions. Karl Lashley trained rats on mazes and about this same time. During some of not be identified. another is able to assume the func- actions of the pineal gland (Descartes. The result These conclusions were summarized in two princi. but This patient was treated for intractable epilepsy within such regions the parts are functionally by the bilateral removal of the temporal lobes. sagittal view.

That is. Earlier neural circuits that contribute to human memory. 3. in spite of the high degree of anatomi.. The underlying assumption. is that there is conserva- and served more as tests of recency rather than recog. To better approximate the recognition tasks systems. 1974) used small stimu. 1984). M.09. whereas amyg- (1) the hippocampus and possibly other structures dalectomy or hippocampectomy alone produced in the human medial temporal lobe (MTL) was mild impairments at best. in stimulus was novel (presumably known because the spite of this devastating memory impairment. Mishkin and Delacour (1975) used brain is sufficiently similar to that of the human brain large pools of stimuli. M. known as trial unique. As is discussed elsewhere in these volumes. This what specific effects in rat or monkey memory (e. M. that is. 1958). also shown to be spared in the operated monkeys coming was similar evidence from the animal (Zola-Morgan and Squire. and at the cellular and monkeys. which is supported by lus pools and were very difficult for monkeys to learn detailed anatomical studies. came directly from example. which has been a major 1967. yielded at last an impairment in monkeys the structure commonly damaged in patients with apparently equivalent to the memory deficits suf- memory impairments (Penfield and Milner. for spared even in global amnesics. memory systems in the brain. The memory). M. damage to the hippocampus in animals lobe damage led to the idea that there are multiple initially appeared to have very limited (Douglas. effec- required to encode some new memories. termed ‘declarative’ or ‘episodic/ revealed much about the neural basis of human mem. the organization of the infrahuman used in humans. and even rodents.e.g. and deficit as being limited to particular long-term mem- other amnesic patients in the subsequent years ory processes. guiding force in the research of the last 30 years. Mishkin (1978) Several important points came from this research: demonstrated for the first time that. Cohen. the 1981.. but some tively replicating H. i. Continued work in human amne- (3) to the extent it was involved in retrieval. this role sic patients further characterized the nature of the must be time limited as well. The benefit is that animal stud- version required the subject to indicate which ies can provide a degree of experimental control that . versions of this task (Gaffan.’s damage. operatively.. combined lesions. semantic’ or ‘explicit’ memory (Cohen and Squire.3. and delays (Figure 4). creating a ‘trial-unique’ version to permit such comparisons.e. Animal Models of Amnesia 145 particularly interesting about this patient was that. recency memory). M. 1984. memory for the hippocampus or related MTL structures were not an individual stimulus rapidly decayed across brief the permanent storage site of long-term memory. level in invertebrates. ory. cal homology between human and nonhuman primates. the goal in the creation of models of human recognition memory. Olton. 1971.09. The study of H. the new scale (see Figure 2). fered by human amnesic patients. However. 1972). What was not immediately forth. but even once relearned. Combining this task with damage to the evidence from H. (2) because some past memories survived. research has vastly increased our knowledge about Mahut. tion of structure and function among mammalian nition. ‘nondeclarative’ or ‘implicit’ memory were patients like H. rodents. 1975) and some. 1977). in 1975. For some time it appeared the neural bases of memory and generated model that there was no comparison even between humans systems in monkeys. Mishkin and Delacour described a modifica.3 Animal Models tion to the matching-to-sample memory task 3. 1968). human amnesia has been to better understand the delayed nonmatching-to-sample (Figure 3). impairment and sparing of memory functions in In spite of the effects of MTL damage on human both humans and monkeys with medial temporal memory. rather than one which required ences in size and complexity. with respect to the med- ial temporal lobe structures (see Figure 2). The pattern of literature. the basic organizational the subject to remember which stimulus had been principles in the brain apply up the evolutionary seen most recently (i. Weiskrantz and Warrington. not only produce memory functions remained intact after MTL a severe impairment on relearning the task post- damage. In spite of great differ- of the task. and those pro- idea of multiple memory systems. and other patients with milder medial temporal lobe similar to that sustained by impairments seemed to point to the hippocampus as H.1 Why Animal Models? developed by Gaffan (1974) that could be readily learned in monkeys and that was closer to tests of From the outset. some other stimulus was remembered. some of which are cesses that are spared in human amnesics termed. recognition learning capabilities survived (Corkin. Tulving.

With very few exceptions. only at long delays). in the case of organic causes how different circuits interact with each other in the of amnesia. and rat brains illustrating some components of the medial temporal lobe important for memory function: the amygdaloid complex. it smaller lesions do impact memory. mem- ing normally (a caveat for all lesion studies). memory for objects vs. M. Although physical damage can be detected and mapped out. either with is not always clear whether ‘spared’ tissue is function. and fornix.g. the neural damage asso. tigation of how individual structures contribute to lation of the individual contributions of brain memory will better enable the understanding of structures to memory. methods available to animal researchers improve these restricted lesions do not produce dense amnesia selectivity in the assessment and manipulation of in animal or human subjects. As described earlier. the development of treatments. service of normal memory. monkey. Nevertheless. initial studies suggested that only when damage ciated with human amnesia has been fairly extensive.03.. but also for the assessment of amnesic symptoms and.146 Animal Models of Amnesia Human Fornix Amygdala Hippocampus Monkey Fornix Amygdala Hippocampus Fornix Rat Amygdala Hippocampus Figure 2 Midsagittal view of the human. is almost impossible to achieve with human studies. which allows not only for the reve. . See also Chapter 3. hippocampus.. Furthermore. though memory impairment observed in monkeys.g. Relative sizes are not to scale. greater specificity (e. was a clear and not always determinable. approximated that sustained by H. the inves- brain function. The ory for space) or degree (e.

but the novel object possible temporally graded retrograde amnesia for is now rewarded (i. whose prolonged a specific context (i. and Aggleton. because they can recall and Sample describe the event in detail at a later time. there is an ongoing debate as to how best to ITI define the animal equivalents of declarative and non- Trial 2 declarative memory systems (See Chapter 1. can determine whether only information learned in ture for human Korsakoff patients. Trial 2: For this and short-term memory. the conditioning. It is not possible to measure recall in the same way in animals. We can then assess whether animals forget The focus thus far has been on the history of human newly learned information more rapidly as a result of amnesia resulting from damage to the MTL. 1999. or if the alcoholism has produced damage to the diencephalic processes that allow encoding of context are some- structures of the brain. it can be verified that someone has a memory of an event. Thus. a new pair of objects serves as the steeper in amnesic patients than in normal subjects.. After a delay.2 What Is Being Modeled? seen memorandum and use that information to guide behavior. how are we to ask a nonverbal animal to + ‘declare’ what is in its memory store? Thus in animal models. These include global Trial 1: The monkey is presented with a sample object (i. the research has been more These ‘memory systems’ are reviewed elsewhere in extensive for structures in the MTL. 1990).3. In addition to the development of specific tests of Squire et al. although we cannot determine whether an animal can recall a specific memory. containing rich memory representations that + can be recalled at will. trial unique)..09. as well as anterograde how impaired. there + are some clear consistencies in the human amnesic syndrome that may be taken as core deficits that can Figure 3 Delayed nonmatching-to-sample (DNMS) task. 2004). In humans. that there is a wide litera. with a forgetting curve that is all subsequent trials. In this instance we infer memory by observing Choice a change in the animal’s behavior. Animal Models of Amnesia 147 Trial 1 the remainder of this chapter. and intact intertrial interval (ITI). Squire.g. Finally. we should be noted. Whereas such divisions have a Choice clear basis and intuitive appeal when applied to humans... declarative amnesia (See Chapter 3.. Aggleton and Brown.. Although animal studies have animal memory. during which the test tray is obscured.04. we can deter- mine whether an animal can recognize a previously 3.e. After a brief information closer to the onset of amnesia. be assessed with animal models. or Delay implicit system such as skill learning (for which there is not always a particular recollection of having acquired the skill). Vann or lesser degree.e. + however. 1982. Nevertheless. multiple modality) anterograde amnesia for facts covering a food reward (þ) in the central well of the testing and events. tied to specific neural structures.e. the ques- tion must be raised: what exactly is it that is being modeled? Human memory has been defined as com- Sample prising two main systems: a declarative. and so we will these volumes. sample and choice (i. or explicit system. so discussion will be limited to point- focus on research in these structures of the brain in ing out where different theories make specific . and a nondeclarative. we can use several means to determine Delay whether an animal can demonstrate that it has experienced an event and remembers it on some level. It particular brain damage than controls. nonmatch-to-sample). episode) is forgotten. however. spared memory for remote events.e.04). To this end. the next trial begins. to a greater in memory (e. in the presence of intact skill learning or tray. but sample object and a novel choice object are presented simultaneously covering the lateral wells. a number of theories of memory demonstrated a role for the diencephalic structures formation have been developed that are.

(empty). hiding either a baited or recall. 3. a food reward can be retrieved. This is a loose distinction. However. Training takes place in two phases for strate that a given stimulus has been previously each trial: sample and choice. for example. In the basic paradigm. the Two paradigms in particular have been used to assess monkey is seated in a sound-attenuated chamber recognition memory in rodents and monkeys: the (Wisconsin General Testing Apparatus. When the the visual paired comparison (VPC) task (also screen is raised. The task is simple in that monkeys learn to displace junk objects to obtain a hidden food reward.4 Recognition Memory recently. After a brief period. can be used This task was the first to successfully demonstrate a to assess spatial or relational information and vice deficit in recognition memory consequent to the same versa. phase. the monkey views a testing tray con- known as preferential looking. During the sample seen (judgment of prior occurrence. animals with aspiration lesions of the amygdala.09. and when displaced. In particular. we will focus on tasks re- quiring recognition memory. when delays are increased from 30 to 120 s or the number of to-be-remembered objects is increased from three to ten. categories.09. N. or WGTA) delayed nonmatching-to-sample task (DNMS) and behind an opaque screen (see Figure 3). while a novel object covers the opposite ticular. typically . taining three equidistant food wells. A. each task assesses whether subjects demon. Like some tests of human recognition or covered with junk objects. with permission from Macmillan Journals Ltd. or spatial memory. unoperated controls. which can be nition). animals with aspiration lesions of the hippocampal formation. potentially. these tasks may be seen as testing memory for lateral well (baited). food is only located under objects that have not been seen 3. H. predictions about a given task and to group the ‘facts’ as well as the duration of the memory trace. That is. substantial differences in the task require. Brown. However. the specific brain regions sample object is moved to cover a lateral well necessary for successful performance of each. a single object covers the central food well. conjunctive/relational memory. animals with combined amygdala and hippocampal lesions. or spontaneous recog. neural damage that produces human amnesia. A þ H. animals are trained to associate novelty with reward. but discussion of particular memory tasks into theoretical with slight parametric modification. Adapted from Mishkin M (1978) Memory in monkeys severely impaired by combined but not by separate removal of amygdala and hippocampus. Nature 273(5660): 297–298. an empty well. ments may alter both the demands made on memory The screen is then lowered and the now-familiar processes and. 1996).148 Animal Models of Amnesia 100 N A 80 H Percent correct 60 A+H 40 30 60 120 3-item 5-item 10-item Delay (s) List length Figure 4 Performance of monkeys with damage to the medial temporal lobe on DNMS. In par.1 Delayed Nonmatch to Sample as recognition memory tasks.4.

1993. When a distrac- impaired performance capabilities of amnesic patients tion is inserted into the delay period (opening the such as H. it is clear that the perirhinal (1978) initial findings were due in part to the cortical cortex is important for object memory. both by the temporal cortical region and also by The use of neurotoxic lesions of the hippocampus active mechanisms when necessary. 1984). a meta-analysis of the results across several to be remembered (i. 1960. motor task at longer delays. Indeed.. damage to these cortical areas (collectively 1999. to alternate performance strategies that allow the These results seemed to match the spared and animal to bridge the longest delays. 1998. Using more selective lesion techniques. 2004).. possibly con- has not entirely solved the question. M. 2001. Animal Models of Amnesia 149 ranging from 10 s to 10 min. who could not retain new information screen at short delays. 2001). or nonvisual laboratories suggested that there was a negative cor- modalities. 1994.. the screen is raised. 1983) while sparing relation between lesion size and memory deficit at performance on tasks such as visual discrimination delays of 10 min (Nemanic et al... but see also Hampton. thus with 24-h delays (Malamut et al. which were then tested in reverse order.. in this become possible to parse out the individual contribu. impaired performance (Mahut et al. list also Beason-Held et al. even within studies. for similar findings). short delays. see Zola and Squire.. For duced an even greater impairment on performance example. or by changing the nature of the information However. 1982. with the amount of The early work by Mishkin and others on this task damage on performance of the task (Baxter and confirmed that large lesions to the MTL dramatically Murray. 1989. by contrast.. 1996). in particular damage sustained by the animals with combined for object identity (Ungerleider and Mishkin. it is currently being argued that the performance of this task in monkeys (Figure 5. and Mishkin (1998) showed that monkeys with selective the monkey must choose one of the objects. (1989) showed that the of DNMS. Nor were they impaired on lists of 40 item. 1999. by chan. Indeed. Squire and Zola-Morgan. Baxter and Murray. hippocampal damage impaired DNMS per- delay between the sample and choice phases. memory can be and colleagues (2000) found that regardless of lesion further manipulated by changing the length of the method. campus and amygdala produced as large an Hippocampal damage.. damage to the perirhinal cortex pro- tions of MTL structures to recognition memory. for reply). When the animal reliably resulting in delays ranging from 30 s to 40 min. even after addition of cortical-sparing amygdala damage to ani.. lesions.. Zola masters the rule governing the task. for a different interpretation of the data). as a mea- pocampal damage at the longest delay. 1993.g. 2001. . learning). amygdala in the initial Mishkin studies. 2005. 2004). formance at delays of 10 min and beyond (see ging the number of items to be remembered (i.. Eacott et al. Buffalo et al. 2001.’ then the novel item is chosen and the food items. perirhinal cortex may be important for visual percep- Zola-Morgan et al.. suggesting that Mishkin’s anatomy of the region. such as tactual or olfactory). having the animal perform a for more than a few seconds without active rehearsal. which is maintained Mishkin. the results from correlates positively with the amount of hippocampal the DNMS task have been called into question.e. relation. 2005).. Buckley et al. those learning in which monkeys learn a set of 20 concurrent results suggested that the lack of impairment in mon- discrimination problems that are presented only once keys with hippocampal damage might be due in part per day. By contrast.. same study. Mishkin. one study has suggested a positive cor- 1978. see Milner. removed along with the hippocampus and (Suzuki et al. 1982). Murray and trolled by the prefrontal cortex. with or without distraction. However. or removing it from but could improve over many trials to perform tasks the apparatus during 10-min delays) an impairment such as mirror drawing or the incomplete figures task is revealed in animals with selective neurotoxic hip- (Gollin. If the neurotoxic damage to the hippocampus and amyg- animal remembers the sample object and correctly dala showed normal DNMS performance out to applies the nonmatching rule ‘choose the unfamiliar 2-min delays. tion as well as memory (e. Zola-Morgan et al. Murray and Bussey. further studies demonstrated that and that damage limited to this region produces a damage to the subjacent perirhinal and entorhinal severe impairment in visual recognition memory cortices. devastates 1999). Indeed. reward can be retrieved. leaves short- impairment as the combined lesion (Murray and term memory intact on this task. However. which sure of hippocampal memory function. Meunier et al. However. referred to as ‘rhinal’ cortex) that spares the hippo. stimulus location. it has damage (Nemanic et al.e. mals with hippocampal damage did not exacerbate Taken in context with what we now know of the their memory impairment.

. followed have also questioned the view that the hippocampus is by removal of the entire hippocampus. familiarity processes et al. Rothblat or to abnormal discharges in the remaining cortical and Hayes. determines the degree to which recollection or famil- ticularly detectable through the use of specific iarity contributes to performance (Yonelinas. Furthermore. actually spared necessary for object memory. matching-to-sample (DMS) have shown no effect on Finally.150 Animal Models of Amnesia 100 90 N Percent correct 80 Erh 70 PRh 60 Rh 50 10 30 60 120 3-item 5-item 10-item Delay (s) List lengths Figure 5 Effects of entorhinal (Erh). It has been sug- where cortical damage was also present (e. however. although ischemic damage in curve reflects the contribution of both recollection human amnesic patients impaired performance on a and familiarity. the shape of the ROC 1996). Clark gested that two processes support recognition memory et al. it is worth noting recent developments in recognition memory in rodents. For normal memory. 1997). at best. histological methods (Bachevalier and Meunier. 1996). Neurosci. thus seemingly providing a direct linear shape (Figure 6(a)). processes are in play when an animal or human must pocampus is extremely sensitive to ischemia.g. Several rodent versions of the DNMS task for showed that ischemia-related memory impairments objects have been developed (e..g. They showed that producing ischemia. later work in rats by Mumby and colleagues (1996) vilinear shape. except in instances the nature of recognition memory. However. The hip. These two processes can be measured by the use a single hippocampal cell field (CA1) in human of receiver operating characteristic (ROC) (see patients (Squire and Zola.. for review). Rempel-Clower et al. the effect of hippo- Mumby (2001).. showing an asymmetrical and curvi- DNMS-like task. 1994. the majority of studies assessing the campal damage on DNMS for objects is mild and effects of hippocampal damage on DNMS or delayed restricted to long delays and lists. 1986. 1993.. and as has determine whether or not a given item was previously been shown in humans. and rhinal (Rh. results from rodent studies tissue. Erh þ PRh) lesions on DNMS performance as compared to unoperated (N) animals. Thus.. and Murray EA (1993) Effects on visual recognition of combined and separate ablations of the entorhinal and perirhinal cortex in rhesus monkeys. Adapted from Meunier M. Bachevalier J. with permission from the Society for Neuroscience. Mumby and Pinel. ischemic events produce quite seen (see Aggleton and Brown. Aggleton. In the profound memory deficits in monkeys (Zola-Morgan absence of strong recollection. 2001) or in those studies where hippocampal performance: recollection and familiarity. 1994. 1987). 1999. 1985. 1996). and par. J. the familiarity component has a symmetrical and cur- ory. It has been proposed that link between the hippocampus and recognition mem. there is evidence to Figure 6) curves in which the shape of the curve suggest that the damage is more widespread. were likely due to either extra-hippocampal damage Kesner et al. perirhinal (PRh). Mishkin M. 13: 5418–5432. whereas the recollection component is . As recently reviewed by learning on DNMS. These two damage was produced by forebrain ischemia. will determine whether they recognize the item or although ischemia was thought to have damaged only not.

Wixted and ROC curve by manipulating the rat’s ‘bias’ level.6 0. shifting their tendency to 3. Wais et al.. Animal Models of Amnesia 151 (a) Word list recognition in humans (b) Familiarity (c) Recollection 1 0. This familiarization period may range from in rodents is necessary for recall. linear. monkeys passively view the ROC curve for control rats was asymmetrical and a visual stimulus.. (2004) showed that rats demonstrated both Cipilotti et al. illustrating the recollection component of the curve. By Squire. 2005. (f) Performance of control rats after 75-min delay. rats shifted their response criterion to more or less strict levels. after very long delays. rats. 2006). Nature 431(7005): 188–191. the VPC task utilizes a famil- alarm’).4 0. suggesting that both recollection and familiar. Thus.4 0.6 0.6 0.2 Visual Paired Comparison/ correctly identify previously encountered odors Spontaneous Recognition (‘hits’) or incorrectly identify new odors as old (‘false. reprinted with permission from Macmillan Publishers Ltd. the image Although these results remain to be replicated. 2006)..4 0. In an odor recognition task. obtain rewards. Aggleton et al. and Eichenbaum H (2004) Recollection-like memory retrieval in rats is dependent on the hippocampus. For VPC.6 0. typically a black/white image.. selective tage of monkeys’ natural preference for looking at hippocampal damage produced symmetrical and cur. and in contrast vilinear ROC curves (Figure 6(e)).8 1 Probability of false alarms Probability of false alarms Probability of false alarms Figure 6 Example of receiver operator characteristic (ROC) curves for normal human performance (a) and the portions of the curves contributing to (b) familiarity and (c) recollection.8 Probability of hits 0. although there is active debate the asymmetrical and curvilinear components of the about the interpretation of the results (e. From Fortin NJ. they cease visual explora- these results suggest that the hippocampal formation tion).2 0. and humans .8 1 0 0. but the task ity play roles in normal rodent recognition memory demands are quite different. Similar to DNMS. Wright SP.e. 2004. Monkeys.. 2004). Fortin et al. indicating that only to the DNMS. 1998. This task takes advan- performance (Figure 6(d)).. At this point.2 0.09. side by side show similar shifts in the ROC characteristics with a novel image. it in fact requires no specific learn- familiarity was contributing to their performance ing and no forced choices between two objects to (Fortin et al.4 0. to show habituation (i. (e. Finally. (e) Performance of normals vs. Yonelinas et al.4 0. but not recognition.2 0 Odor list recognition in rats Odor list recognition in rats Lengthened delay period (d) (preoperative performance) (e) (postoperative performance) (f) in control rats 1 Controls Hippocampus 2 1 0. iarization phase and a choice phase. studies disappears. novel things in their environment.. By contrast. changing the amount of reward and the effort required to retrieve it.g. 15 to 30 s of looking time.4.2 0. and after some delay period (as brief as in humans with selective hippocampal damage also 1 s) the image reappears on the screen..8 3 Con F Probability of hits 4 5 0. (d) Performance of normal rats on odor discrimination tasks.g. hippocampectomized rats illustrating characteristic familiarity curve in the hippocampal group.2 0 0 0.8 1 0 0. asymmetrical and linear. suggesting that only recollection contributed to and are allowed to look at it for a sufficient period performance after long delays (Figure 6(f )).6 0.

but the role of latter study used five daily familiarization sessions. .. 1994. Nemanic recognition of single items. thus Bachevalier. Spared novelty longer retention.. deficits in the VPC task. (b) Control task: A similar spatial array is presented in the sample phase. Clark et al. rodents levels on VPC are much lower than on DNMS. although the the task more spatial in nature. an identical old object and a novel one effect is reliable. 1988). it should not be surprising that performance (Ennaceur and Delacour. olfaction. By contrast. Clark and relationship to each other. In this task. whereas parahippocampal neurotoxic damage to the hippocampus. though the sensitivity but did find an effect on recognition for location may be greater in detecting recognition memory (Ennaceur and Aggleton. 2005). the preference has been reported in rats with selective role of the hippocampus and area TH/TF in VPC hippocampal damage at delays of 48 h (Forwood becomes more prominent when arrays or locations et al. 2004) and recently as of objects are used as stimuli (see Figure 7). 2000. Following a brief delay. and 10 s find effects of fornix damage on object recognition. Similar to DNMS.. but this not present. at delays as short as 1 min (Pascalis and Bachevalier. and ion. and touch during a sample phase. locations. the sample array and a novel arrangement (in which three objects are repositioned) are presented to the monkey. where animals are trained to a 90% correct criter.. 2004). 2002). Similar to the DNMS task. Ennaceur et al. using methodology similar to least three regions of the MTL maintain a similar the monkey visual VPC experiments. colleagues showed deficits as early as 1 min in tributes in the initial encoding and short-term rodent visual preference (Clark et al..152 Animal Models of Amnesia naturally prefer to look at (explore) the stimulus regardless of their spatial relations (Nemanic and they have not yet seen (novelty preference). making long as 3 weeks (Mumby et al. damage to the peri-postrhinal region (a) (b) Spatial VPC Control task Delay Delay Figure 7 Spatial variant of the visual paired comparison ( VPC) task using arrays of objects. Initial studies failed to delays (30 s for parahippocampal gyrus. however. Unlike the DNMS task. Thus.. we infer that they remember the familiar stimulus. VPC has are placed equidistant from the rat. perirhinal cortex is evident in all object conditions. Variants of the task have been used to explore 1999. Winters et al. 2005. the contributions of at 1997). Zola et al. for perirhinal cortex). By contrast. damage to figurations as well as the effects of specific neural the temporal cortex produces impairments at shorter damage on performance. 2000) after retention of visual stimuli. but the novel array has one object replaced with a new one. Studies in rodents initially employed what Given that the subject is not actively performing a was termed the spontaneous recognition paradigm task. typical novelty preference is in the then after some delay during which the object is range of 65–70% preference for novelty. are allowed to examine a junk object using vision. perirhinal con. With VPC. Rats preferen- proven to be very sensitive to hippocampal damage tially spend time exploring the novel object. 2001. Later stu- area TH/TF and hippocampus are required for dies have also been equivocal.. (a) Spatial VPC: Five objects are presented in a visual array. or item con- et al. that is.

test specific memory functions in humans with simi. touch. required by the matching tasks that may encourage sumed damage was either diencephalic (Korsakoff’s the subjects to rely on familiarity in the absence of patients) or hippocampal (ischemic/anoxic patients) recall to support good performance. to lesion differences. Animal Models of Amnesia 153 showed impaired novelty preference (Ennaceur on the finding that performance levels on the task et al. however. patients with damage restricted to digms allow the rat to physically explore the object.04). demonstrating preference for novelty at the 0-s better understand specific impairments in human delay only.. pal damage impairs VPC performance at much shorter Tests of human amnesic patients on animal tests of delays than on DNMS (Clark et al. are largely in agree. a number of memory systems presumably is a model of procedural memory. 2001).5 s) but not when delays were extended to 3. Squire et al. severe impairments in memory on VPC (Pascalis et al. while relatively few. smell. Squire et al. tively. but the reported results map onto those from other species. differences. hippocampus or temporal cortex. rodents and suggests that the impairment seen in the In sum. she was not tested at longer larly selective damage and to compare the results with delays on DMS. her patients.09. models and the human impairment may be due to task This patient had previously shown impaired recall. were severely impaired on the DNMS task.3 Performance of Human Amnesics 2 min or 2 h.. Thus differences other obvious pathology as measured by magnetic in the nature or extent of impairments between animal resonance imaging (MRI) (Holdstock et al. (1988) suggested that.5 Conjunctive/Relational on DNMS in patients with damage to the hippocampal Memory formation. who Human amnesia has. unlike in part from the methods used. By contrast. 2004). 1996. Similarly. they were also impaired on a concurrent discrimination task As is described in detail elsewhere in these volumes that is routinely spared in monkey studies and that (See Chapter 1.. the recent performance fell to chance. most rodent para. the hippocampus have shown spared performance even using any or all of three sensory systems: vision. For region damaged.. tests (Holdstock et al. by and large. Nemanic et al. 2004). as described reports of patients with selective damage to individual above. She also demonstrated Thus. however. Zola et al. 2000b)..08). It is possible that were directly related to declarative knowledge of the this difference between rodent and monkey results solution. Similarly. but the absolute animals.. Reed and Squire (1999) reported marked impairments 3. it is important to test humans on animal mem. humans normally learn this task declara- of the Clark et al. 2000b).. For example. 2000a). example. been tested with became amnesic after a possible ischemic infarct. these impairments are often larger than those delay for which impairment occurred differs for each obtained from matching or nonmatching tasks. The primate studies were limited to which follows more closely the results in monkeys and visual recognition (see Mumby. Winters et al. Unlike the monkey studies. or to evolutionary but intact recognition on standard human memory forces.. DNMS performance is related to the forced choice (1988) reported that a group of patients whose pre.. 2000a).04).4. her performance was as good as controls on the components of the MTL provides an opportunity to DMS task with delays up to 30 s (Holdstock et al. numerous tasks that are quite different from those resulting in reduced hippocampal volume and no used with animals (See Chapter 3. Regarding the lesion differences. As noted. By contrast. McKee and Squire (1993) reported novelty preference in human amnesic patients when the delays were short (0. each region contributes to recognition been demonstrated on the VPC task. At the longer delays of 5 and 10 s. on DMS at delays up to 30 s (Holdstock et al. 2000. 1978). and as in memory in the same relative manner.. (2000) study. As for the animal studies. Unfortunately. recognition. Based have been proposed to uniquely account for the . recognition memory impairments have in both tasks. it ment with the animal findings when large temporal is possible that the difference between VPC and DMS/ lobe lesions were used. ory tasks both to validate the behavioral assays and to 2004). in addition to the lack of language capabilities. With the exception monkeys. and eventually may inform those reported in rats and monkeys in that hippocam- the assessment of human patients. these two measures of recognition memory early studies of Squire may have resulted from damage show very different sensitivity to selective damage to to the adjacent MTL structures (See Chapter 3.. R.09. similar to the initial monkey studies (Mishkin. Direct comparisons of VPC and DMS on Animal Tests of Recognition performance in human amnesic patient Y. 2000.

. O’Reilly and mance. <CA> allows for the disambiguation of the indivi. The impairment for conjunctive information. and C stand in unique 2001). but were unimpaired at learning a set of that this task requires a relational solution. temporal cortex were impaired when stimulus pre- Bþ versus C. although they could manage C are individual stimuli.b) basic findings.b). Rats with hippocampal lesions were able to perform points address the way in which the hippocampus the task only when an elemental solution was possible forms associations among individual elements in the (i. but not when the third was added. respec. 1995a. they found that tions. for the Driscoll et al. Dusek and Eichenbaum (1998) 3. which is the addition of the third nonspatial memory tasks except by invoking ‘con. however. In another swim version of the task. 2005).09. and those which contend that this is too point in training beyond which configural representa- narrow a view and does not account for deficits on tions are needed.g. have generated much attention in the hippocampal formation (Alvarado and Rudy. but further found a retrograde that are representative of the two viewpoints. and distinctions between these accounts will be addressed Eþ vs. More recently. they were unimpaired at learning three and colleagues taking the relational viewpoint. of the possible elemental solution (pick A. relationships to each other (Dusek and Eichenbaum. It is formed of three concurrent discrimination animals with fornix transection or damage to the problems that are presented as follows: Aþ versus B–.’ Two of these viewpoints. and þ or  indicates the to perform when the problems were presented in rewarded or unrewarded member of the pair. Bachevalier. fied such that they have no N-methyl-D-aspartate tation of the three conjunctions <AB>. and Cþ versus A. Aþ vs. Within the animal literature.. These impairment for both elemental and configural informa- two tasks have been selected because they have been tion learned presurgically. F). By contrast. the elemental tested in a variety of species. B. A). D. It is at this point that text.e. 1989). B. in detail in the relevant chapters of this book. but learning new elemental damage.. Thus. Cþ vs. discrimination (Cþ vs. Sutherland. Alvarado and Rudy (1992) trained normal rats Alvarado et al.. <BC>. Using a solved on the basis of individual stimulus representa. which have been genetically modi- are inconsistently associated with reward. 2002) and adult macaques with on a visual discrimination version of the transverse neurotoxic hippocampal damage (Alvarado and patterning task. Because the individual elements (A. C) this discussion has often centered around theories produced configural learning in normal rats. remembering previously learned elemental prob- animals and humans with hippocampal or MTL lems was impaired.1 Transverse Patterning used an olfactory digging version of the task in which The transverse patterning task was first described by three scented sand-filled cups served as stimuli.. slightly different training paradigm. the hippocampus and subjacent cortex (but not those Using a swim-maze version of the Lashley jump with damage to the amygdala and subjacent cortex. monkeys with neonatal damage to 1998). in that the concurrent odor discriminations (Rondi-Reig et al. B. and O’Reilly resolved). were also impaired on the odor version of transverse dual elements. and impairment was in the retrograde direction only have provided evidence for nonspatial deficits in (i. and sentation was random.5. in spite proposing that the hippocampal formation is pre. where A. 1989. the conjunctive hippocampal damage was predicted to impair perfor- account (Sutherland and Rudy.154 Animal Models of Amnesia types of deficits observed in animals with selective they were able to determine that learning two pro- hippocampal damage. (2005) replicated Alvarado and Rudy’s purposes of this chapter.e.. including humans. blocks. subject must learn that A. was both anterograde and retro- grade (Driscoll et al. 2005b) were impaired in an object . In a related study. However. By adding problems incrementally.. avoid C) the disposed toward spatial or contextual memory animals could have used. there is a clear processing. problems was intact). blems at once (e. Both of these view. Similarly. we will focus on two tasks (1995a. a represen. and a Spence (1952) as an example of a task that cannot be food reward was buried in the correct cup. search for hippocampal function. and (NMDA) receptors in hippocampal cell field CA1. when two problems had to be simultaneously environment. B. with Rudy. The elemental discriminations (Aþ vs. taking the conjunctive viewpoint and Eichenbaum Furthermore. B and Bþ vs. and C) knockout mice. stand. these authors showed that NR1- tively. a prediction that has been empirically proven Rudy 2001) and the relational account (Eichenbaum in further rodent studies using neurotoxic lesions of the et al. it has also been suggested patterning.

Eichenbaum.2 Transitive Inference sensitive to perirhinal damage but not fornix damage This task specifically tests predictions from the rela.. Saksida E.. it is noteworthy (Dusek and Eichenbaum. correct within 90–300 trials for rats and 300–500 or parahippocampal cortices in conjunctive or rela- trials for monkeys). 1989. with the excep. Animal Models of Amnesia 155 version of the transverse patterning task. The reason for the differences between the respond correctly. Similarly. 2007). but the precise nature of this role is yet to test between B and D. The fact that perirhinal and and Eichenbaum. in the touch screen version of the transverse 2003. 2003. 1994. tional memory view of hippocampal function.. O’Reilly series of representations. combined and have implications for the degree to which the perirhinal-entorhinal lesions in rodents impaired hippocampus or cortical structures are preferentially transverse patterning performance (Dusek and engaged. for alternative patterning task. sequence of discriminations Aþ versus B.. and Dþ versus E (Dusek on transitive inference. Bþ ver. Finally.e. tive memory that allows for the inferential or flexible 2005. van Elzakker et al. 1995. parahippocampal lesions do impair performance on tion of the endpoints A and E... For example. Cþ versus D.. such as the biconditional discrimination (e. 2002. 2003. it should be able to infer that B is greater than D sus C. Saksida task encourage the use of specific strategies that et al. In support of this hypothesis. allows for the formation of these orderly repre- tive hippocampal damage was able to impair a touch sentations. Saksida et al. so that in novel situations. if the ated animals were unimpaired in performing a linear subject has learned a set of hierarchical relationships. have been shown to be 3. These oper. Eichenbaum representations. control rats and monkeys required interpretations). The crucial test of this hypothesis is a novel et al. John or Peter?). Variants of purely configural tasks. 1998) or hippocampal damage particular. Although these two stimuli be determined. 1997) are impaired C. monkeys (Alvarado and Bachevalier.09. . (Whishaw and Tomie. (Saunders and Weiskrantz. 1984. neither fornix transection nor selec. Saksida et al. but not memory processes engaged for task performance elemental discriminations. 2004) and Eichenbaum. It has been the cortex in the use or formation of configural or suggested that humans and animals form an orderly relational cues (e. Eichenbaum et al. 2001. AD. damage to the perirhinal in future studies whether the different versions of the cortex (Alvarado and Bachevalier. it assesses a proposed property of declara. (Bussey et al. if John is taller than Bill and Bill is taller than impairment was only for the use of conjunctive Peter. but see Frank et al. Killiany et al. 1998. and colleagues have suggested that the hippocampus By contrast. who is taller. 2005a. Cþ versus X. whereas some 88% of normal the transverse patterning task are not yet clear. monkeys with use of memories in new situations (Cohen. more trials to learn the task and reached lower per- formance levels (performing at 60–70% correct after several thousand trials when the three problems are intermixed) as compared to control rats trained on 3. In 1998. ABþ.09. For example. it should be noted that a number of recent in an object version of the task (performing at 90% studies have implicated the perirhinal.5. CB).3 Role of the Temporal Cortex the swim or digging versions of the task and monkeys Finally. that. Bussey et al. that these relational screen version of the task in either rats or monkeys representations permit the flexible use of memory. all other stimuli are several relational tasks certainly suggests a role for equally often rewarded or nonrewarded..g. 2007) or parahippocampal areas TH/TF in determine the success or failure of control subjects. entorhinal lesions (Buckmaster et al.. 1992). 2006). Brasted et al... CDþ. confirming that the (i. Bþ versus lesions (Dusek and Eichenbaum. and importantly. have never before been paired together. entorhinal. animals and humans can 2007).. methodological differences between the studies animals with hippocampal damage were at chance could be responsible. 1997. The logical structure of the rodents with combined perirhinal and entorhinal task takes the form of Aþ versus B. such that A > B > C > D > and Rudy. Note that. 2005a) impai- Such information will likely reveal the types of red performance on transverse patterning. they findings on the effects of hippocampal damage on have shown that.5. It will be informative to assess tional tasks.g.. 1997). 1998). Buckley and Gaffan.. Bussey and Saksida. but animals correctly choose B over D on the probe..

This game. With respect to configural form the transverse patterning problem (Rickard tasks. it remains to be Astur and Constable. Pihlajamaki et al. 2004). In a study that supports both the relational and Rock crushes. By contrast. Importantly.g. simply repeat. who not only found impaired performance on trans. Scissors cut). One indication of specific the task (Hanlon et al. Memory However. but have suggested that damage to the posterior hippo- also reported that these patients were impaired campus was particularly correlated with impairment on six-pair elemental. 2006). 1995a. these hippocampal involvement is provided by Driscoll same patients were unimpaired and showed normal and colleagues (2003). humans (Heckers et al.. hippocampal activity (which also correlated with Several behavioral studies of human amnesic memory). elaborate processing improved memory Alvarado and Rudy. 2004. however. 2003. damage was not quantified. 2004)..5. Rickard and Grafman (1998) compared (Davachi and Wagner. a role for the hippocampus in relational learning sia on transverse patterning and elemental tasks. Furthermore. these same patients were also unimpaired in elemen. 1992. as well. at the time of encoding. Human studies of transitivity have also suggested or hypoglycemic (insulin overdose)-related amne. or process elaborately by mentally dies of transverse patterning in humans. concurrent discriminations on transverse patterning (Alvarado and Bachevalier. but required twice as many to per- neuroimaging studies. because their trol subjects perform the tasks (Hanlon et al. the chil. 2003). they were required to either it provides an interesting control mechanism for stu. By contrast.. of the triplets over simple repetition.156 Animal Models of Amnesia 3... Performance of relational memory tasks is asso- discriminations. Preston et al. recent reports et al. these ciated with changes in hippocampal activation in patients performed like controls on elemental dis. 2004) and impaired perfor- answered whether damage to the hippocampus or mance in schizophrenic patients who also showed temporal cortical areas played a greater role in the abnormal hippocampal activity while performing observed impairments. 2004). basis of the inherent properties of the stimuli (e. This elaborate processing was thought to encourage tal discriminations or a feature-neutral task that have encoding of the relationships between items.. whereas item-based processing increased patients also reported deficits on transverse patt. Thus. patterning (but not elemental discriminations) dren’s Rock-Paper-Scissors hand game). though present. surprisingly. can be solved verbally as well as on the volumes (Driscoll et al.. Activation in the tem. This report contrasts et al.09. did not differ elaborate processing was associated with greater between controls and patients. criminations and two problems of transverse Titone et al. 2005). which suggested that amnesic patients 2005b). 2004. and in the anterior hippocampus patterning. Preston their performance drop. those The use of conjunctive and relational tasks to tap patients required the same number of trials to cri- human memory processes has flourished in the last terion whether solving three or six elemental few years.g. Only when all three problems were in particular for humans performing the transi- trained and a configural solution was required did tive inference task (Heckers et al. further reference to the other discriminations. ordering them along the lines of subjective desirability.... Not intact hippocampus (Gallagher and Holland. of which transverse pat- on Animal Tests of Conjunctive/Relational terning was a particularly difficult example.. This hypothesis was confirmed with have confirmed hippocampal activation while con- additional amnesic subjects. Davachi and Wagner (2002) be compared at the time of presentation without asked human subjects to memorize triplets for which. activity in the entorhinal and parahippocampal cortices erning. who showed that aged hippocampal activation when performing the human subjects who were impaired on transverse uniquely human version of this task (i. 2002). encoding versus retrieval . when comparing performance per problem. 2004. Paper wraps.. or the also been shown in animals not to depend upon an conjunction of the members as a unique triplet. (2006). spatial relational memory somewhat with another by Reed and Squire (1999) in humans activates posterior hippocampus (e.e. poral cortical areas. as suggested by Rickard et al. showed correspondingly reduced hippocampal however. such as transverse patterning. in particular in the service of functional discriminations. which can conjunctive accounts. performance of four patients with ischemic..4 Performance of Human Amnesics discrimination problems. It is possible that these discrepancies reflect had a more general deficit in solving multiple differences in task demands. and studies in primates verse patterning in human amnesic patients. anoxic.b). Similar to animal results.

be they contextual. tial navigation based on extramaze cues. Olton. or is updated from trial to trial. In this version. in the absence errors than anterograde working memory errors will be of extrahippocampal damage. on a variety of spatial tasks. whereas repeatedly tion for studies of the hippocampus in spatial entering an arm not visited prior to the delay constitu- memory. M. 1983) and somewhat on the spatial reference clearly shown the key role played by the hippocam. Reentry of an arm visited prior to the delay published their landmark work that laid the founda. navigational. 1979.09. 1983). Normal rats learn this task quickly. the radial arm maze was a departure from previous the rat swims in a large round tank full of opaque maze tasks. impairs performance generated (Olton et al. This task differed from most maze cues in the room. 1979. a food fairly accurate direct headings in order to get there. which refers to information that then swims until it locates the platform. and in the case of reference memory. The rat to reference memory. then the rats The final performance test entails the removal of the . This version of the task contrasted with a refer- ence memory version in which four of the arms were The initial research into the role of the hippocampus consistently baited. but not reference memory. In the years following the first description of by reentering an arm that was already visited on that H. A reentry counts as an 3. In this case. In their book (O’Keefe and Nadel. as it was designed to test whether the water. For each trial. decades. If a has been whether or not spatial memory is impaired. and the position of the invisible tasks. which were more concerned with some aspect platform in relationship to those cues. In contrast to the radial maze. although disconnection 3. Olton and Papas. Submerged somewhere in the tank is a platform hippocampus was required for working memory. reward was placed at the end of each arm. as opposed water at one of four cardinal starting points. the swim task developed by Morris (1981) is generally a reference memory task 3. Thus. Indeed. the rat is randomly released into the that changes.. in the first case rodents. 1978) tes an anterograde error (Olton. O’Keefe and Nadel pleted. the litmus test for receiving hippocampal lesions selectively disrupts functional damage to the hippocampal formation working memory. lesion and electrophysiological studies have Olton. onto which the rat can climb to escape from the water. 1983). both working and in spatial memory was conducted for the most part in reference memory errors are possible.. the information was it by the experimenter if it fails to find it.6 Spatial Memory error. by entering a than that in the monkey with respect to finding a link never-baited arm (Olton et al. the rat must form a spatial representation of the specific location. 1979. learning defined. In 1978. the routes are pre. delay imposed after a certain number of choices per- ery of hippocampal place cells by O’Keefe and mits the assessment of retrograde and anterograde Dostrovsky (1976) firmly linked the hippocampus working memory as the remaining choices are com- with spatial mapping. they suggested that the unique contribution of hip. 1979). Animal Models of Amnesia 157 effects. which measures work- ing memory for location.6. Rats with lesions to the hippocampal formation are pocampus to memory was the formation and storage quite impaired on the spatial working memory variants of cognitive maps (See Chapter 1.6. While the radial arm maze does require trial find the quickest trajectory to navigate to the animals to discriminate locations. platform. The addition of a between the hippocampus and memory. rats had to learn not to reenter arms already visited on a given trial to achieve the optimum foraging strategy. constitutes a retrograde error.09. In the following of the task (Olton et al. In this task.