Joke University and University of Toroste,
PANEER, aan
ENCODING SPECIFICITY AND RETRIEVAL
PROCESSES IN EPISODIC MEMORY‘
ENDEL TULVING® ane
DONALD M, THOMSON
Aorist Univeveite, Cloyiou, Edetwrin,
Toronto, Canada
Recent changes in pretheorciloal eriewtatlon tomar problems of hums
Memory have Drought with thes oisceri wilh cetwievsl processes, bed a
mumber of early versions uf theories uf retrieval have heen consteacie This
weaer chacribes an] cvahinies explanations offered by these theories to ac
ccomt-for the eect of extralist cuing, facilitatioe of rerall of list items by om
list items. Experiments designed to test the currently mast popalar theary al
retrieval. the generation-recngnition Uhesry. yielded rewalts incompatible not
ly wits geuerationrecognition meets, tut most utber thevcies as well:
ceria conditions subjects cousissently failed fo recognise sna
recallable ist weeds Seevral tectative explanstion: of this phenomenon of
recognition failure were subvemed under the encoding specificity principle
ancarding to the mewioey teave of an evee asd hence the peopertics
tf effective retricral eur are deterenined by the specific mending operations
cstralia
performed] kyr he system sn the ingat
‘The current transition from. traditional
asanciationism to information processing amd
ongattizational points of view about human
inemory manifests itself in many ways,
of the éleareat signa of change hina to do with
the experimental and theoretical separation
between storage and retrieval processes, Ti
sn important early paper, Malton {1962},
for instance, pointed out that "the princip
issues in theory of memory... are about
either the storage or the retrieval of traces
{p.4]." Only 10 years befare Melton snade
the statement, it would have puzzled most
students of verbal learning, At thas time
memory was still a matter of uceptisition, re
tention, transfer, acd interference of asso»
clationa hetwen stiswli and responses.
White everyone was aware af the logical
clistinction “Letween acquisition and reten-
fiom on the one bated and retertion and recall
on the other hand, these distinctions shaped
2 This research was umported by the Naticaal
Science Foundation Grant 24171X,
‘The paper was written daring the first sathor's
residence a8 a Fellow at the Center for Advanced
Sty in the Behavioral Sciences in Staniont,
Callionia,
*Reoueals fur repriuts shoukd be sent ta Entel
‘Tolri, cliher at Department of Paycluloay, Yale
University, New Havee, Conmectieut 08910 ee at
the Department of Pepchology, Univesity of
Toreeau, Taruste MSS 1A1, Canaea,
linac
neither experiment sor theory. At the bevel
of conceptual analysis, the mechanism of 7e-
call was incladed in the concept af nesocia-
tiow sat the level af experimental operations,
recall waa observable behavior whinge mes-
surable aspects simply served to provide evi-
dence about strengti of associations. More-
ver, the set of recall wns empirically newtral
ii that it diel not alfect the state of the sys-
tem it was theoretically uninteresting Le-
cause it could nat be studied independently
cof acquisition
‘The Inst 10 or 13 yenrs have changed the
itleatinnal framework for studying memory.
‘Today the orienting attitudes clearly include
the notion that bath recall and recognition
are more or lets complex retrieval operations
tor processes that can be studie! amd analyzed
in some sense separately of Morage opera
tions oF processes. Retrieval operations
complete the act of remenibering that kegins
with encoding of infaemation about an ever
into the memary store, Thus, remembering
is regneded a3 4 joint product of inforination
stored in the past and information present
in the immediate cognitive envieonment of
the rememberer, Tt is also becoming in-
creasingly clear that remembering docs nat
involve a mere activation af the learned asuo-
ciation or atausal af the stored trace by a
stimulus, Some sort of 4 mare camplex in-Escoome Srecaricmy ano Revmeva.
1966), ‘The theory is nlso highly compatible
with a well-known principle according to
which probability of recall of an item is a
direct function of the similarity between the
recall situation sand the original learning en-
vironment (e.g,, Hollingworth, 1928; Mel-
ton, 1963),
Encoding Specificity Principle
‘The encoding specificity principle i the
final idea abaut retrieval and extrallat cuing
voffects we disewss. Tn its broadest form the
principle asserts that only that can be re-
‘irieved that haa been stared, and that how it
can be retrieved depends am how it was
stored. In its move restricted senses, the
principle becomes less truistic and hence
theoretically imore interesting, For instance,
we assume that what is stored about the oc-
currence of a word in an experiinetital lst 1s
information about the specific encoding of
that word an that context that situation.
‘This information may or may nat include
the relation that the taryet word has with
some other word in the semantic system
Tf it does, that other word may be an effec-
tive retrieval cue. Tf it does not, the other
word ¢annot provide access ta the stored
information because: its relation ta the tanget
ward is not stored.
‘Thus, the effectiveness of retrieval cues
depends on the properties of the trace af the
word event in the eplondie system, Tt is
independent of the aemmritie properties of
ithe word except insofar as these properties
were enceded as a part of the trace of the
event. The distinction between semantic
characteristics of words ag lexical units and
words as to-be-remembered events can be
readily demonstrated with homographs—for
instance, if viowet fs encocled and sored as
a color name, it normally cannot be re-
triewad as an instance of the estegory of
flowers, or girls” names—but the same prin.
ciple presumably holds for all verbal items.
‘The cue tude facilitates recall of the target
word crare if the original encoding of coat
asa to-hé-renenibered ward inelaced seman
ie Information of the kin’ that defines the
relation between two chjects in the sme
conceptual category, Most intelligent sab-
jects in episodic memory experinents rou-
359
tnely encode to-be-remembered words se-
mandeally, and hence words measlaglilly
related to target items will serve as effective
retrieval cues,
A recent application of the encoding speci-
ficity principle to the interpretation of effec.
tiveness of retrieval cues appeared ina study
by Tulving and Osler (1968); one of its
mare interesting implications was explicitly
tested in three experiments by Thomson and
Tulving (1970); ond its bearing on results
from intralist cuing experiments has been
discussed by Postinan (1972), Since the
prlnekple asserts that it 4a the enceded trace
of the target word rather than the character-
istics of the target word in semantic memory
that determines the effectiveness of extralist
retrieval cues, as well ns all other cues, it
can be experimentally contrasted with thea-
ries that attribute the effectiveness of extra-
list cues to their preexperimental relatinns
with target words, Such contrasts, how-
ever, are possible only under spesial
conditions.
Logie af Experimental Comparison berswern
Theories
‘The main difference between the genera-
ton-recognition models of retrieval and the
encnding specificity principle that is subject
to test lies in the encoding stage of an item's
processing as the locus af the effect nf ences,
Asarding lo the geieration-recognition
modela the encoding stage is not important,
as long as it does not disturh the capacity
of the extralist eve to produce the target item
as un implicit response, According to the
encoding specificity principle, the target Item
must be encoded in some sort of reference
to the cue for the cue to be effective,
Both theories can account equaliy well for
the finding that a given cuc in fact is effec-
tive, Thus, for instance, if table does faeili-
tate the recall of the target word crate, ES ia
possible that an implicit response “chair™
mare tn the cue at retrieval ancl uhaequently
recognized, Tris also possible that the target
CHAIR was semantically encoded at the time
of presentation in a specific way that ren-
dered the cue word ¢able effective. Experi.
ments in which specific encoding conditions