Prcepalnaece,
BEC TSERSIRSS!-a00
Innate and Universal Facial Expressions:
Evidence From Developmental and Cross-Cultural Research
Carroll E. Izard
‘The idea of innate and universal facial expressions that have links with human emotions was
_ziven the status of scientific hypothesis by Darwin (1872/1965). Substantial evidence, old and
‘ew, supports his hypothesis. Much ofthe evidence is independent of language, but Russell's
(1994) criticisms of the hypothesis focus on language dependentdata, Inthisartile itis argued
that Russell's critique was off target in that his arguments relate only to a hypothesis of the
universality of semantic atibutions and overstated in that he used questionable logic in
designing studies tosupporthis claims. tisalsoargued that Russell misinterpreted the relation
between the universality hypothesis and differential emotions theory. Finally, newevidenceis
resented thatsupportsthe Darwinian hypothesis of the innateness and universality ofthe facial,
expressions ofa limited setof emotions and the efficacy ofthe mostcommonly used method of
testingit
‘The purpose of this article isto present new evidence and argu-
‘ments for the innateness and universality ofa limited set of facial
expressions of emotions. | attempt to demonstrate that Russell
(1994) overstated hiscriticismsof the data supporting the universal-
ity hypothesis and did not accurately assess the relevance of the
universality data or his criticisms for differential emotions theory.
[Newevidence from studies of emotion recognition shows the robust-
ness of the forced-choice format for assessing the accuracy of
subjects’ classification of facial expressions intoemotioncategorie.
Although the language-dependent data that Russell (1994) cit
ized may never convince everyone, asubstantial body oflanguage-
independentevidence supporsthehypothesisoftheinnateness and
universality ofat least some of the patternsof facial movements that
signal emotion states in human and nonhuman primates and other
species (Chevalier-Skolnikoft, 1973; Darwin, 1872/1965; Harlow,
1962; Mason, 1985; Sackett, 1966, vanHooff, 1962). Thisevidence
and that from the language-lependent methods are consistent with
the separate and much older popular notion that specificexpressions
‘may be either signs or symbols of specific feelings and intentions
Inthe framework of folk psychology, aparticular expression ike
asmile isa sign when theres sufficient converging information to
indicate that theexpresserishappy orexperiencingjoy.A smile may
bbeonly asymbol, withouta simultaneous underlyingjoy experience,
whenitisasetofcortcally mediated facial movernents used merely
Editor's Note land's rebttal andthe ove by Eksvan that procedes it were
prepared in response to an invitation from the Bulletin to write replies that
‘would appear in the same issue withthe target article by James Ruse. The
Rossel atc inadvertently appeared alone inthe January issue.
“This work was supported by National Science Foundation Grant BND,
s706146
‘Correspondence concerning his article shouldbe addressed to Caroll E.
1uard, Department of Psychology, 220 Wolf Hall, University of Delaware,
Newaik, Delaware 19716-2577
288
as a social grecting or to deceive the observer. However, even the
status of the cortical smile as a symbol is due to the frequent
association ofthe genuine smile with concomitant feelings of joy. I
refer to the popular belief in an association between a specific
‘emotion expression and aspecific corresponding conscious experi-
ence emotion feeling, feltaction tendency, or motivational state) as
the expression-feeling link.
Russell’s (1994) review focused on the question as to whether
“emotions are universally recognized from facial expressions” (p.
103). This use of the term emotion apparently identifies it withthe
‘experiential or conscious component of emotion. I thisis the case,
Russell's question might be more precisely phrased in terms of the
cexpression-feeling link or in terms ofthe question as to whether
specificemotionexpressions are associated with specificconscious
experiences. And this question clearly has two facets. First i there
sufficientregularityin the correspondence between particularemo-
tion expressions and particular conscious experiences or feeling
sates to justify classifying the expressions in terms of the verbal
‘categories that describe the conscious states? In the pages that
follow, I presenta case for an affirmative answer to this question.
‘The second part of the issue of an expression-feeling link is that
if one assumes substantial regularity in correspondence of specific
expressions and specific feelings or genetically influenced connec:
tions amongthem, thencanexpression-feclingrelationsbemodified
through experience, even othe point at which emotion expressions
and emotion feeling states become dissociated? Common observa
tion overeons, as wellas modern empirical research, indicates that
‘sometimes the message ina voluntarily controlled facial expression
‘may be unrelated to the corresponding feeling, ambiguous, or
deliberately deceptive (ef. Ekman, 1994)
There appears to be an evolutionary-biologically based connec:
tion between particular expressive behavior anda particular feeling
Cor motivational state, but this connection can be uncoupled or
‘modified by virtue of thecapacity ofhumanand nonhuman primates
toexercite voluntary contol over innate emotion expressions (lurgens
&Ploog, 1970, 1981). That voluntary control of expressive behavior
isapparentlyafaily recent developmentin phylogeny (P!oog. 1986)INNATE AND UNIVERSAL FACIAL EXPRESSIONS 289
suggests thatthe modifiability of connections between feeling or
‘motivational states and expressions might be indicative of the
‘complexity ofthe species" social life andthe capacity fordeveloping
social-cognitive skills. To have the very sizable repertoire of self-
revealing signals of higher primates without thecapacity oregulate
them would seem a highly maladaptive, if not fatal, tum of nature.
Undoubtedly, natural selection saved us from this awkward fate
Ancient Origins of the Expression—Feeling Link
Despite the imperfectrcliability and modifiability ofexpression—
feeling relations, the need for veridicality of expressions as an
indicator of feelings and intentions in situations critical to social
relationships, adaptive behavior, and survival Andrew, 1963) sug-
‘gests that the idea ofan expression-feeling link was a precursor of
the formal hypothesis of the innateness and universality of certain
facial expressions. There is, as Russell (1994) observed, a widely
beldandapparently pancultual beliefin the expression-fecling link:
the power of the face to reveal feelings, intentions, and even
character or personality.
Russel (1994) traced the idea of an expression-feeling link to
‘Aristotle, bt itis much older than that. There are clear indications
‘ofageneral acceptance of theideain Homer and in the Bible Perhaps
itisabiosocial givenoraffordance(. Gibson, 1966;Neisser, 1976).
(One ofthe prophets quoted God as saying that “My fury shall come
vp in my face” (Ezekiel 39:18). And Homer depicted belief in the
expression-feeling link poignantly in an episode in which two of
‘Agamemnon’ ides had t confrontthe mighty warrior, Akhilleus:
“Shamefaced, pale with fear of him, they stood without a word; but
the knew what they felt" (Homer, Iliad Book One, p. 22).
Nooneknows who made thefirstattribution aboutthe meaning of
‘facial expression. Chances are good thatthe atribution was made
‘automaticaly, silently, and without benefitoflanguageby one of our
primate ancestorsinanefforttoadaptand survive cf. Andrew, 1963,
1965), Ethologists and primatologists generally agree that at least
some of these simian signals were emotion specific and not just
positively or negatively valenced (Chevalier Skolnikoff, 1973;
Hauser, 1993; Mason, 1985),
‘The Vicissitudes of Human Facial Expressions
Despite the testimony of venerable observers andtheevidence and
arguments of ethologists and evolutionists, doubt and controversy
have probably always challenged the reliability of the expression—
feeling link. Inmustering hismen forbatle, Shakespeare's Henry V
said, “Disguise fair nature with hard favout'drage.... then end the
eyeaterribleaspect... let the browo'erwhelmit“ (Henry V, Act!
Scene IV). But Shakespeare's Duncan, afte learning thatthe Thane
‘of Cawdor, a longtime friend, had treasonously deceived him, said,
“Thereisno arttofind the mind'sconstructionin theface” (Macbeth,
‘Act I, Scene IV),
Few doubt that people observe facial expressions and habitually
make inferences about the expresser’s feelings and intentions, but
thedebate aboutthe veridicality of messagesin the face and concem
‘with the noise or unreliability inemotion communication arequite
likely aboutasoldas the practice of makinginferences.'Therecanbe
litle doubt that there is degree of inaccuracy in interpreting facial
expressions in daily life and in the laboratory. As already noted,
evolution and adaptation required that human beings be genetically
endowed with the capacity w selfregulate, manipulate, and alter
expression-feelin relations in keeping with the perceived contin-
sgencies of veridcal expression in socal interactions. Ths may be
ttuc of nonhuman primates as well, forithas been shown tha they
have somecortical eontroloverinnate vocal expressions ofemotons
(lrgens & Ploog, 1970; loog, 1986).
Inaddition, muchof thenoiseinemotion communication through
facial expression can reasonably beattributedtocivilizationandits
constrains, Socialization processes and social learning aimed at
controlling and modifying emotion expressions and expression
feelingselations begininearly infancy and have measurable effects
by 4 to 6 months of age (Camras, Oster, Campos, Miyake, &
Bradshaw, 1992;Malatesta& Haviland, 1982). Yet, despitethe fact
thatexpresions can be voluntarily managed to conceal subjective
feeling states or simply for purposes of communicating with or
influencing conspecifics, theideaof anexpression-Feeling ink lives
on. Though not held in blind faith, it remains a common belief
becauseitis generally consistent with what actually happensinmany
socialexchanges.
Darwin and the Idea of an Expression—Feeling Link
Russelland Ihave different perspectives on thehistory of theidea
of an expression-feeling link and the universality hypothesis. T
disagree with his implication that Darwin was justanother believer.
‘The idea ofan expression—feeling link was for many millennia just
that, an idea or folk belief abstracted from common observation or
zeligious doctrine, Pre-Darwin scientists, whom evolutionists now
describe as creationists, studied facial expressions in awe of what
God had wrought (ef. Russell, 1994)
Darwin brought anentirely different perspective. Hisideasabout
knowledge through experience and observation, hypothesis formu
Jation, and hypothesis testing area lot like what we hearin modern
‘courses on scientific method. Whereas evidence against an expres-
sion-feeling link might have been seen by Bell (1806/1966) and
Duchenne de Boulogne (1876/1990) as heresy, it would have been
seen by Darwin as data calling for a change in postion
"have steadily endeavored to keep my mind fee, 808 give up any
hypothesis, however beloved (and I cannot resist forming one onevery
subject). assoonasfactsareshowntobe opposed tot. Indeed Thave had
‘0 choice but to acti this manner, for with the exception ofthe Coral
Reefs, cannot remember a singe fist formed hypothesis that had nat
after sine tobe given up or greatly modified. (Darwin, 1887/1958, p
141)
‘The central thesis ofthis artcleis that the expression-feelingidea
of Homer, Ezekiel, Aristotle, and Shakespeare has become more
tan a robust folk belief. I was given the status of a scientific
hypothesis by Darwin, both by virtue of his attitude (willingness to
‘modify it in the face of contrary evidence) and by his observations
that placed human facial expressions in evolutionary-biological
perspective. Darwin's (M Notebook, 1838/1974; 1872/1965) ex-
planation ofthe origin of discrete emotions and ther expressionsis
‘consistent withthe contemporary conceptof modularity (the decom-
posability of the apparent wholeness of complex systems into
relatively independent and dissociable units or modules). Gould
(1993) defended Darwin's use of the principle of modularity as an
‘explanatory concept for discrete emotions and argued forts appli-
‘ability tothe domains of human talent and cognition.290 ‘CARROLL B. IZARD
Darwin's treatise on emotions, The Expression of Emotions in
Man and Animal (1872/1965), 33 yearsin the making, inspired not
only the research on facial expressions now being debated, bt work
fon the neuroanatomy and the biological and social functions of
‘emotions (e.., Izard, 1989; Kling, 1986; LeDoux, 1987; Lorenz,
1965; Panksepp, 1986). So the culture-common idea that specific
‘expressionsareassociated with specificemotionstates took the form
of the scientific hypothesis of the innateness and universality of
‘emotion expressions inthe hands of Darwin, He was keenly aware
‘ofthe work ofhis predecessors ike Bell (1806/1966) and Duchenne
de Boulogne (1876/1990), and in his early years he was an avid
reader of Shakespeare. Yet he was able to add the perspective and
insight that changed a folk belief into a testable hypothesis.
Russell's (1994) critique did not directly address the innateness—
hypothesis (TUH). He focused on an aspect ofthe folk
belief about expression-feeling relations or subjects’ ability 10
associate verbal representations of subjective emotion states with
‘posed facial expressions as measured by emotion-recognition and
‘emotion-labeling techniques. Both of these techniques of assessing
emotion knowledge are language dependent, the former depending
‘on recognition memory and the latter on recall memory. Russell's
(1994) critique is about the universality of semantic attributions to
facial expressions, or the semantic attribution hypothesis. Although
‘much of the data that relate in any way to the IUH came from
‘emotion-recognition and emotion-labeling studies, there are, as
already noted, other mportanttypesof supporting data (see Ekman,
1994)
Russell's (1994) critique of thecross-culturalemotion recognition
and emotion- labeling research and some new data of his own seems
tosuggestthatitis time for us staunch supporters of theTUHto show
‘our Darwinian colors and modify the hypothesis (See Ekman, 1994,
forcommentsregarding the accuracy of Russll’streatmentof other
scholars who have studied facial expression.) That may be the case,
but believe that what Russell's review requires us tomodify ae the
impressions and inferences that have heen drawn fromthe research
relating to the IU. Note that data supporting the semantic attribu-
tion hypothesis are only one of the kinds of evidence that can be
‘marshalled in support ofthe TUH.
Universality Hypotheses and Emotion Theory
Before developing the empirically based arguments, I present a,
few comments in response to Russell's (1994) statements about the
relation of the universality hypotheses to my differential emotions
theory (DET). The universality ofcertainexpression-feelinglinks—
‘or, more pertinent to Russell's review. the equivalence of emation
labels across cultures asindexed inadultsubjecs through language-
dependent techniques-—may have litte to do with DET's central
thesis. The overarching principle of DET is that emotions (which
consist of neural, expressive, and experiential components) are
inherently adaptive and that each discrete emotion has unique
‘organizational and motivational properties (Izard, 1989, 1991, 1992)
Clearly, however, evidence for natural connections between specific
expressions and specific motivational-feeling states and evidence
for the IUH strengthen the biosocial-developmental emphasis in
DET.
Russell (1994) quoted the following statement: “Emotion at one
Jevelisneuromuscularactivity inthe face” (izard,1971,p.188).One
ceaninger from this statementthatemotionexperience (motivational
feeling state) is dependent on facial expression or, atleast, that
expression is alwaysan integral part of emotion, 'm not sure that |
ever meant for that inference to be made without qualification. At
any rate, DET has been modified a numberof times since its frst
formulation, and ater versionsallow for thecomplete independence
ofexpression and feling (Izard, 1977). Current DET views expres-
sive-behavior feedback as only one of four different systems for
‘emotion activation (Lard, 1993),
Idonotmean to denigrate the signficanceof theidea of universal
facial expressionsin the original formalationof DET orinitscurrent
hypotheses relating to expressive-behavior development and the
‘development of interpersonal relationsin the prelingvalinfant. Yet,
‘evenin these cases, the idea of innateness and universality of facial
cexpressionsininfancy (asa function of ourevolutionary-biological
heritage) is more important to DET than is confirmation of the
semantic attribution hypothesis with adult subjects who have had
‘many yearstolearnto inhibitand modify facial behaviorstosuittheir
individual personalities and their culture.
PPechaps these thoughts help explain why my cross-cultural re-
search focused on demonstrating the human capacity to recognize
facial expressions of emotions and not on the veridicality and
"universality of spontaneous expressions in daly fe. Ibelieve there
is veridicality in spontaneous expressions when they are free of
personal and cultural bias regarding thei display. Inthe following
ppages, I argue that there is universality in recognition of facial
expressions when the expressions meet certain criteria, These crite
riaare difficultto state in purely theoretical terms, butthey relate to
the communicative functions of expressions inevolution and adap-
tation. However, even ifthe criteria were absolutely clear and all
‘expression stimuli metthem exactly, thereare several reasons notto
‘expect perfect correspondence among emotion words and emotion
expressions (Izard, 1971),
‘The Ecological Validity of Facial Expressions
Russell (1994) stated that ultimately the question ofthe universal-
ity of emotion expressions is about what happens in natural situa-
tions. One could put this the other way around. What happened in
natural situations (Darwinian selection processes) brought about the
innateness and universality of facial expressions. Ifexpressions are
anything, they area functional communication system that preceded
language in evolution and that precedes language in ontogeny.
Arguably, twasthe relatively specific messages ofdiscreteemotion
expressions that facilitated the evolution of primate social life.
Hamburg (1963) maintained that emotions were critical in the
evolutionary process, serving many important functions, some of
which scem tobe dependent on emotion-specific communications
‘Andrew (1963) argued that primate facial expressions evolved
because they were effectivein social communication, Mason(1985)
concluded thatin rhesus monkeys, species-fypical motoric patterns
of facial expressions are innate relatively impervious to environ-
‘mental influences, elictedby certain stimulus situations onthe same
predictable schedule in all individuals (despite preat variations in
rearing conditions), and stable overtime in signal value, The signal
value holdsevenafterthe monkeyshaveleamed to simulate expres-
sions to manipulate social events.
Inhumanevolution, effective communication situations where
adaptive response or even survival was at stake would seem tohave
called foremotion-specific communication and not merely expresINNATE AND UNIVERSAL FACIAL EXPRESSIONS 291
sive behavior consistent with broad dimensions of experience such
as valence and arousal. To see the point more clearly, imagine an
adult and juvenile trampling through the savannah when the leading.
foot of the adult lands adjacent toa stimulus thatshould be avoided:
afoul-smelling, deteriorating carcass. Thishappens again, buton the
second occasion the stimulus is a deadly viper. A display on both
‘occasions that was consistent only with undifferentiated negative
arousal would provide the railing juvenile with itl information for
Tearningto discriminateadisgusting encounter fromaterrifyingand
deadly encounter. Surely, the speed and repertoire of behavioral
responses in the two situations need tobe different tobe adaptive
‘Although all developmental psychologists do not agree on the
time of emergence of discrete facial expressions, few doubt their
significance in socioemotional development. Ithasbeen well docu-
mented that discrete facial expressions of human mothers have
predictable and differential effects inregulatingtheiinfants’ behav
jor in happy circumstances and in what can be perceived as @
‘dangerous situation (Malatesta & Haviland, 1982; Sorce, Emde,
‘ampos, & Klinnert, 1985). Nevertheless, Russell (1994)is correct
innoting the scarcity ofthe ecological validity of spontaneous adult
expressions
Conceptual Components of the IUH.
‘The IUH has two distinct components: inateness and universa-
ity. tis conceivable that a facial expression could be innate but not
displayed in adults ofall cultures. This could be explained in terms
of cultural differences in socialization and in genetically based
individual differences inthe neural substrates that serve expression
encoding. Conversely, an expression mightbeuniversalnot because
itisinnate,butbecause of culture-constantlearing. Furthermore, an
expression might be inaate and universal but show individual and
coltral variability inthe semantic meaning attributed tot. Russell
(1994) cite data on facial exprssionsin the bom bind, the ontog-
eny of expressions inearly infancy, and comparisonsof expressions
in different cultures and granted that some patterns of facial move-
tment are universal, I guess that he might also agree that these
universal expressions arealso innate
Treview additonal evidence forthe innateness of facil expros-
sions because, ike Darwin, believe thatthe universality ofexpres-
sionsisrootedintheirinnateness. However, Russell (194) demon-
strated thatitis possible wo disregard the inatenessissue and grant
that there are some universal expressions and still raise critical
«questions about the semantic atsibution hypathesis—theuniversal-
ityofcorrectoreross-culturally reliable atributionsofthesemantc
‘meaning of expressions. The thrust of Russell's (1994) argumeatis
that we have not established the hypothesis of universal semantic
anuibutionsasitrlatestodisereteemotion expressions particularly
certainnegativeemotion expressions. Heddnotaddress the UH ot
the supporting data from prelingual infants derived through such
language-independent methodsas the anatomically based objective
coding ystems (eg .Ekiman & Fiesen, 1978 Izard, 1979; Oster &
Rosenstein, in press)
{think that methods dependent on language and semanticattribu-
tion are not the best way to establish the IUH. Objective coding
systems to establish onfogenies of facial expressions inthe born
blind, cross-cultural comparisons of the ontogeny of expressions
through naturalistic observation, anderost-culturaleomparisons of
expressive responses to standardized eliciting events are examples
of stronger methods. However, Russell's (1994) review focused on
the research based on the language-dependent methods relevant to
the semantic attribution hypothesis, because it was largely the
impact of this research that put the idea ofthe universality of facial
expressions intocontemporary psychology textbooks.
Overview of Arguments for the 1UH.
Despite the foregoing distinction between the semanticatribution
hypothesis and the IUH, my first argument bears onthe innateness
ccomponentof the IUH, because Ithink iti of importance inits own
‘ight independentof the semanticissues relating toemotion-tecog-
nition and emotion-labeling techniques and the semantic attribution
hypothesis, For example, firm knowledge of the innateness and
universality of a set of facia expressions has far-reaching implica
tions for cross-cultural research on socialization and its effects on
normal and abnormal development (Izard & Harris, in press; ard &
Malatesta, 1987). Thesupporting data for this argumentareindepen-
dentoflanguage-dependent methods. The gistof theargumentisthat
evidence on the ontogeny of facial expressionsineatly development
supportsthe notion that certain facial expressionsareinnateand that
their innateness links them to the human evolutionary-biological
heritage and hence to universality
My second argument is that there is not much weong with what
Russell (1994) identified as the forced-choice format that 1 and
others use for assessing subjects" ability to classify facial expres-
sions into emotion categories. Given thatithas some ofthe inherent
‘weaknesses of any language-dependent method, the results were
rather robust support for the IUH (ef. Ekman, 1994). My third
argumentisthat fixing the forced-choice formatby adoptingsomeot
‘Russell's (1994) suggestions doesnot change profoundly the kindof
results reported in my early eross-cultural research (Izard, 1971),
Facial Expressions in Early Infancy
Because no one denies that some discrete emotion expressions
existin adulthood, evidence ofthe existence of these same expres-
sions in early infancy would be consistent with the IUH. Evidence
that young infants can discriminate discrete emotion expressions
‘would also support the IUH. In presenting data relating to discrete
‘emotion expressions in early (preverbal) infancy, Tam aware that
such evidence can support the IUH without having a direct bearing
(on the semantic attribution hypothesis. The encoding, and even
aspects of the decoding, of expressions is quite different from the
‘business of matching words with expressionsorthe taskof verbally
labeling expressions.
The Ontogeny of Facial Expression Encoding
‘Weexamined the ontogeny offacialexpressions inapproximately
100 infants, who were observed during mother-infantinteractionsat
ages2.5,3.0,4.5,6,9.0,and9.5 months (Izard, Haynes, Fantauzz0,
Slomine, & Castle, 1993). There weretwotypesofconditions. Inthe
positive conditions, mothers playfully expressed interest and joy
facially and vocally. In the negative conditions, mother assumed a
still and silent face or expressed sadness or anger facially and
vocally, High reliability was obtainedin coding theinfants’expres-
sions with Max Izard, 1979) and Affex (ard, Hembree, Dougherty,
& Spizzirsi, 1983),292
By 2.5 months, infants encoded full-face and partial expressions,
of interest, joy, sadness, and anger with sufficient frequency of
statistical analysis. The basic morphology of these expressions
(pattems of muscle movements and appearance changes) remained
stable through the first 9 months. Furthermore, the same Max-
‘dentiiedpatternsof fecal movements presentedin theadultexpres-
sions used in my original cross-cultural research (Izard, 1971).
Although these movements were too infrequent for formal analysis,
the infants also displayed surprise, disgust, and fear with the same
facial movements that characterize these expressions in adults. We
also demonstrated that these infant emotion expressions were reli-
ably recognized by adults (see also Emde, Gaensbauer, & Harmon,
1976; Ieard, Hucbner, Risser, McGinnes, & Dougherty, 1980; and
see Oster, Hegley, & Nagel, 1992, foradifferent point of view),
‘As Russell (1994) noted, some researchers have challenged the
notion that Max-specified expressionsin very young infants younger
than 2 or 3 months) are accompanied by corresponding emotion-
specifiemental (feeling-motivational state (Camras,1992;Matias
& Cohn, 1993; Osteretal., 1992). However, the expression-feeling
congruence hypothesis (Izard & Malatesta, 1987), as it relates 10
carly infancy, isa separate issue from either the IUH or the idea of
an expression-feeting link in older children and adults
Even investigators who disagree with the idea that thereisexpres-
sion-feeling congruence in very young infants have obtained evi-
‘dence of the identity of emotion expressions in infants of widely
ifferent cultures. During an experimental procedure that included
armrestraint, American and Japanese 5-mmonth-old infants produced
‘anatomically identical Max-specified expressions of anger most
frequently, but also enjoyment, fear. and sad-anger and sad—fear
blends (Camras etal, 1992). Camrasetal. interpreted the observed
infant expressions as variants of distress or ery faces, yet they
concluded thattheir cross-cultural evidence of young infantsencod.
ing identical expressionsin response to the same event was support
for tho TUK.
The Ontogeny of Emotion Expression Decoding
‘Sackett (1966) reared rhesus monkeys in complete socal iso
tion, and from ther Ist weeks oflife, he exposed them to photographs
(rear-projected slides) of juvenile and adult monkeys. The photo-
graphs included facial expressions ranging from the play face
(comparableto the humaninterestexpression) to the threat and fear
faces (which have components similar to the human anger and fear
expressions). Primatologists agree on the meaning of these expres-
sions in monkeys (Chevalier-Skolnikoff, 1973; vanHiooff, 1967).
‘Theisolate-eared monkeys responded differentially and appropri
ately to other monkeys’ expressions. They did not make their first
appropriate response to friendly and threatening expressions at the
same age, butona schedule similar to thatof the maturation of what
Harlow (1971) called the affectional, aggressive and fear systems.
They responded to friendly expressions with play faces well before
they responded with ear faces and avoidant behaviors tothe threat
expression,
‘Human infants are characterized by an innate preference for
looking atthe face whenitis in competition withotherstimuli (Fante,
1963, 1965, 1966; Stechler, 1964). Infants 2 months of ageshowed
‘more interest expressions and attention (looking time) to a live
hhuman face than to the face of a mannequin or an inanimate object
wih scrambled facial features(Langsdorf, ard, Rayias, &Hembree,
‘CARROLL E, IZARD
1983), Reviews of the literature on the development of emotion
‘recognition inhuman infants agree that infants iscriminatebetween
positive and negative expressions by about 3 months and among
‘negative expressionsby 6or7 months e.g.Nelson, 1987),Ludemann
and Nelson (1988) demonstrated that infants of 7 months could
represent oy and surpriseexpressions categorically, but Ludemann
and Nelson failed to show thatthe infants could discriminate these
expressions from the fear expression. However, Serrano, Iglesias,
and Locches (1992) used a habituation-recovery procedure and
‘concluded thatinfantsof4to 6 months discriminated andrecognized
anger, fear, and surprise expressions. Each type of expression was
portrayed by several different models, and discrimination general-
ined across models.
The foregoing evidenceisstronger support for the IUH than forthe
semanticattribution hypothesis, Infants’ bility todiscriminate and
categorize facial expressions does nottellus when orhow they come
{oderiveinformation fromanexpression that could motivate behav
jor appropriate to that expression. Such discrimination and catego-
‘ization certainly tel us nothing about the derivation of semantic
information fromexpressions. Eventhe categorization andrepresen-
tation of some types of expressions may require some amount of
visual experience with those expressions (cf. Ludemann & Nelson,
1988),and surely the derivationofsemanticinformationrequiresthe
acquisition of receptive vocabulary and perhaps extensive social
learning as well
Monkey and human infants of only a few months of age can tell
‘usalotaboutthe recognition and categorical representation of facial
expressions, but they cannot tell usanything about matching expres-
sions with names of emotion categories (emotion recognition) orthe
free-response labeling of expressions (emotion labeling). Ipropose
that individual and cultural differences inthe basic processes that
lead to emotion-recognition and emotion-labeling abilities help
explain why emotion labeling scoresare always ower than recogni-
tion scores by a substantial margin and why recognition scores can
vary with the manipulation of the sort of variables identified by
Russell (1994), There are anumber of important questions here that
require more researc,
How Bad Is the Forced-Choice Format?
Russell's (1994) review was critical of the method most fre-
‘quently usedin the early cross-culturalemotion-recognition studies.
‘The principal features ofthe procedure consist of expression photo-
‘graphs and emotion categories selected by a priori criteria and a
‘multiple-choice response format in which the number of response
options is determined by the number of hypothesized emotion
categories. Russel called this the forced-choice format and argued
that it had several characteristics, which, taken together, might
influence responses.n favor ofthe semantic attsibution hypothesis,
and, indirectly, the TUHL
However, the forced-choice format did notnecessarily lead toan
overestimation of emotion-recognition accuracy, depending on the
4efinitionof recognition accuracy. As Russell (1994) observed, the
goal of my cross-cultural research was to tet the hypothesis that
people of different cultures had the capacity to identify aparticular
set of facial expressions in terms of a particular set of a priori
categories. intended this research to provide evidence for the UH,
Dut the evidence is actually more directly related to the semantic.
attribution hypothesis, because the methods were dependent onINNATE AND UNIVERSAL FACIAL EXPRESSIONS 293
language and meaning attribution. Itwas because lsoughtevidence
(of people's capacity to identify a limited set of expressions that 1
‘selected emotion categories and expression exemplars by certain
criteria. This is quite different from the very important question of
theaccuracy with which people attribute meaningto thespontaneous
expressions of daily life.
“The selection of expressions and emotion categories (emotion-
recognitionresponse options) was guidedby existingempirical data
‘as wel as the theories of Darwin (1872/1965) and Tomkins (1962,
1963) and DET. The idea was that an evolutionary-biosocial ap-
proach to understanding emotions suggested that limited number
of facial expressions could be given corresponding categorical
rubrics and that subjects in different cultures could match expres-
sions and categories.
Tthink the forced-choice format, perhaps with some modification,
is an acceptable method for estimating adult subjects” capacity 10
classify facial expressions intoemotioncategories. Thisisespecially
soifthereissome theory to guide the selection of emotion categories
orlabels and the expressions that exemplify them. Darwin's (1872/
1965) theory provides the general principle that in evolution the
expressions served communicative functions and motivated adap-
tive behavior in observing conspecifics (cf. Andrew, 1963).
“The forced-choice format, in keeping withthe Darwinian tradi-
tion, assumes thatemotionscanbeadequatcly described or classified
interms of categories. Inthe domainofexpressions,asctofemotion
categories is tantamount tothe idea ofa limited set of discrete and