Feeling and ThinkingPreferences Need No Inferences

R. B. ZAJONC University of Michigan

ABSTRACT: Affect is considered by most contempo-rary theories to be postcognitive, that is, to occur onlyafter considerable cognitive operations have been ac-complished. Yet a number of experimental results onpreferences, attitudes, impression formation, and de-_cision making, as well as some clinical phenomena,suggest that affective judgments may be fairly inde-pendent of, and precede in time, the sorts of percep-tual and cognitive operations commonly assumed to bethe basis of these affective judgments. Affective re-actions to stimuli are often the very first reactions ofthe organism, and for lower organisms they are thedominant reactions. Affective reactions can occurwithout extensive perceptual and cognitive encoding,are made with greater confidence than cognitive judg-ments, and can be made sooner. Experimental evi-dence is presented demonstrating that reliable affec-tive discriminations (like-dislike ratings) can be madein the total absence of recognition memory (old-newjudgments). Various differences between judgmentsbased on affect and those based on perceptual andcognitive processes are examined. It is concludedthat affect and cognition are under the control of sepa-rate and partially independent systems that can influ-ence each other in a variety of ways, and that bothconstitute independent sources of effects in informationprocessing.

The intellectual contact between psychology andpoetry is scarce and, when it takes place, oftentends to be exploitative. If we happen to comeacross a poem that appears to support one of ourfavorite generalizations, we are tempted to citeit (not as evidence, of course, but more in theform of a testimonial). Or we might confer uponit the status of an epigraph in one of our forth-coming chapters (commonly, to the detriment ofboth the poem and the chapter), But when poetrydisagrees with us we are apt to ignore the conflictaltogether. Nevertheless, this paper begins witha poem by E. E. Cummings (1973), the firststanza of which affirms a premise tacitly rejectedby psychology many decades ago:

Vol. 35, No. 2, 1S1-17S

since feeling is firstwho pays any attentionto the syntax of thingswill never wholly kiss you (p. 160)

In it, Cummings takes for granted that feelingsare primary and, by implication, that they arefundamental. They are precedent to the intellec-tive qualities and elements of experience, and theyare nearer to its essence: They are nearer to aninner "truth."

In contrast, contemporary psychology regardsfeelings as last. Affect is postcognitive. It iselicited only after considerable processing of in-formation has been accomplished (see Figure 1).An affective reaction, such as liking, disliking, pref-erence, evaluation, or the experience of pleasureor displeasure, is based on a prior cognitive pro-cess in which a variety of content discriminationsare made and features are identified, examined fortheir value, and weighted for their contributions.Once this analytic task has been completed, acomputation of the components can generate anoverall affective judgment. Before I can likesomething I must have some knowledge about it,and in the very least, I must have identified someof its discriminant features. Objects must becognized before they can be evaluated,

Most of us will not be deeply distressed by dis-covering that our current theories are in conflict

This article was the Distinguished Scientific ContributionAward address given at the meeting of the AmericanPsychological Association, New York, New York, Sep-tember 2, 1979. It was prepared with the support of aJohn Simon Guggenheim Fellowship.

I benefited greatly by discussing these ideas with severalpeople, and I am very indebted to them. I am especiallygrateful to Hazel Markus, Phoebe Ellsworth, Allan Paivio,and Robyn Dawes, who all made extensive and helpfulcomments on an earlier draft.

Requests for reprints should be sent to R. B. Zajonc,Research Center for Group Dynamics, University of Michi-gan, Ann Arbor, Michigan 48106.

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 1S1Copyright 1980 by the American Psychological Association, Inc.

0003 -066X/80/3 S02-0151 $00.7 5

with a controversial poet of the 1920s. But con-temporary psychology not only contradicts Cum-mings, it also contradicts one of its very ownfounding fathers. Thirty years before Cummingspublished his poem on feelings, Wundt (1907)wrote in a similar vein:

When any physical process rises above the threshold ofconsciousness, it is the affective elements which as soonas they are strong enough, first become noticeable. Theybegin to force themselves energetically into the fixationpoint of consciousness before anything is perceived of theideational elements^. . . . They are sometimes states ofpleasurable or unpleasurable character, sometimes theyare predominantly states of strained expectation. . . .Often there is vividly present . . . the special affectivetone of the forgotten idea, although the idea itself stillremains in the background of consciousness. . . . In asimilar manner . . . the clear apperception of ideas inacts of cognition and recognition is always preceded byfeelings, (pp. 243-244)

Whatever happened to Wundt's affective pri-macy idea? Is there compelling evidence to rejectit? - Or to accept it, for that matter? Strictlyspeaking, we have no better evidence today thanWundt had in 1896. Perhaps a bit better.

In part, my concern in this paper is withWundt's assertion. More specifically, building onthe scanty evidence we now have, I have tried todevelop some notions about the possible ways inwhich affect is processed as part of experience andhave attempted to distinguish affect from process-ing of information that does not have affectivequalities. This article is confined to those aspectsof affect and feeling that are generally involved in

1 The italics are mine. The original is even more tothe point. "Affective elements" were "Gefuhlselemente,"and the italicized part of the citation was "ehe noch vonden Vorstellungselementen irgend etwas wahrgenommcnwird" (Wundt, 1905, p. 262) .

2 It is a fact that only 12 years after the first editionof Wundt's Grundriss was published, Nakashima (1909a;1909b) tested Wundt's assertion by collecting reactiontimes of psychophysical (pitch, hue, temperature, etc.) andaffective (preference) judgments made on the same setsof stimuli. He did not find shorter reaction times forjudgments of preference than for judgments of pitch, hue,temperature, etc., and thus disagreed with Wundt withregard to the primacy of feelings. But his study alonecould not have buried Wundt's idea. Actually, Naka-shima's data were rather inconclusive, since he failed tocontrol for levels of discriminability associated with thetwo types of judgments. Thus, for example, subjects candetect very small differences in hue yet feel quite indif-ferent in their preference for stimuli that differ so little.Since reaction times for comparisons vary with the sizeof the difference, these times can be compared meaning-fully only if the stimuli are preselected so that differencethresholds for the two types of judgments are the same.

preferences. These aspects are reflected in theanswers to such questions as "Do you like thisperson?" "How do you feel about capital punish-ment?" "Which do you prefer, Brie or Camem-bert?" "Are you pleased with the review yourrecent book received?" In short, I deal with somehot cognitions (as Abelson [1963] christenedthem) and try to distinguish them from the coldones. The class of feelings considered here is thatinvolved in the general quality of behavior thatunderlies the approach-avoidance distinction.Thus, for the present purposes, other emotionssuch as surprise, anger, guilt, or shame, whichhave been identified in the literature and exten-sively analyzed by Tomkins (1962, 1963), Izard(1977), and others, are ignored.

Unlike experimental psychologists,3 social psy-chologists are deeply concerned with affect andwith hot cognitions. The extensive work on atti-tudes, research on cognitive dissonance and cog-

3 Contemporary cognitive psychology simply ignores af-fect. The words affect, attitude, emotion, feeling, andsentiment do not appear in the indexes of any of themajor works on cognition (Anderson, 1976; Anderson &Bower, 1973; Bobrow & Collins, 1975; Crowder, 1976;Kintsch, 1974; Lachman, Lachman, & Butterfield, 1979;Norman & Rumelhart, 1975; Schank & Abelson, 1977;Tulving & Donaldson, 1972). Nor do these conceptsappear in Neisser's (1967) original work that gave rise tothe cognitive revolution in experimental psychology. Andin the six volumes and the 2,133 pages of the Handbookof Learning and Cognitive Processes (Estes, 1975-1978),there is only one entry for affect and only one for attitude.It is worth noting that both of these entries are in Volume3 in a contribution written by a social psychologist. Inthe last three volumes—those principally devoted to cog-nition—there are no references to affect whatsoever.

The notable exceptions are Handler's (1975) work onthought and emotion, Neisser's 1976 essay, and Miller andJohnson-Laird's (1976) recent volume on language andperception from which the following revealing quotationis taken:

The information-processing system that emerges fromthese remarks is fearfully cognitive and dispassionate.It can collect information, remember it, and worktoward objectives, but it would have no emotional re-action to what is collected, remembered, or achieved.Since in this respect it is a poor model of a person, weshould add at least one more predicate to this list ofthose that take "person" as their first argument. Wewill use Feel (person, x) to indicate that people havefeelings as well as perceptions, memories, and intentions.It might be possible to subsume Feel under Perceive onthe grounds that our feelings are a special class of per-ception of inner states. Or we might discuss feelingsunder Remember, the recognition that some word orobject is familiar, is after all, a matter of feeling acertain way about it. Or, since we have already jecog-nized that there is a strong affective component to ourintentions, we might link Feel to Intend. . . . All these

1S2 • FEBRUARY 1980 • AMERICAN PSYCHOLOGIST

PHYSICALENCODING

HIGHERORDER

ENCODING

COGNITIVEREPRESENTATIONS

OF STIMULUSINFORMATION

AFFECTIVEREACTION

I1 p+(JUDGMENTJ

Figure 1. Typical information-processing model of affect.

nitive balance, the Schachter and Singer (1962)studies on emotion, and Heider's (1958) attemptsto describe the cognitive representation of affectthat characterizes interpersonal relationships areall clear manifestations of this concern.4 Thereare practically no social phenomena that do notimplicate affect in some important way. Affectdominates social interaction, and it is the majorcurrency in which social intercourse is transacted.The vast majority of our daily conversations entailthe exchange of information about our opinions,preferences, and evaluations. And affect in theseconversations is transmitted not only by the verbalchannel but by nonverbal cues as well-—cues thatmay, in fact, carry the principal components of

considerations testify to the systematic importance ofthis psychological predicate. Nevertheless, we will havelittle to say about Feel in the following pages, (pp.111-112)Thus, Miller and Johnson-Laird explicitly acknowledge

the significance of feelings as part of experience, yet theydecide to devote minimal attention to them. Their de-cision is noteworthy in the light of their belief that "Feelis an indispensable predicate for any complete psychologyand that it probably lies much closer than Perceive, Re-member, and Intend to the basic sources of energy thatkeep the whole system running" (p. 112).

Beyond these volumes there are some isolated theoreti-cal attempts directed toward the understanding of therole of motivational and emotional factors in perceptionand cognition (Broadbent, 1977; Erdelyt, 1974; Posner &Snyder, 197Sa).

4 While such studies as those of Byrne (1961), Berscheidand Walster (1978), or Rubin (1973), which deal withinterpersonal attraction, also have a concern with-affect,they do not contain specific analyses of how affect isrepresented as part of experience. And in studies thatcompare the effects of conditions that differ on the affec-tive dimension (such as self- vs. nonself-relevance, ego-involvement), it is generally not the affective quality perse in these conditions that is examined as the major sourceof variation. -

information about affect. It is much less impor-tant for us to know whether someone has just said"You are a friend" or "You are a fiend" than toknow whether it was spoken in contempt or withaffection. Argyle and his colleagues (Argyle, Sal-ter, Nicholson, Williams, & Burgess, 1970) foundthat 22 times more variance is accounted for bythe tone of one's voice than by the content ofthe utterance when people are asked to interpretutterances. In fact, even when the content ofrecorded utterances is nearly completely obliter-ated by means of electronic masking, filtering, orrandom splicing of the tape, subjects still canencode the emotions expressed in these utterancesquite reliably (Dawes & Kramer, 1966; Scherer,Koivumaki, & Rosenthal, 1972). And we have nodifficulty in identifying emotions expressed bymembers of unknown cultures speaking unknownlanguages. In a recent volume on person per-ception, Schneider, Hastorf, and Ellsworth (1979)noted that "inferences based on nonverbal cuesare primarily inferences about relationships andfeelings, and thus are among the most importantinferences we make" (p. 142). One cannot beintroduced to a person without experiencing someimmediate feeling of attraction or repulsion andwithout gauging such feelings on the part of theother. We evaluate each other constantly, weevaluate each others' behavior, and we evaluatethe motives and the consequences of their behavior.And you have already made up your mind aboutthis paper!

Nor is the presence of affect confined to socialperception. There are probably very few percep-tions and cognitions in everyday life that do nothave a significant affective component, that aren'thot, or in the very least tepid, And perhaps all

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 153

perceptions contain some affect. We do not justsee "a house": we see "a handsome house," "anugly house," or "a pretentious house." We donot just read an article on attitude change, oncognitive dissonance, or on herbicides. We read an"exciting" article on attitude change, an "impor-tant" article on cognitive dissonance, or a "trivial"article on herbicides. And the same goes for asunset, a lightning flash, a flower, a dimple, ahangnail, a cockroach, the taste of quinine, Sau-mur, the color of earth in Umbria, the sound oftraffic on 42nd Street, and equally for the soundof a 1000-Hz tone and the sight of the letter Q.5

Feeling and Thinking

According to the prevalent models for affect (e.g.,Figure 1), preferences are formed and expressedonly after and only as a result of considerableprior cognitive activity. How fully and completelymust objects be cognized before they can be evalu-ated? I argue, along with Wundt and Cummings,that to arouse affect, objects need to be cognizedvery little—in fact, minimally.

In order to consider this possibility more spe-cifically it is important to distinguish betweenthoughts and feelings. At the genotypic level, thisdistinction is not an easy one to make, for ithovers dangerously near the mind-body duality.Some conceptual elements of this distinction, how-ever, may be identified for purposes of clarity.While feelings and thoughts both involve energyand information, the first class of experiences isheavier on energy, whereas the second is heavieron information (e.g., Inhelder & Piaget, 1QS8;pp. 347-348). In the pure case, the analysis offeelings attends primarily to energy transforma-tions, for example, the transformation of chemicalor physical energy at the sensory level into auto-nomic or motor output. In contrast, the analysisof thoughts focuses principally on informationtransformations. In nearly all cases, however,

5 This conjecture probably does not apply to incidentalperceptions where the attentive processes are at minimum,although it is not inconceivable that the traces of theseincidental perceptions still might recruit affect upon re-trieval and thus become hot. In fact, Izard (1979) as-sumes that some emotion is always present in conscious-ness. Normally, it is the emotion of "interest" that domi-nates behavior. This emotion, which directs and sustainsattention and exploration, is absent only when other emo-tions such as distress or anger "achieve consciousness"(p. 16S).

feeling is not free of thought, nor is thought freeof feelings. Considerable cognitive activity mostoften accompanies affect, and Schachter and Singer(1962) consider it a necessary factor of the emo-tional experience. Thoughts enter feelings at vari-ous stages of the affective sequence, and the con-verse is true for cognitions. Feelings may bearoused at any point of the cognitive process:registration, encoding, retrieval, inference, etc. Butthis converse relation is not totally symmetrical.I will later argue for Wundt's conjecture thataffect is always present as a companion to thought,whereas the converse is not true for cognition.In fact, it is entirely possible that the very firststage of the organism's reaction to stimuli and thevery first elements in retrieval are affective. It isfurther possible that we can like something or beafraid of it before we know precisely what it isand perhaps even without knowing what it is.And when we try to recall, recognize, or retrievean episode, a person, a piece of music, a story,a name, in fact, anything at all, the affectivequality of the original input is the first elementto emerge. To be sure, the early affective reactionis gross and vague. Nevertheless, it is capable ofinfluencing the ensuing cognitive process to a sig-nificant degree. Needless to say, after some cog-nitive activity has been executed, there may benew feeling to the stimulus. But the fact thatcognitions can produce feelings—as in listeningto a joke, for example, where affect comes at theend with a punch line or as a result of post-decision dissonance—need not imply that cogni-tions are necessary components of affect. What Iwant to argue is that the form of experience thatwe came to call feeling accompanies all cognitions,that it arises early in the process of registrationand retrieval, albeit weakly and vaguely, and thatit derives from a parallel, separate, and partly in-dependent system in the organism.

At the phenotypic level, we can support Wundt'sconjecture by spelling out in somewhat greaterdetail some of the ways in which affective judg-ments and reactions, or hot cognitions, differ fromtheir cold cognitive counterparts, keeping in mindthat the first category is represented by the proto-type "I like Joe," and the second by "Joe is aboy."

Affective reactions are primary. Wundt andCummings are joined by Bartlett and Osgood inthe view that feelings come first. Bartlett (1932)observes in his book on remembering,

1S4 • FEBRUARY 1Q80 • AMERICAN PSYCHOLOGIST

Attitude names a complex psychological state or processwhich .it is very hard to describe in more elementarypsychological terms. It is, however, as I have oftenindicated, very largely a matter of feeling, or affect. . . .[When] a subject is being asked to remember, veryoften the first thing that emerges is something of thenature of attitude. The recall is then a construction,made largely on the basis of this attitude, and its gen-eral effect is that of a justification of the attitude, (pp.206-207)

In his analysis of environments as perceptualtargets, Ittelson (1973) asserts that "the first levelof response to the environment is affective. Thedirect emotional impact of the situation, perhapslargely a global response to the ambiance, verygenerally governs the directions taken by subse-quent relations with the environment. It setsthe motivational tone and delimits the kinds ofexperiences one expects and seeks" (p. 16). Pref-erences influence language comprehension and lan-guage production as well (Premack, 1976). Os-good (1962) was impressed with the primacy ofaffect in a different way:

First, I must confess that, when we began this researchover ten years ago, I had the expectation that the majorfactors of the semantic space would represent the waysin which our sensory apparatus divides up the world—e.g., would parallel Boring's "dimensions of consciousness.". . . The accumulating data have proved my expectationwrong . . . the dominant factors of evaluation, potencyand activity that keep appearing certainly have a re-sponse-like character, reflecting the ways we can react tomeaningful events rather than .the ways we can receivethem.

But these major factors also seem to have an affectiveas well as a response-like character. As a matter of fact,the similarity of our factors to Wundt's (1896) tridimen-sional theory o f . jeeling—pleasantness-unpleasantness,strain-relaxation, and excitement-quiescence—has beenpointed out to me." (pp. 19-20)

It is significant also that at least three social-psychological conceptions labeled "cognitive" con-sistency theories focus not on consistency of con-tent but on the consistency of affect (Abelson &Rosenberg, 1958; Heider, 1958; Osgood & Tan-nenbaum, 1955).

Decisions are another area where thought andaffect stand in tension to each other. It is gen-erally believed that all decisions require someconscious or unconscious processing of pros andcons. Somehow we have come to believe, tauto-logically, to be sure, that if a decision has beenmade, then a cognitive process must have pre-ceded it. Yet there is no evidence that this isindeed so. In fact, for most decisions, it is ex-

tremely difficult to demonstrate that there hasactually been any prior cognitive process whatso-ever. One might argue that these are cases inwhich one alternative so overwhelmingly dominatesall the others that only a minimum of cognitiveparticipation is required and that that is why thecognitive involvement preceding such decisions isso hard to detect. But this argument must con-front the observation that if all decisions involvethe evaluation of alternatives, then when choicesappear quite lopsided to the decision maker, it iseven more important to scrutinize the alternativesthat appear inferior, for it is entirely possible thatone of them possesses some hidden but overridingvirtue. It is therefore not without merit to sup-pose that in many decisions affect plays a moreimportant role than we are willing to admit. Wesometimes delude ourselves that we proceed in arational manner and weigh all the pros and consof the various alternatives. But this is probablyseldom the actual case. Quite often "I decided infavor of X" is no more than "I liked X." Mostof the time, information collected about alterna-tives serves us less for making a decision than forjustifying it afterward. Dissonance is prevalentjust because complete and thorough computationis not performed before the decision (Festinger,1964). We buy the cars we "like," choose thejobs and houses that we find "attractive," andthen justify those choices by various reasons thatmight appear convincing to others who never failto ask us, "Why this car?" or "Why this house?"We need not convince ourselves.8 We know whatwe like.

In a study of consumer behavior, Quandt(1956) found that buyers often do not attend tothe features of the article that they consider cri-terial for their decisions and often base theirchoices on features that they previously dismissedas irrelevant, And Kahneman and Tversky (1979)have demonstrated that numerous axioms of de-cision theory that give decisions their rationalflavor are blatantly contradicted by experimentalresults.

»Phoebe Ellsworth (Note 1) illustrates the role ofaffect in her own recent decision experience. In tryingto decide whether to accept a position at another uni-versity, she says, "I get half way through my Irv Janisbalance sheet and say, 'Oh hell, it's not coming out right!Have to find a way to get some pluses over on the otherside!' "

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 155

Affect is basic. In one of her last books, whichbears the provocative title of Mind: An Essay onHuman Peeling, Susan K. Langer (1967) tried toshow "that the entire psychological field—includ-ing human conception, responsible action, ration-ality, knowledge—is a vast and branching develop-ment of feeling" (p. 23). Affect is the first linkin the evolution of complex adaptive functions thateventually differentiated animals from plants. Andunlike language or cognition, affective responsive-ness is universal among the animal species. Arabbit confronted by a snake has no time to con-sider all the perceivable attributes of the snakein the hope that he might be able to infer fromthem the likelihood of the snake's attack, thetiming of the attack, or its direction. The rabbitcannot stop to contemplate the length of thesnake's fangs or the geometry of its markings. Ifthe rabbit is to escape, the action must be under-taken long before the completion of even a simplecognitive process—before, in fact, the rabbit hasfully established and verified that a nearby move-ment might reveal a snake in all its coiled glory.The decision to run must be made on the basis ofminimal cognitive engagement.

It is thus significant that in categorizing facialexpressions, about 50% of the variance is ex-plained by the pleasant-unpleasant dimension(Abelson & Sermat, 1962; Hastorf, Osgood, & Ono,1966), and the same value is obtained for themultidimensional scaling of similarities among pho-tographs of faces (Milord, 1978). Similarly, itis a typical result in semantic differential studiesthat among the three factors Evaluation, Potency,and Activity, all of which Osgood considers to beaffective components of meaning, it is the firstthat accounts for about 50% of the variance.7

And it is no accident, according to Osgood (1969),that these three factors of the semantic space arefound repeatedly among diverse sets of concepts:

In my opinion, it is the innatcness of the emotional re-action system of the human animal that underlies theuniversality of the affective E-P-A components of mean-ing. In other words, the "innateness" of E-P-A . . . isreally the pan-humanes of emotional reactions, and theseobviously have evolutionary significance for the survivalof any species. Organisms without other specialized adap-tive mechanisms (e.g., armor, coloration, poisons, etc.)which were unable to represent for themselves the goodversus bad implications of things (antelope versus saber-toothed tiger), the strong versus weak of things (saber-toothed tiger versus mosquito), and the quick versus slowof things (saber-toothed tiger versus quicksand) wouldhave little chance of survival. In the human speciesthese "gut" reactions to things appear as the affectivemeaning system (the E-P-A components of total mean-

ing), and it is these components which provide us withwhat might most appropriately be called the "feeling-tones" of concepts as a part of their total meaning, (p.195)

Affective reactions are inescapable. Unlikejudgments of objective stimulus properties, affec-tive reactions that often accompany these judg-ments cannot always be voluntarily controlled.Most often, these experiences occur whether onewants them to or not. One might be able to con-trol the expression of emotion but not the experi-ence of it itself. It is for this very reason thatlaw, science, sports, education, and other institu-tions of society keep devising ever new means ofmaking judgments "objective." We wish somedecisions to be more independent of these virtuallyinescapable reactions.

We may completely fail to notice a person'shair color or may hardly remember what it wasshortly after meeting the person. But we canseldom escape the reaction that the person im-pressed us as pleasant or unpleasant, agreeable ordisagreeable, as someone to whom we were drawnor someone by whom we were repelled. And theseaffective reactions—and, more important, the re-trieval of affect—occur without effort. In con-trast, some cognitive judgments require substan-tial effort. Chess contestants typically lose sev-eral pounds of their weight in the course of atournament.

Because affective judgments are inescapable,they cannot be focused as easily as perceptual andcognitive processes. They are much more influ-enced by the context of the surround, and theyare generally holistic. Affective reactions are thusless subject to control by attentive processes.8

7 It is therefore something of a paradox that so littleattention is paid to affect in information-processing studies.Most of the tasks in experiments on information process-ing are verbal. Most of them involve some forms ofsemantic memory. If the semantic space is primarily anaffective space, as Osgood argues, then the affective com-ponents and qualities of information need to be given asmuch attention as their phonemic, graphemic, lexical,semantic, conceptual, or pictorial counterparts.

8 The existentialists (e.g., Sartre, 1947) ascribe a sub-stantial voluntary component to emotion. "The existen-tialist does not believe in the power of passion. He willnever agree that a sweeping passion is a ravaging torrentwhich fatally leads a man to certain acts and is thereforean excuse. He thinks that man is responsible for his pas-sion" (pp. 27-28). Because of the participation of sen-sory, cognitive, and motor processes, the argument thatemotions have some voluntary component is not withoutbasis.

156 • FEBRUARY 1980 • AMERICAN PSYCHOLOGIST

Affective judgments tend to be irrevocable.Once a cognitive judgment has been made—forexample, that at the forthcoming social hour therewill be more scotches drunk than bourbons-—onecan still be persuaded that it may turn out other-wise. It can be pointed out, say, that the dis-tribution of ages of the guests is different thanthat we really like scotch better than bourbon,is greater than the supply of scotch. We canreadily accept the fact that we can be wrong.But we are never wrong about what we like ordislike. Hot cognitions are seldom subjectivelyfalse. It would be much harder to persuade usthat we really like scotch better than bourbon,given that we feel otherwise. Once formed, anevaluation is not readily revoked. Experimentson the perseverance effect, the strong primacyeffects in impression formation, and the fact thatattitudes are virtually impervious to persuasion bycommunication all attest to the robust strengthand permanence of affect. Affect often persistsafter a complete invalidation of its original cog-nitive basis, as in the case of the perseverancephenomenon when a subject is told that an initialexperience of success or failure has been totallyfabricated by the experimenter (Ross, Lepper, &Hubbard, 1975).

The reason why affective judgments seem soirrevocable is that they "feel" valid. We are noteasily moved to reverse our impression of a personor of a piece of music. We trust our reactions,we believe that they are "true" and that they ac-curately represent an internal state or condition.Perhaps the subjective validity of affective judg-ments and reactions and our confidence in thesejudgments derive from the Cartesian traditionB

that allows us to doubt everything except our ownfeelings, especially the feelings of doubt. Perhapsit reflects a basic reality.10

Affective judgments implicate the selj. Whenwe evaluate an object or an event, we are describ-ing not so much what is in the object or in theevent, but something that is in ourselves. Cog-nitive judgments deal with qualities that reside inthe stimulus: "This cat is black," "Camembertand Brie are soft-ripened cheeses." These judg-ments are made on I-scales that are orders ofstimuli (Coombs, 1964). Affective judgments,however, are made on J-scales, that is, scales onwhich are located jointly the various stimulias well as the ideal preference point of the per-son. "I dislike this black cat" or "I prefer Cam-

embert to Brie" are judgments on J-scales. Thus,affective judgments are always about the self.They identify the state of the judge in relation tothe object of judgment.

Affective reactions are difficult to verbalize. Theremarkable aspect of first impressions of personsis their immediacy. When we meet a stranger, weknow within a fraction of a second whether welike the person or not. The reaction is instan-taneous and automatic. Perhaps the feeling isnot always precise, perhaps we are not alwaysaware of it, but the feeling is always there. If ourlater experience with the stranger conflicts withthe first impression, we are terribly surprised. Weconsider it an exception. Paradoxically, this sub-jective validity of affective reaction, this certaintythat we "know what we like," is often accom-panied by our inability to verbalize the reasons forour attraction or repulsion to the person.11 Whenasked why we like someone, we say that we likethe person because he or she is "nice," "pleasant,"or "interesting." But these adjectives describeour reactions to the person, not the person. Theresimply aren't very effective verbal means to com-municate why we like people and objects or whatit is that we like about them.

The communication of affect, therefore, reliesmuch more on the nonverbal channels (Ekman &Friesen, 1969; Schneider, Hastorf, & Ellsworth,1979). Yet it is remarkably efficient. And it isin the realm of nonverbal expression of feelingsthat their basic nature is again revealed. Theuniversality of emotional expression strongly sug-gests our evolutionary continuity with other spe-cies and the fundamental nature of affect. Thefacial expressions of humans upon biting into a

9 Hume (1898), too, held that emotions (passions) can-not be false. "A passion must be accompanied with somefalse judgment, in order to its being unreasonable; andeven then 'tis not the passion properly speaking, which isunreasonable, but the judgment" (p. 196).

10 Because nonverbal cues exchanged in social inter-action are dominated by affect, they are perceived as hav-ing such properties as trustworthiness and freedom fromvoluntary control (Schneider, Hastorf, & Ellsworth, 1979,pp. 123-127).

"Mandler (197S), Neisser (1967), and Nisbett andWilson (1977) pointed out that individuals have no accessto the cognitions that occasion, mediate, or cause theiractions, that are parts of their attitudes, or that deter-mine their preferences. On the basis of an extensive re-view of the social psychological literature, Nisbett andWilson (1977) concluded that introspective reports aboutinfluences on the subjects' evaluations, decisions, and ac-tions Were so unreliable as not to be trusted.

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 157

sour apple and their expressions of surprise, anger,delight, or serenity are remarkably similar acrossall cultures and are not far removed from theexpressions of the great apes. Perhaps we havenot developed an extensive and precise verbalrepresentation of feeling just because in the pre-linguistic human this realm of experience had anadequate representation in the nonverbal channel.

The role of affective communication is par-ticularly significant in the social interaction amonganimals. The effectiveness of communication ofaffect and the accuracy of recognition of affectiveexpression are illustrated by the results of Prattand Sackett (1967). They raised rhesus monkeysin conditions that allowed complete contact withpeers, in conditions that allowed only visual andauditory access, and in complete isolation. Themonkeys were then examined for the kinds ofanimals they preferred to approach. Those raisedunder the same conditions preferred each othertwice as much as those raised under different con-ditions, even when the stimulus animals were totalstrangers to the test monkeys. While it could notbe determined what sorts of cues allowed the ani-mals to make these fine discriminations, it is verylikely that the three groups developed during thecourse of their previous experience distinct pat-terns of emotional responding to new stimuli andto strange individuals, and that the animals raisedunder the same conditions found each other moreattractive because of the familiarity of these emo-tional patterns.

The reliance of affect on nonverbal means ofcommunication has, I believe, implications for theway it is processed. For if affect is not alwaystransformed into semantic content but is insteadoften encoded in, for example, visceral or muscularsymbols, we would expect information containedin feelings to be acquired, organized, categorized,represented, and retrieved somewhat differentlythan information having direct verbal referents.Recent electromyographic research provides strongevidence for the participation of muscular activityin the imagination, recall, and production of emo-tional states (Lang, 1979; Schwartz, Fair, Salt,Mandel, & Klerman, 1976). In light of theseintuitions, it is not unreasonable to speculate thatthe processing of affect is closer to the acquisitionand retention of motor skills than of word lists.

Affective reactions need not depend on cognition.At the turn of the century, Nakashima (1909a,1909b) tried to find support for Wundt's affective-

primacy conjecture by comparing reaction timesfor psychophysical judgments and for preferences.He failed. But he did find evidence that judg-ments of pleasantness were independent of sensoryqualities and that these judgments could not havebeen mediated by these qualities. Similar inde-pendence, based on multidimensional scaling, hasbeen reported more recently, for example, in study-ing the perceptions of and preferences for softdrinks. Cooper (1973) found that similarity scal-ing yielded a space dominated by a "cola-ness"dimension, whereas preference scaling generated aspace dominated by popularity of the drinks.Generally, it appears that similarity judgmentspredict preferences only when the similarity judg-ments are themselves highly evaluative, as in thecase of admissions officers judging college candi-dates (Klahr, 1969) or art-trained students judg-ing paintings (Berlyne, 197S; O'Hare, 1976). Os-good (1962) took it as a given that the affectivereaction system "is independent of any particularsensory modality" (p. 21).

If there is indeed a separation between affectand cognition, then it is not surprising that re-search on preferences, attitudes, attractions, im-pressions, aesthetic judgments, and similar affec-tive responses—research that commonly has in-voked cognitive mediators—has not been terriblysuccessful. If overall preferences were simply amatter of calculating the combination of weightedcomponent preferences, and if component prefer-ences were nothing more than cognitive representa-tions of object features marked with affect, thenthe problems of predicting attitudes, decisions,aesthetic judgments, or first impressions wouldhave been solved long ago. After all, these prob-lems have been around for nearly a century. Yetexcept for trivial cases or cases in which theresponses are highly cognitive (e.g., Yntema &Torgerson's [1961] study of judgments of ellip-ses), the cognition-based solutions to these prob-lems have rarely predicted more than 20% of thetotal variance.

The dismal failure in achieving substantial atti-tude change through various forms of communica-tion or persuasion is another indication that affectis fairly independent and often impervious to cog-nition. If attitudes consist of information unitsthat have affect or utilities attached to them, thento change an individual's attitude, what could besimpler than providing the individual with'alter-native information units that have the same sort

158 • FEBRUARY 1980 • AMERICAN PSYCHOLOGIST

of affect as that attached to the desired attitude?If a person believes that Candidate A is honest,we can simply give the person information provingthat A is not honest. Or, we could change thecentrality or the weight of honesty. Yet thisapproach has been the least successful in attitudechange, Even the most convincing arguments onthe merits of spinach won't reduce a child's aver-sion to this vegetable. Direct persuasion effectshave been so weak that researchers have insteadturned to more pernicious avenues of attitudechange, such as insufficient justification, persuasionthrough distraction, the foot-in-the-door technique,or the bogus pipeline.

It is unlikely that calculations based on dis-criminable component features and their affectivevalues will reliably predict our overall affectivereactions to objects and events. These reactionsdo not seem to be composites of such elements.An affective reaction to a person we meet emergeslong before any of these features can be identified,let alone evaluated. The assumption that com-ponent affect, utilities, or values attach themselvesto the very same features that the subject attendsto in a typical detection, recognition, discrimina-tion, or categorization task is likely to be wrong.12

The analysis of preferences is not simply an analy-sis of cold cognitive representations that have be-come hot, that is, cognitive representations thathave some affect attached to them.13 The stimulusfeatures that serve us so well in discriminating,recognizing, and categorizing objects and eventsmay not be useful at all in evaluating these ob-jects. If this is indeed the case, then there mustexist a class of features that can combine morereadily with affect and thereby allow us to makethese evaluations, to experience attraction, repul-sion, pleasure, conflict, and other forms of affect,and to allow us to have these affective reactionsquite early after the onset of the sensory input.These features might be quite gross, vague, andglobal. Thus, they might be insufficient as a

121 did not have the slightest doubt of this assumption,however, when I wrote my dissertation (Zajonc, 1955),which employed it without question.

13 The term hot cognition has been used fairly indis-criminately, although it generally refers to cases whehaffect accompanies or qualifies information. "I have amalignant tumor" is a hot cognition. However, theemotional experience of listening to one's favorite pieceof music performed by one's favorite artist is less likelyto receive the label of hot cognition. It is even less mean-ingful to speak of hot cognitions when affect becomes sepa-rated from the original cognitions.

basis for most cognitive judgments—judgmentseven as primitive as recognition, for example. Inorder to distinguish this class of features fromsimple discriminanda, I call them preferenda (Za-jonc, Note 2).

I cannot be very specific about preferenda. Ifthey exist they must be constituted of interactionsbetween some gross object features and internalstates of the individual—states that can be alteredwhile the object remains unchanged, as, for ex-ample, when liking for a stimulus increases withrepeated experience. Color preferences are a casein point. Similarity scaling of color yields threedimensions—brightness, hue, and saturation—thatexplain almost all of the variance in similarityjudgments. But on the basis of Nakashima's(1909a) research and according to unpublishedwork of Premack and Kintsch (Note 3), the scal-ing of color for preference would not reveal thesethree factors. If we did not know from othersources that brightness, hue, and saturation ex-haust the entire range of differences among colors,then we would not discover them by means ofpreference scaling. Abstract preferences for colorand color preferences for classes of objects, suchas hair, cars, or houses, are still more problematicif we insist on using brightness, hue, and satura-tion in quantifying them. And the same appliesto face recognition: Physical features do not serveas discriminanda for faces (Milord, 1978; Patter-son & Baddeley, 1977). It is therefore an inter-esting problem to discover what it is in color that"holds" affect if it isn't brightness, hue, and satu-ration and what it is in a face that "holds" affectif it isn't physical features. The answer to thisproblem is probably that some physical aspects,perhaps vague, gross, or configural, are involved,but not alone. Preferenda must consist of aninteraction of these global features with some in-ternal state or condition of the individual.

Affective reactions may become separated fromcontent. It sometimes happens that we are re-minded of a movie or of a book whose contents •we are unable to recall. Yet the affect presentwhen leaving the movie or our general impressionof the book are readily accessible. Or we are re-minded of an interpersonal conflict of long ago.The cause of the conflict, the positions taken, thematter at issue, who said what, may have all beenforgotten, and yet the affect that was present dur-ing the incident may be readily retrieved. Suchexperiences, together with such clinical phenomena

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 1S9

as free-floating anxiety, hysteria, or posthypnoti-cally induced moods, all point to the possibilitythat some aspects of affective processes might wellbe separate and partly independent of cold cog-nitions. Occasions when they are not include thosewhen an affective experience has been communi-cated to someone else or when it has been thoughtof a great deal. On such occasions an elaboratecognitive representation of affect occurs that maybe processed very much like any other type ofinformation. It is important to observe, however,that not all affective experiences are accompaniedby verbal or other cognitive representations andthat when they are, such representations are im-precise and ambiguous.

Preferences Need No Inferences:Empirical EvidenceThe prevalent approach to the study of prefer-ences and related affective phenomena holds thataffective reactions follow a prior cognitive process:Before I can like something I must first knowwhat it is. According to this prevalent view,therefore, such cold cognitive processes as recog-nition or categorization are primary in aestheticjudgments, in attitudes, in impression formation,and in decision making: They come first. If wesay, for example, that we like John because he isintelligent, rich, and compassionate, it follows thatwe must have gained some impression of John'sintelligence, wealth, and compassion, and combinedthem, before we formed an attraction to him. Thismust be especially so in the case of judgments ofnovel stimuli before the component units becomefused into an integrated structure. Thus, if thecomplexity of polygons is an important basis oftheir attractiveness, then polygons that are judgedpleasing (or displeasing) must have previouslybeen somehow examined for their complexity.Otherwise, the calculus of preferences makes littlesense.14

The first indication that affect may not requireextensive participation of cold cognitive processesappeared in studies of the exposure effect, that is,the phenomenon of increasing preference for ob-

14 Affective reactions to objects that have been en-countered and evaluated many times may become auto-mated, thus gaining some independence from the compo-nent processes (Shiffrin & Schneider, 1977). As such, theymay have different properties than first reactions, It isthose first affective reactions (that is, those elicited whenindividuals are asked to evaluate objects totally novel tothem) that I wish to consider at this point.

jects that can be induced by virtue of mere re-peated exposure (Harrison, 1977; Zajonc, 1968).While the empirical results that established thephenomenon were quite consistent, their explana-tion continued to be very elusive. Theories thatattempted to account for the mere exposure effect,such as Harrison's (1968) response competitionhypothesis or Berlyne's (1970) optimal arousaltheory, treated affect as resulting from a priorcognitive process. Both theories contained theremnants of Titchener's (1910) thesis on familiar-ity. In explaining the preference for familiar ob-jects, Titchener attributed a critical role to recog-nition, which he thought gave the individual a"glow of warmth, a sense of ownership, a feelingof intimacy" (p. 411). The majority of subse-quent findings bearing on the explanation of theexposure effect, however, have revealed that recog-nition must play a relatively minor role, as mustthe subjective feeling of recognition.

Matlin (1971) was the first to discover thatthe role of recognition in the exposure effect mayhave been overstated. During an initial experi-mental session, she presented Turkish-like wordseither three times or six times. Subsequently,these words, together with others that were notshown at all, were rated for liking and also forfamiliarity. That is, for each word the subjectshad to decide whether they saw it previously inthe exposure series and to report how much theyliked it. Table 1 shows Matlin's results. Likingis averaged as a function of objective familiarityand as a function of subjective familiarity. Notethat there is an effect due to subjective familiarity,that is, when the subjects thought a stimulus wasold they rated it more positively than when theythought it was new. However, the objective his-tory of the individual's experience with the stimu-lus is just as effective in influencing liking. Stim-

TABLE 1

Average Stimulus Affect Ratings as a Function ofObjective Familiarity (Old-New) and SubjectiveFamiliarity ("Old"-"New")

Subjective familiarity

Objective familiarity

OldNew

M

"Old" "New"

4.90 4.204.20 3.90

4.55 4.05

4.474.01

Note. Data are from Matlin (1971).

160 • FEBRUARY 1980 • AMERICAN PSYCHOLOGIST

V, -39

E2={.24)

Subjective Familiarity

Recognition Confidence(•<—E4=-81

Recognition Accuracy p E5=.56V2-.48

Figure 2. Causal model for independent affect; goodness of fit is x2(5) = 39.0. V = variable; E =error. (From "Exposure Effects May Not Depend on Stimulus Recognition" by R. L. Morelandand R. B. Zajonc, Journal of Personality and Social Psychology, 1979, 37, 1085-1089. Copyright1979 by the American Psychological Association, Reprinted by permission.)

uli that the subjects had actually seen were likedbetter than stimuli not seen, independently ofwhether the subjects thought of them as "old"or "new,"

Similar results were obtained recently by More-land and Zajonc (1977, 1979), using Japaneseideographs. Subjects were given 0, 1, 3, 9, and27 prior exposures, counterbalanced, of course,with the stimuli. Following these exposures, thesubjects made a variety of recognition and likingjudgments. A number of findings are of interest.Many stimuli shown in the first series, some ofthem 27 times, were not recognized as familiarwhen shown later. Taking only those stimuli thatwere so judged, and relating the rated attractive-ness of these stimuli to their actual number ofexposures, we obtained correlations of .43 in oneexperiment and of ,50 in another. An objectivehistory of exposure influenced liking of stimuli forwhich the subjects could not have felt a "glow ofwarmth" or a "sense of ownership."

We also performed another type of analysis.Because we had a sufficient number of measures,we were able to use linear structural equationanalyses to evaluate various causal models of ourdata. We used the LISREL III program (seeJoreskog & Sorbom, 1977) to calculate maximumlikelihood estimates for causal models that assign

different roles to the recognition factor. Theprogram distinguishes between latent variables(constructs) and their observed indicators (mea-sures). By estimating the unknown coefficientsin a system of simultaneous equations for anyparticular model, the program describes the pat-tern of relations among the latent variables, dis-tinguishing causal effects from unexplained varia-tion in each case.

The results of this analysis are shown in Figure2. Latent variables are shown in ellipses, whilemeasures of those variables are shown in rec-tangles. The coefficients linking the ellipses withthe boxes represent the validities with which par-ticular latent variables were assessed by their mea-sures. Path coefficients linking the latent vari-ables to each other represent causal relations.Unexplained variation in the latent variables (Viand V2) and error in the various measures (Eithrough E5) are also shown. Some parameters(shown in parentheses) had to be set equal tosome a priori value in the maximum likelihoodsolution so that variance in all of the latent vari-ables could be identified,

The first model tested was one postulating thatstimulus exposure has two mutually independenteffects, one cognitive and one affective, or onecold and one hot. We supposed that under the

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 161

Ef(.00>*"

-Ev=.45

Recognition Confidence *"—£4=82

Recognition Accurocyp—E5=.68

V, =.33

Figure 3. Causal model for mediated affect; goodness of fit is x 2 (6) =83.6. V = variable; E =error. (From "Exposure Effects May Not Depend on Stimulus Recognition" by R. L. Moreland andR. B. Zajonc, Journal of Personality and Social Psychology, 1979, 37, 1085-1089. Copyright 1979 bythe American Psychological Association. Reprinted by permission.)

impact of repeated exposure, people gain an in-creasing ability to recognize the stimulus—Theyachieve a feeling of subjective familiarity and anawareness of recognition, which authors sinceTitchener have thought to be the necessary condi-tions for an increased positive affect toward thestimulus. This is the purely cold effect that iscapable of generating the eventual "glow ofwarmth." However, we wanted to know as wellwhether, quite independently of this cold cogni-tive effect, there is also an affective change, orhot effect—that is, whether subjects acquire amore positive attitude toward the object as ex-posure increases, independently of recognition.They do. While the path coefficient from stimulusexposure to subjective recognition is substantial(.53), indicating that recognition improves withexposure, there also is a hot effect: There is astrong path from stimulus exposure to subjectiveaffect that is independent of recognition (.66).

We can compare this model with one that isentirely cold, that is, with one that requires theentire process to be mediated by cognitive factors,by the discriminanda. This model, shown in Fig-ure 3, says essentially that whatever affectivechanges take place as a result of exposure areentirely mediated by stimulus recognition. Theresult of requiring affect to be mediated by recog-nition is a substantial reduction in the efficiencyof prediction. The x2 in the previous model was39.0 (df - 5) and in this model is 83.6 (df = 6),generating a significant (p < .01) difference be-tween the two models of x 2 ( l ) = 44.6.

The experiments just described all involved pre-sentation of stimuli under optimal conditions;that is, there was nothing to prevent the subjectsfrom registering what was shown and from memo-rizing the information presented to them. Sub-

jective recognition and the likelihood of recogni-tion were controlled by statistical techniques. Andthe results showing that stimulus recognition wasnot a necessary condition for the exposure effectwere correlational.

Much firmer evidence, however, that hot cogni-tion is quite short on cognition was collected byW. R. Wilson (197S), who controlled for recogni-tion experimentally by means of an ingenious tech-nique. He employed the method of dichotic lis-tening in order to reduce recognition to a chancelevel. Random sequences of tones, such as thoseconstructed by Vitz (1964), were presented to oneear, and a story was simultaneously presented tothe other. Subjects were asked to track the storyon a written page to verify whether what theyheard corresponded to the printed text. The melo-dies were played five times each. The subjectswere subsequently given a recognition memorytest in which the earlier melodies and other melo-dies that they had never heard were played. Butnow there was no interference from the other chan-nel, and no other task was required of the sub-ject. The subjects also rated all the melodies forliking, some subjects giving their recognition mem-ory judgments before, others after, the ratingsfor liking. The procedure succeeded in reducingrecognition memory nearly to the chance level.The accuracy of recognition was 59% in one ex-periment and only 53% in another.

Table 2 shows the results of these experiments.Again, as in the case of previous results, likingvaries with subjective recognition. But apart fromthis effect, liking also varies with the objective his-tory of stimulus exposure. With recognition reducednearly to the chance level, differential affectivereaction to the stimuli is obtained as a consequenceof mere repeated exposure. Random melodies

162 • FEBRUARY 1980 • AMERICAN PSYCHOLOGIST

presented five times were liked better than melo-dies never heard, even though the subjects couldnot discriminate the former from the latter forfamiliarity.

In a follow-up of these studies, Kunst-Wilson(who is the same person as W. R. Wilson) and Itried to reproduce the effect in a visual mode(Kunst-Wilson & Zajonc, 1980). Random poly-gons were constructed and presented for an ex-tremely brief time interval—in fact, only 1 milli-second. Subsequently, the subjects rated the poly-gons for liking and were tested for their recogni-tion memory. Judgments were made in pairedcomparisons to avoid possible response bias.Again, recognition was at a chance level: 48%.However, of the stimuli that were liked, 60%were old and 40% were new. Sixteen of 24 sub-jects liked objectively old stimuli better than newstimuli, but only 5 of 24 recognized them as suchat better than chance level. And of the 24 sub-jects, 17 showed better discrimination betweenobjectively old and objectively new stimuli in theiraffective judgments than in their recognition re-sponses, while only 4 showed such superiority ofrecognition over affective judgments. Thus, thesubjects were able to distinguish between the oldand new stimuli if they used liking as their re-sponse, but they were not able to distinguish be-tween them if they had to identify them as "old"or "new." This result may be taken as, evidencethat a Class of features (preferenda) exists thatallows individuals to experience affect toward ob-jects but does not allow them to accomplish cog-nitive tasks as simple as those in recognitionmemory tests.

TABLE 2

Average Stimulus A feet Ratings as a Function ofObjective Familiarity (Old-New) and SubjectiveFamiliarity ("Old"-"New")

Subjective famil iar i ty

Objective familiarity

OldExperiment IExperiment II

NewExperiment IExperiment II

MExperiment IExperiment II

"Old"

4.203.51

3.7S3.03

4.023.29

"New"

4.033.85

3.073.02

3.523.40

M

4.123.66

3.303.03

Note. Data are from Wilson (1975).

These experiments establish, I believe, that af-fective reactions to a stimulus may be acquiredby virtue of experience with that stimulus even ifnot accompanied by such an elementary cold cog-nitive process as conscious recognition. Thus, atheory that assumes that subjective experiences ofnovelty and familiarity mediate the affective re-sponse acquired during the course of exposuresmust contend with the results showing that withthe subjective experience of novelty held constant,systematic variations in affect can be obtainedjust by means of an objective manipulation of ex-posure.

However, one should not assume that no formof recognition occurred. Obviously, some discrimi-nation, however primitive or minimal, must havetaken place, even though it must have been ata level not accessible to the subject's awareness.It is somewhat surprising that any effect at allwas obtained with exposures as short as 1 milli-second, but it should be noted that the stimuliwere high contrast (black on white) and that nomask was used. Detectable effects with 1-milli-second exposures were also obtained by Shevrinand Fritzler (1968) and by Shevrin, Smith, andFritzler (1971). These authors reported differ-ential evoked potentials and word associations tocritical and control stimuli presented for 1 milli-second—stimuli that the subject could neitherrecognize nor identify. Even more pertinent isthe work of Marcel (Note 4). He presented overa large number of trials either a single word ora blank always followed by a mask. The exposureduration of the word was varied. The subjectswere then asked whether anything had been pre-sented before the mask. If they answered yes,two words were then presented to them underoptimal conditions. The subjects were then askedwhich of these two words was more visually simi-lar to the one shown before the mask. Finally,they were asked which of these same two wordswas more semantic ally similar to the stimulusshown before the mask. With decreasing stimu-lus exposure, all three types of judgments tendedto become less accurate, and eventually all threereached the chance level. But the first to becometotally unreliable were judgments regarding theactual presence of the stimulus words. The secondtype of judgment to be reduced to a chance levelby the decreasing exposures was that concernedwith physical similarity. And when the subjectswere totally unable to rise above chance in com-

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 163

SURE

to 2.5i

Ul

z 2.0UJQU-

8 1.5

GUESS

a ..IKE-DISLIKE

OLD- NEW

6w,LIKING RECOGNITIONFIRST FIRST

Figure 4. Confidence in liking and recognitionjudgments. (Drawn from data reported by Kunst-Wilson & Zajonc, 1980.)

paring physical similarities of the words, theywere still judging their semantic similarities quitereliably.

Marcel's results are reminiscent of those re-ported by Broadbent and Gregory (1967), whofound that unpleasant words (such as "blood")were more often misperceived as other unpleasantwords (such as "death") than as equally probableneutral words. Marcel's results, moreover, are ofparticular interest if we consider the consistentfindings from the semantic differential literatureshowing that meaning is very highly saturated withaffect. If it is indeed affect that allows subjectsto make a semantic match in the absence of con-scious recognition, then deciding which of twogiven words is emotionally more similar to a stim-ulus word should be at least as easy as decidingwhich is semantically more similar. This experi-ment, in fact, is now being carried out by More-land and myself.

Another consequence derives from the prevalentapproach to affect and cognition. Prevalent the-ories, especially the one developed by the lateDaniel Berlyne (1967), generally assume that theaffective reaction occurs in response to the levelof arousal, which in turn is mediated by collativevariables such as complexity, novelty, or congru-ity. If complexity, congruity, and novelty medi-

ate liking, in that objects and events are likedjust because they are optimally complex or simple,novel or familiar, then the judgments of objectsalong these dimensions should, in general, be morestable, more consistent, and made sooner thanaffective judgments. At the very least, these judg-ments should not be slower, more inconsistent,unstable, or inefficient than affective ratings. Inparticular, we would expect that recognition judg-ments, for example, which reflect the operation ofthe collative variable of novelty, should be madewith greater confidence than liking judgments.Figure 4 shows the results from our previousstudy with Kunst-Wilson using 1-millisecond ex-posures. The results show that compared toliking judgments, recognition judgments are madewith much less confidence. The differences are,in fact, huge—more than 6 times their standarderrors. Even if we take only the recognition judg-ments on which the subject was correct, this effectremains true.1-"1

One more bit of data. According to the preva-lent view, attending to discriminanda alone shouldbe easier and quicker than attending to discrimi-nanda tagged with values. Since the latter in-volve more information, more detail must be at-tended to, and the subject would consequently

lr' We suspected that these results may be due to thefact that the subjects knew they could be wrong onthe old-new judgments, and awareness of this fact mighthave induced caution in them. But they could not be"wrong" on their liking judgments. These latter judg-ments express opinions, and people generally feel free tohold any opinions whatsoever. We tried, therefore, to"objectify" affective judgments and to "subjectify" recog-nition judgments in order to determine whether the con-fidence ratings would be reversed. To obtain "objectified"affective ratings, subsequent to stimulus exposures, weasked subjects in another experiment to rate the polygonsfor their "aesthetic value." We told them also in thisconnection that our polygons had all been rated for aes-

, thetic value by art critics. To obtain "subjectified" recog-nition judgments, we told the subjects that one of thetwo polygons in each slide might appear more "familiar"than the other and asked the subjects to indicate whichone did in fact appear more familiar. Thus, the subjectscould now be "wrong" in their affective judgments, where-as, recognition became much more a matter of subjectiveimpression. The results did not change a great deal.Confidence was a little greater for subjective familiarityjudgments than for the old-new judgments and a littleweaker for aesthetic judgments than for judgments ofoutright liking. But these differences were quite small.The means were 2.01 and 2.41 for familiarity and aes-thetic judgments, whereas they were 1.60 and 2.29 forrecognition and liking.

164 • FEBRUARY 1980 • AMERICAN PSYCHOLOGIST

require more processing time. If familiarity medi-ates the affective reaction generated as the resultof repeated exposures, then judgments of famili-arity should be made quicker than judgments ofliking. If anything, however, our results showedthe opposite. Although of only borderline signifi-cance, affective judgments of polygons were madefaster than recognition judgments.

Feeling and Thought: Two Systems?About 10 years ago, Hyde and Jenkins (1969)carried out an experiment in which subjects weretested for recall of word lists to which they wereexposed under different conditions. Some sub-jects were simply exposed to the words with theinstruction to pay close attention. Of three othergroups^ one was instructed to count the numberof letters in each word, another to report thepresence of the letter E, and a third to rate eachword for pleasantness. In some groups the sub-jects were warned that they would be tested forrecall; in others they were not warned. Hyde andJenkins's results were quite strong. Among boththe subjects who were warned about a future recalltest and those who were not warned, those askedto rate pleasantness showed the best recall. Hydeand Jenkins took their data to mean that itemsare "arranged" differently in storage dependingon the context prevailing during acquisition. Thesuperior performance of subjects who rated thewords for pleasantness was due to the fact thatthese subjects acquired the words as "units ofmeaning" and could therefore recruit for themsupportive components of associative structures.When words are examined for number of lettersor the presence of the letter E, such "structuresare not activated and the recall is unorganized"(Hyde & Jenkins, 1969, p. 480).

Since, as we have seen earlier, the semantic con-tent of verbal material is saturated with affect,the facilitation that occurred as a result of priorpleasantness ratings in the Hyde and Jenkins ex-periment could have strong affective components.Two conditions of a recent experiment by Rogers,Kuiper, and Kirker (1977) are of particular in-terest in this respect. In all conditions, their sub-jects were tested for recall of previously shownadjectives. In one condition, the subjects had tocheck the adjectives to see if they were printed inthe same or different type as a sample print (struc-tural), and in another whether they rhymed withcomparison words (phonemic). But in two condi-

tions the subjects were required to engage in ex-tensive semantic processing of the adjectives: Inone they checked the adjectives to see if theymeant the same as comparison words (semantic)and in another to see whether they described thesubject (self-reference). Note that while theselast two conditions both activate semantic struc-tures within which the adjectives are imbedded,the self-reference condition brings the subject intoa cognitive domain greatly charged with affect.Rogers, Kuiper, and Kirker's findings are espe-cially significant in view of the virtual discontinu-ity of the self-reference effects. Of the 10 self-reference adjectives, 2.84 were correctly recalled.In contrast, only .34, .68, and 1.33 adjectives werecorrectly recalled in the structural, phonemic, andsemantic treatments, respectively.

Another group of similar studies that used rec-ognition memory rather than recall also suggestsstrong participation of affect in information pro-cessing. Formulating their experimental problemin depth-of-processing terms (Craik & Lockhart,1972), Bower and Karlin (1974) showed photo-graphs of faces to subjects with instructions tojudge the photographs for gender, honesty, orlikeability. Following exposures, subjects weretested for recognition memory in two experiments.The hit rate was higher when the subjects ratedphotographs for honesty or likeability than whenthey reported gender. Strnad and Mueller (1977)replicated Bower and Karlin's results in a between-subjects design, and Warrington and Ackroyd(197S) found parallel effects when comparing theseeffects for faces and words, also in a between-sub-jects design. According to Bower and Karlin(1974), deeper processing facilitates recognitionbecause it forces the subject to attend to a greatervariety of detail. "Judgment of honesty of facewould appear to require comparison to an idiosyn-cratic set of vague prototype criteria regardingthe patterning of features such as distance be-tween the eyes, size of pupils, curvature of themouth, thickness of lips, and so on" (p. 7S6).They went on to say that "if you want to re-member a person's face, try to make a number ofdifficult personal judgments about his face whenyou are first meeting him" (pp. 756-757). Pat-terson and Baddeley (1977) asked subjects to dojust that: In one condition their subjects ratedphotographs of faces for the length of nose, dis-tance between the eyes, roundness of face, or full-ness of lips. In another condition the ratings

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 16S

were vague and less detailed but much more likelyto recruit affect: nice-nasty, reliable-unreliable,intelligent-dull, and lively-stolid. Recognitionmemory, as reflected by d' and by hit- and false-alarm rates, was clearly superior for what Patter-son and Baddeley called "personality" ratings.Patterson and Baddeley (1977) thus disagreedwith Bower and Karlin and concluded that theirown "results clearly did not implicate analysis offacial features as a critical or optimal basis forface recognition" (p. 411). Instead, they believedthat should they "ever find an optimum strategyfor encoding of faces, analysis of individual fea-tures is unlikely to be its focus" (p. 417) .

There seems to be general agreement that whenjudgments of pleasantness are made of faces or ofadjectives, individuals engage in forms of deeperinformation processing. What is not agreed uponis the type of content that is accessed at thesedeeper levels. Patterson and Baddeley (1977)doubt that face recognition is based on the sortsof discriminanda that we would intuitively suspectof serving recognition. But if these discriminandaare not the basis of face recognition, what is?Is face recognition, then, based on preferenda?Recall in this respect that the scaling of faces forsimilarity yields pleasantness as the major factor,explaining about 50% of the variance, whereasphysical features play a relatively minor role.Clearly, the contribution of affect to face recogni-tion has been underestimated. Early face dis-crimination is based primarily on affective reac-tions. Infants smile at an approaching face asearly as 10 weeks of age, and at 12 weeks theysmile differently at familiar and unfamiliar faces(Izard, 1978).

Curiously enough, none of the above studiesmention the possible role of affect in processingface information. And none of the studies onrecognition memory of faces collected reactiontime data to verify whether the assumed deeperprocessing was accompanied by longer responselatencies. However, in one recent study, Keenanand Bailett (1979) used methods similar to thoseof Rogers, Kuiper, and Kirker (1977) but insteadadministered recognition memory tests. They re-port results that have an important bearing onthe form of information processing that mayemerge when affect is involved. As in the previousstudies, a number of adjectives were presented,and the subjects were required to check themagainst a number of criteria. For example, sub-

jects were asked whether the given adjective de-scribed themselves, a best friend, a parent, an-other friend, a teacher or boss, a favorite TVcharacter, or Jimmy Carter. Also asked for someadjectives was a semantic encoding question:"Means the same as ?" Following theinitial series, subjects were given a recognitionmemory task in which the original adjectives wereinterspersed among an equal number of similardistractor items. Keenan and Bailett's results arevery clear. Self-reference generated by far thehighest recognition performance (over 90%),whereas reference to Jimmy Carter produced arecognition rate of less than 65%. The otherrecognition rates were arranged according to thesocial signfkance that the target had for the sub-ject: best friend, parent, friend, and teacher.

If the superior recognition memory for the self-reference items was due to deeper processing, onewould expect that response times for these itemswould be longer than response times for itemsprocessed at shallow levels. However, the resultswere quite the opposite and very strikingly so.Encoding times for self-reference items were byfar the shortest. The longest reaction time wasfound for items referred to Jimmy Carter (notethat the experiment was run in 1977 when Carterwas not quite as well known as he is now). More-over, the other targets had response times thatvaried directly with the proportion of correct rec-ognitions.

Keenan and Bailett (1979) attempt a varietyof cognitive interpretations, but at the conclusionof what is truly a valiant effort, they offer thepossibility that in the course of processing self-referent information, "the crucial dimension un-derlying memory is not what the subject knowsor the amount of knowledge that is used in en-coding the item, but rather what the subject feelsabout what he knows" (p. 2 5 ) , It is no longerclear that deeper processing necessarily requiresmore time. Structures that are highly integratedand that have been frequently "tuned in" mayprocess information quite rapidly. The relationbetween reaction time and depth of processing can-not be predicted, therefore, for all tasks (Bad-deley, 1978). Keenan and Bailett's study may betaken as evidence against the levels-of-processingapproach. But it may also be taken as evidencethat the participation of affect in processing in-formation of some types may increase efficiencyto a remarkable degree. The beneficial role of

166 • FEBRUARY 1980 • AMERICAN PSYCHOLOGIST

affect in memory is dramatically illustrated in apaired-associates study in which Sadalla and Loft-ness (1972) asked subjects to form pleasant, un-pleasant, and nonemotional images for each pairand found considerably poorer performance for theneutral pairs than for either the pleasant or theunpleasant pairs.

It is this type of result that suggests the possi-bility of some separation between affect and cog-nition. Consider the task in those experimentswhere the subject is asked to verify if a givenadjective, say "honest," describes him or her. Itis most unlikely that the process of this verifica-tion involves checking the item for its presencein a list, as some information-processing modelswould have it. For one thing, no evidence sug-gests even vaguely that the self is represented asa list of trait adjectives (Markus, 1977). Foranother, the question is probably not interpretedby the subject to mean "Is the trait 'honest' trueof you?" but more likely to mean "Is the trait'honest' consistent with your perception of your-self?" If this is indeed the interpretation thatthe subject imposes upon the task, then we mustinquire what may be meant by "consistent withyour perception of yourself?" To some extentthis consistency may involve absence of contentthat is mutually contradictory; for example, theperson could not be both tall and short. Butmore important, some form of affective consistenc}'is probably involved. That is, the self as usedin this task is probably some global and generalimpression suffused with affective quality. Whatis matched is primarily the affective quality of theitem with the affective quality of the impression.Of course, the shorter processing times for self-referent items may be due to the fact that wehave more integrated and better structured im-pressions of ourselves and of people who are im-portant to us. But it is equally true that theself is a target charged with strong, widespread,and clear affect, and an emotional match wouldtherefore be quite easy for the subject to verify.There is a need in these studies to separate theelaboration and integration of the cognitive struc-ture from the affect that pervades it, but such acontrol procedure is difficult, for the two proper-ties are highly correlated.

That the affective qualities in impression forma-tion are processed differently and perhaps sepa-rately from the cognitive content that "carries"that impression is shown both by Anderson and

Hubert (1963) and by Posner and Snyder (1975b).In a typical impression formation task, the firstauthors found strong primacy effects for impres-sions (i.e., the overall affective rating of the per-son was influenced more by early trait adjectivesin the list) and an equally strong recency effectfor the recall of the adjectives. Anderson andHubert (1963) suggested that "the impressionresponse is based on a different memory systemthan that which underlies the verbal recall" (p.388). They did not go on to specify how thesetwo systems might differ except to say that "aseach adjective is received, its meaning is extractedand combined with the current impression, thusyielding a changed impression. Once this is done,memory for the adjective per se is no longer neces-sary for the impression process" (pp. 390-391).

Dreben, Fiske, and Hastie (1979) found similarorder effects for impressions, and Hamilton, Katz,and Leirer (in press) obtained better recall whensubjects organized items into an impression of aperson than when subjects regarded these itemsas discrete units. More important for the dual-process hypothesis, however, is the finding ofDreben, Fiske, and Hastie that the weights calcu-lated for the adjectives did not predict their recall.That is, the adjectives assumed to be contributingthe most to impression are not necessarily also theones that are best recalled. Following his cogni-tive response theory, Greenwald (1980) suggestedthat cues effective in helping the individual re-trieve content may not be the same ones that areeffective in helping retrieve the evaluative aspectsof the content. It is not unreasonable to supposethat the major difference between these two typesof cues may be the difference between discrimi-nanda and preferenda. And it is perhaps the differ-ence between these cues that is also involved inthe perseverance effect (Ross, Lepper, & Hubbard,1975), in that details of initial information aboutsuccess (or failure) are used only to construct anoverall impression of one's own task competenceand are soon discarded. Thus, in debriefing, whenthe experimenter tells the subjects that their suc-cess (or failure) was rigged, this new informationmay no longer be capable of making contact withthe original input (which by then has been re-coded and discarded) and may therefore havelittle effect on its original affective consequences.

Posner and Snyder (197Sb) also argue for adual memory. In their experiments, subjects areshown a sentence such as "James is honest, loyal,

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 167

and mature," and in a subsequent display a probeword such as "foolish" is flashed. Two tasks arestudied. In one the subject is asked to verify ifthe word itself was among those in the precedingsentence. In another, the required match is be-tween the emotional tone of the word and that ofthe preceding sentence. The interesting resultthese authors obtain is that, as the length of thelist increases, reaction times increase for wordmatching and decrease for emotional tone match-ing. Posner and Snyder (197Sb) agree with An-derson and Hubert about the two memory systemsfor the component adjectives and for the overallimpression, but they doubt that the "emotionalinformation concerning impression is handled inany different way than other semantic dimensionsin the memory system" (p, 80). Their doubtsshould be weakened by a recent impression-forma-tion experiment in which the pattern of recall ofindividual adjectives was effectively manipulatedin the hope of thereby affecting the primacy ofimpressions. Riskey (1979) was able to changethe recall of adjectives, but the primacy of impres-sions nevertheless remained unchanged.

While these authors propose separate systems,it is always separate cognitive systems that theypropose. In contrast, the separation being con-sidered here is between an affective and a cogni-tive system—a separation that distinguishes be-tween cliscriminanda and preferenda and that takesus back to Wundt and Bartlett, who speculatedthat the overall impression or attitude has anexistence of its own, independent of the compo-nents that contributed to its emergence. Thequestion that cannot be answered with the datathus far collected is whether the affect-contentseparation is simply a matter of separate storage(as Anderson and Hubert, on the one hand, andPosner and Snyder, on the other, have proposed)or whether there isn't some separation already atthe point of registration and encoding. The rapidprocessing times of affect suggest a more completeseparation of the two processes at several junc-tures.

One is necessarily reminded in this context ofthe dual coding hypothesis proposed by Paivio(1975) for the processing of pictures and words.Paivio (1978a) suggested a number of differencesbetween the processing of these types of content,for example, that representations of picturesemerge as perceptual isomorphs or analogs (ima-gens), whereas parallel units in the verbal system

are linguistic components (logogens). He alsoproposed that pictorial information is organizedin a synchronous and spatially parallel manner,whereas verbal information is discrete and se-quential. Finally, he suggested that the process-ing of pictures is more likely to be the business ofthe right-brain hemisphere, whereas the processingof words is the business of the left. Paivio's pro-posal for a dual coding theory kindled a contro-versy of some vigor. While Anderson (1978) hasrecently argued that the controversy cannot beresolved with what we now know about theseprocesses, it has nevertheless stimulated some ex-citing empirical and theoretical work (e.g., Banks& Flora, 1977; Kerst & Howard, 1977; Kosslyn &Pomerantz, 1977; Paivio, 1978b; Pylyshyn, 1973;Shepard, 1978).

Most relevant for my discussion, however, isPaivio's (1978c) finding that reaction times forpleasant-unpleasant ratings are faster for picturesthan for words. Paivio takes this result to indi-cate that "the analog information involved inpleasantness and value judgments is more closelyassociated with the image system than with theverbal system" (p. 207) . Thisanalog pleasantness information is "carried by" affectiveand motor processes that are closely associated with visualmemory representations of things. Such processes pre-sumably originate as reactions to things and persist asaffective or motor memories that can be activated' bypictures of the referent objects, or, more indirectly, by theirnames when accompanied by the appropriate contextualcues. More specifically, pleasantness and value judgmentsmight be based on continuously variable interoceptive re-actions and approach or avoidance tendencies that areactivated jointly by the comparison stimuli and the taskinstructions, (p. 207)

However, the specific responses of the autonomicnervous system are not readily discriminable, sincethere are not many receptors to register the finechanges in autonomic processes (Averill, 1969;Mandler, Mandler, Kremen, & Sholiton, 1961).Moreover, interoceptive process and motor memo-ries are slower than the affective responses theyare presumed to activate.

It is a fact, of course, that all sorts of judg-ments are faster and more efficient for picturesthan for words, and this may be so just becausepictures are able to evoke an affective reactionmore directly and faster than words. An affectivereaction aroused early in the encoding process—earlier than it is possible for the interoceptive andmotor memories to become effective—might facili-tate a complex cognitive encoding sequence by aninitial categorization along affective lines, which,

168 • FEBRUARY 1980 • AMERICAN PSYCHOLOGIST

as we have seen, requires minimal stimulus in-formation. Such facilitation through early affec-tive sorting that relies not only on discriminandabut on preferenda as well may also induce a con-structive process that can more readily recruitstored content by searching for congruent affectivetags,10

This review suggests that a separation betweenaffect and cognition may well have a psychologicaland a biological basis.17 Recall that in contrastwith cold cognitions, affective responses are effort-less, inescapable, irrevocable, holistic, more diffi-cult to verbalize, yet easy to communicate and tounderstand. Consider also that the processing ofaffect is probably an even stronger candidate forthe right hemisphere than the processing of pic-tures (Carmon & Nachson, 1973: Dimond, Far-rington, & Johnson, 1976; Ley & Bryden, 1979;Milner, 1968; Safer & Leventhal, 1977; Schwartz,Davidson, & Maer, 1975). In the context of thisreview it is especially interesting (a) that facerecognition is superior when the stimuli are pre-

18 Another area of research • in which affect may beimplicated (although it had not been so suspected) is thefrequency-judgment paradigm. Typically, in these ex-periments .subjects are shown stimuli in different fre-quencies, and two types of judgments are collected after-wards. In one condition, the subjects are shown the oldstimuli interspersed among new ones and are asked toreport for each item whether it is new or old. In theother condition, the subjects must say how often eachstimulus occurred. It turns out that the frequency judg-ment generates greater accuracy than the binary recogni-tion memory judgment (e.g., Proctor, 1977; Proctor &Ambler,'1975).

Two findings are of interest in the present context.First, subjects have remarkable confidence in their fre-quency judgments (Howell, 1971). Second, warning thesubjects that they will be estimating frequencies of events(vs. simply recalling them) and varying the length ofthe list both influence free recall but have little if anyeffect on frequency estimation (Howell, 1973). It thusappears that frequency judgments behave like affectivejudgments. It is possible, therefore, that frequency esti-mation is more likely to invoke an underlying affectivereaction (which accrues from repeated stimulus exposures)than the binary recognition memory task. It may behotter. Since frequency judgment makes exposure effectssalient and since it requires finer discrimination thanrecognition memory, it may recruit affect as an auxiliarysource of information. In fact, it has been suggestedthat recognition memory responses and frequency estima-tion are not made from the same sources of information(Wells, 1974), although what these sources are and howthey differ from each other is not altogether clear (Hintz-man, 1976).

17 Multiple processing systems and multiple channel con-ceptions are today more the rule than the exception inthe study of sensory processes (Graham & Nachmias,1971; Trevarthen, 1968).

sented in the left visual field (De Renzi Si Spinn-ler, 1966; Moscovitch, Scullion, & Christie, 1976),and (b) that the recognition of emotional ex-pressions shows the same right-brain superiority(Suberi & McKeever, 1977).

It has also been suggested to me by Richard J.Katz (Note 5) that there exists a network in thecentral nervous system, the locus coeruleus, whichis ideally suited for the kind of partially inde-pendent processing of affect that I have suggestedhere. The potential sensitivity of the locus coeru-leus to preferenda can be inferred from a numberof interesting properties and features of this sys-tem. Above all, it is excited differently by noveland by familiar stimuli. Second, self-stimulationstudies have demonstrated that the locus coeruleusis sensitive to incentives. It is further knownthat it is capable of innervating sensory areas(such as the colliculi and geniculate bodies), emo-tional areas (the amygdala and hypothalamus),mnestic areas (the hippocampus), and the cere-bral cortices. Most important, however, is thefact that the locus coeruleus is capable of veryfast responding. Finally, Katz also noted that theenkephalenergic system, which controls the actionof enkephalins (naturally occurring opiates) andis situated at the locus coeruleus, is also involvedin reinforcement and in different reactions tonovelty and familiarity. All of this means, at thevery least, that what I have proposed about theprocessing of affect is not inconsistent with recentknowledge about the relevant neurophysiologicalmechanisms. It means that the organism isequipped with a neurochemical apparatus capableof telling the new from the old and the good fromthe bad, of remembering the old, the good, andthe bad, and of making all these decisions rapidlywithout having to wait for the slow feedback fromthe autonomic system.

Affective reactions are primary in ontogeny.The infant knows to cry and to smile long beforeit acquires any semblance of verbal skills (Izard,1978, 1979). Meltzoff and Moore (1977) reportthat human infants can imitate emotional expres-sions at 12 days of age, long before they acquirelanguage. And good-bad is one of the very firstdiscriminations that children learn.

More important, however, affect is clearly pri-mary in philogeny. Affect was there before weevolved language and our present form of thinking.The limbic system that controls emotional reac-tions was there before we evolved language and

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 169

Figure 5. Time course of the stimulus, sensa-tion, affect, and cold cognitions (R = response).

our present form of thinking. It was there beforethe neocortex, and it occupies a large proportionof the brain mass in lower animals. Before weevolved language and our cognitive capacities,which are so deeply dependent on language, it wasthe affective system alone upon which the organ-ism relied for its adaptation. The organism's re-sponses to the stimuli in its environment wereselected according to their affective antecedentsand according to their affective consequences.Thus, if the most recent version of homo sapiensspecifies that affective reactions are mediated byprior cognitive processes—as contemporary cog-nitive views would have it—then at some pointin the course of evolution, affect must have lostits autonomy and acquired an intermediary in theform of cold cognition. This scenario seems mostunlikely. When nature has a direct and autono-mous mechanism that functions efficiently—andthere is no reason to suppose that the affectivesystem was anything else—it does not make itindirect and entirely dependent on a newly evolvedfunction. It is rather more likely that the affec-tive system retained its autonomy, relinquishingits exclusive control over behavior slowly andgrudgingly. At most, the formerly sovereign af-fective system may have accepted an alliance

with the newly evolved system to carry out someadaptive functions jointly. These conjecturesmake a two-system view more plausible than onethat relegates affect to a secondary role mediatedand dominated by cognition.

Because it is so heavily rooted in verbal skills,the cognitive system in humans has properties thatare quite distinct from those of affect. Above all,the cognitive system is infinitely more diverse andflexible than the affective system. Anything atall can be said and thought with various degreesof precision, and these things can be said andthought in an infinite variety of ways. But thereare only a handful of emotions and feelings thatcan be felt, and they can be felt only in somefew, very constrained ways. And for reasons thatmust be rooted in the partial separation of thetwo systems, affect can be communicated muchmore efficiently and accurately than thought inspite of the fact that its vocabulary is quite lim-ited. It was a wise designer who provided sepa-rately for each of these processes instead of pre-senting us with a multiple-purpose appliance that,like the rotisserie-broiler-oven-toaster, performsnone of its functions well.

Conclusion

It is too early to write a model for affect and forthe various ways that it interacts with cold cog-nitions. The important pieces of evidence are stillmissing. However, we can begin to specify thefacts that such a model must accommodate. Fig-ure 5 summarizes these facts by schematizing thetime course of the stimulus together with theensuing sensory process, the affective response,and some simple aspects of the cognitive process(recognition and feature identification). A stimu-lus is presented for a fixed time interval. Thestimulus triggers a number of processes that canvary in their onset times and offset times. I haveshown these processes as ranges of their onsettimes, ignoring the offset times altogether for thepresent purposes. (I have also ignored the factthat under some conditions, stimulus onset canbe anticipated by the response process.) The on-set times of these four processes are influenced bystimulus conditions and by subject states (e.g.,previous experience with stimuli of the given class,exposure to immediately preceding stimuli thatmay generate contrast or assimilation, knowledge,mood states, priming, or expectation.

170 • FEBRUARY 1980 • AMERICAN PSYCHOLOGIST

Note that a variety of temporal relations holdsamong affect, recognition, and feature identifica-tion. Of course, sensory process must have theearliest onset. Its onset times, too, differ depend-ing on the stimulus, level of attention of the or-ganism, the peripheral processes that are activated,context, etc. Also, an affective reaction alwaysdirectly follows the sensory input. In RI thisreaction is strong and salient, and it might domi-nate the ensuing cognitive process. In R2 and R3,affect .(shown in broken circles) is also arousedimmediately following the sensory process, but it isweak and does not significantly influence the sub-sequent stages of the cognitive process. But for af-fect, recognition, and feature discrimination, all com-binations are possible. I have given examples ofthree of the six possible response patterns. In RI,affect is first, recognition occurs later, and fea-ture discrimination is last. The primacy of affectover recognition in RI reflects our own data(Kunst-Wilson & Zajonc, in press) and the resultsof typical subliminal perception experiments (e.g.,Blum & Barbour, 1979; Shevrin & Fritzler, 1968).The difference between recognition and featurediscrimination that favors the former reflects theresults of Patterson and Baddeley (1977), who,it will be recalled, found that subjects could recog-nize photographs better when they judged themon "personality" characteristics than when theyjudged them on specific physical features. Theresults of Marcel (1976) and of Keenan andBailett (1979) also suggest that recognition canprecede feature identification.18

RQ presents the case typically considered byinformation-processing models: The cognitive pro-cess begins with the individual first discriminatinga critical feature that allows recognition. Andfinally it is recognition that gives rise to the af-fective response. In Rs, recognition precedes fea-ture identification, as is the case when lettersthat form words are recognized better than lettersthat do not form words (Johnston & McClelland,1974) or when meaning is apprehended while theword itself cannot be identified, as in the para-lexic response of certain aphasic patients (Mar-shall & Newcombe,. 1966).

Figure 5 shows the lower temporal limits ofthese reactions. Except for the sensory process,affect is assumed to be capable of the earliest on-set. How can that be?

Perhaps the following analysis, speculative tobe sure, may point to some possible answers. De-

cisions about affect require the least informationand are often based on a different decision schemethan either recognition or feature identification.Each of the three, affect, recognition, and featureidentification, is a form of categorization. Affec-tive reactions of the type considered here are forthe most part unidimensional and sometimes justbinary: safe-dangerous, good-bad, or nice-nasty.Such binary decisions can under some circum-stances be made quite reliably, even in the ab-sence of reliable bases. Moore and Shannon(1956) have shown that reliable circuits can beconstructed using arbitrarily unreliable relays, pro-vided the relays form parallel circuits that aremutually redundant. Zajonc and Smoke (1959)applied this principle to group performance, andSmoke and Zajonc (1962) to group decisions.That is, given certain group decision processes,groups can make judgments much more reliablythan the average group member. An analogoussituation may well exist for affect where the stimu-lus triggers several parallel responses and the de-cision scheme can well be a minimal quorum.10

Recognition, however, even though it also consti-tutes a binary choice (old-new), does not have asimilar advantage because the redundancy of thecomponent criteria (e.g., features of configuralproperties) is seldom as high as in the case ofaffect. Moreover, minimal quorum is seldom adecision basis. In fact, in experimental work onrecognition memory, great care is taken to assurethat criterial features are fairly independent ofeach other. Just because a photograph shows theface of a male is not sufficient for calling it "old"or rejecting it as "new," unless previous expo-sures have shown only female faces.

Each of the responses in Figure 5 can facili-tate the ones succeeding it. An affective reactioncan thus act to precategorize the stimulus for thesubject trying to decide whether it is "old" or"new." And recognition may facilitate featureidentification by a similar prior selection process.Each operation reduces the universe of alterna-tives for the next choice. It is therefore entirelypossible for stimuli that have a strong effectivepotential to evoke affective reactions rapidly, to berecognized sooner than neutral stimuli, and to beremembered better. Thus, Figure S also shows

18 Similar effects are obtained in vision (e.g., Graham& Nachmias, 1971).

10 In the case of decisions or parallel signalling circuitssuch as may be involved in affective reactions, the mini-mal quorum is equivalent to a veto decision or its inverse.

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 171

the representations in memory that are left by thethree processes (affect, recognition, and featureidentification), To the extent that these traces areredundant, the likelihood of a later retrieval wouldbe facilitated.

I began this paper with a quotation from Wundt,and it must be apparent that another spirit hasemerged as I have developed my arguments—that of Freud. The separation of affect and cog-nition, the dominance and primacy of affectivereactions, and their ability to influence responseswhen ordinary perceptual recognition is at chancelevel are all very much in the spirit of Freud, thechampion of the unconscious. In terms of myformulation, there seem to be at least two differentforms of unconscious processes. One emergeswhere behavior, such as that occurring in discrimi-nation among stimuli, is entirely under the influ-ence of affective factors without the participationof cognitive processes. Included here are suchphenomena as perceptual defense and vigilance,subliminal perception and discrimination, state de-pendent recall, and mood and context effects. An-other form of unconscious process is implicated inhighly overlearned, and thus automated, sequencesof information processing; this form includes cog-nitive acts but has collapsed them into largermolar chunks that may conceal their original com-ponent links (cf. Shiffrin & Schneider, 1977). Andthere may be other forms of process in which theseparation between affect and cognition preventsthe individual from apprehending the potentialconnection between them.

Because the language of my paper has beenstronger than can be justified by the logic of theargument or the weight of the evidence, I hastento affirm that one of my purposes was to convinceyou that affect should not be treated as unalter-ably last and invariably postcognitive. The evo-lutionary origins of affective reactions that pointto their survival value, their distinctive freedomfrom attentive control, their speed, the importanceof affective discriminations for the individual, theextreme forms of action that affect can recruit—all of these suggest something special about affect.People do not get married or divorced, commitmurder or suicide, or lay down their lives for free-dom upon a detailed cognitive analysis of the prosand cons of their actions. If we stop to considerjust how much variance in the course of our livesis controlled by cognitive processes and how muchby affect, and how much the one and the other

influence the important outcomes in our lives, wecannot but agree that affective phenomena deservefar more attention than they have received fromcognitive psychologists and a closer cognitivescrutiny from social psychologists,

REFERENCE NOTES

1. Ellsworth, P. Personal communication, May 13,- 1979.2. Zajonc, R. B. Preferenda and discriminanda: Process-

ing of affect. Paper presented at the First OntarioSymposium on Personality and Social Psychology, Lon-don, Ontario, Canada, August 1978.

3. Premack, D., & Kintsch, W. Personal communication,July 1979.

4. Marcel, J. Unconscious reading: Experiments on peoplewho do not know that they are reading. Paper pre-sented to the British Association for Advancement ofScience, Lancaster, England, 1976.

5. Katz, R. J. Personal communication, February 1979.

REFERENCES

Abelson,- R. P. Computer simulation of "hot cognitions."In S. Tomkins & S. Mesick (Eds.), Computer simula-tion of personality. New York: Wiley, 1963.

Abelson, R. P., & Rosenberg, M. J. Symbolic psycho-logic: A model of attitudinal cognition. BehavioralScience, 1958, J, 1-13.

Abelson, R. P., & Sermat, V. Multidimensional scalingof facial expressions. Journal of Experimental Psy-chology, 1962, 63, 546-554.

Anderson, J. R. Language, memory, and thought. Hills-dale, N.J.: Erlbaum, 1976.

Anderson, J. R. Arguments concerning representationsfor mental imagery. Psychological Review, 1978, 85,249-277.

Anderson, J. R., & Bower, G. H. Human associativememory. Washington, D.C.: Winston, 1973.

Anderson, N. H., & Hubert, S. Effects of concomitantverbal recall on order effects in personality impressionformation. Journal of Verbal Learning and Verbal Be-havior, 1963, 2, 379-391.

Argyle, M., Salter, V., Nicholson, H., Williams, M., &Burgess, P. The communication of inferior and su-perior attitudes by verbal and non-verbal signals.British Journal of Social and Clinical Psychology, 1970,9, 222-231.

Averill, J. R. Autonomic response patterns during sad-ness and mirth. Psychophysiology, 1969, S, 399-414.

Baddeley, A. D. The trouble with levels: A re-examina-tion of Craik and Lockhart's framework for memoryresearch. Psychological Review, 1978, 85, 139-152.

Banks, W. P., & Flora, J. Semantic and perceptual pro-cesses in symbolic comparisons. Journal of Experi-mental Psychology: Human Perception and Perform-ance, 1977, 3, 278-290.

Bartlett, F. C. Remembering: A study in experimentaland social psychology. Cambridge, England: CambridgeUniversity Press, 1932.

Berlyne, D. E. Arousal and reinforcement. In D. Le-vine (Ed.), Nebraska Symposium on Motivation (Vol.15). Lincoln: University of Nebraska Press, 1967.

Berlyne, D. E. Novelty, complexity, and hedonic value.Perception and Psychophysics, 1970, 8, 279-286.

Berlyne, D. E. Dimensions of perception of exotic andpre-renaissance paintings. Canadian Journal of Psy-chology, 1975, 29, 151-173.

172 • FEBRUARY 1980 • AMERICAN PSYCHOLOGIST

Berscheid, E., & Walster, E. Interpersonal attraction.Reading, Mass.: Addison-Wesley, 1978.

Blum, G. S., & Harbour, J. S. Selective inattention toanxiety-linked stimuli. Journal of Experimental Psy-chology: General, 1979, 108, 182-224.

Bobrow, D. G., & Collins, A. Representation and under-standing. New York: Academic Press, 1975,

Bower, G. H., & Karlin, M. B. Depth of processing pic-tures of faces and recognition memory. Journal of Ex-perimental Psychology, 1974, 103, 751-757.

Broadbent, D. E. The hidden preattentive processes.American Psychologist, 1977, 32, 109-118.

Broadbent, D. E., & Gregory, M. H. P. The perceptionof emotionally toned words. Nature, 1967, 215, 581-584.

Byrne, D. Interpersonal attraction and attitude simi-larity. Journal o.f Abnormal and Social Psychology,1961, 62, 713-715.

Carmon, A., & Nachson, I. Ear asymmetry in perceptionof emotional non-verbal stimuli. Acta Psychologica,1973, 37, 351-357.

Coombs, C. H. A theory of data. New York: Wiley,1964.

Cooper, L. G. A multivariate investigation of preferences.Multivariate Behavioral Research, 1973, 8, 253-272.

Craik, F. I. M., & Lockhart, R. S. Levels of processing:A framework for memory research. Journal of VerbalLearning and Verbal Behavior, 1972, 11, 671-684.

Crowder, R; G. Principles of learning and memory.Hillsdale, N.J.: Erlbaum, 1976.

Cummings, E. E. Complete poems (Vol. I). Bristol,England: McGibbon & Kee, 1973.

Dawes, R. M., & Kramer, E. A proximity analysis ofvocally expressed emotion. Perceptual and Motor Skills,1966,.22, 571-574,

De Renzi, E., & Spinnler H. Facial recognition in brain-damaged patients. Neurology, 1966, 16, 145-152.

Dimond, S. J., Farrington, L., & Johnson, P. Differingemotional response from right and left hemisphere.Nature, 1976, 261,,690-692.

Dreben, E. K., Fiske, S, T., & Hastie, R. The. inde-pendence of evaluative and item information: Impres-sion and recall order effects in behavior-based impres-sion formation. Journal of Personality and Social Psy-chology, 1979, 37, 1758-1768.

Ekman, P., & Friesen, W. V. The repertoire of nonverbalbehavior: Categories, origins, usage, and coding. Semi-otica, 1969,1, 49-98.

Erdelyi, M. H. A new look at the New Look: Perceptualdefense and vigilance. Psychological Review, 1974, 81,1-25. _ . .

Estes, W. K. (Ed.). Handbook of learning and cognitive. processes (Vols. 1-6). Hillsdale, N.J.: Erlbaum, 1975-

1978.Festinger, L. Conflict., decision, and dissonance. Stanford,

Calif.: Stanford University - Press, 1964.Gainotti, G. Emotional behavior and hemispheric side of

the lesion. Cortex, 1972, 8, 41-55.Graham, N., & Nachmias, J. Detection of grating pat-

terns containing two spatial frequencies: A comparisonof single-channel and multiple-channels models. VisionResearch, 1971, 11, 251-259.

Greenwald, A. G, Cognitive response analysis: An ap-praisal. In R. E. Petty, T. M. Qstrom, & T. C. Brock(Eds.), Cognitive responses in persuasive communica-tion. Hillsdale, N.J.: Erlbaum, 1980.

Hamilton, D. L., Katz, L. B., & Leirer, V. 0. Organiza-tional processes in impression formation. In R. Hastie,T. Qstrom, E. Ebbesen, R. Wyer, D. Hamilton, & D.

Carlston (Eds.), Person memory. Hillsdale, N.J.: Erl-baum, in press.

Harrison, A. A. Response competition, frequency, ex-ploratory behavior, and liking. Journal of Personalityand Social Psychology, 1968, 0, 363-368.

Harrison, A. A. Mere exposure. In L. Berkowitz (Ed.),Advances in experimental social psychology (Vol. 10),New York: Academic Press, 1977.

Hastorf, A. H., Osgood, C. E,, & Ono, H. The semanticsof facial expressions and the prediction of the meaningsof stereoscopically fused facial expressions. ScandinavianJournal of Psychology,- 1966, 7, 179-188.

Heider, F. The psychology of interpersonal relations.New York: Wiley,-1958.

Hintzman, D. L. Repetition and memory. In G. H.Bower (Ed.), The psychology of learning and motiva-tion (Vol. 10). New York: Academic Press, 1976.

Howell, W. C. Uncertainty from internal and externalsources: A clear case of overconfidence. Journal ofExperimental Psychology, 1971, 89, 240-243.

Howell, W. C. Representation of frequency in memory.Psychological Bulletin, 1973, 80, 44-53.

Hume, D. A treatise on human nature (Vol. 2). Lon-don: Longmans, Green, 1898.

Hyde, T. W., & Jenkins J. J. The differential effects ofincidental tasks on the organization of recall of a listof highly associated words. Journal of ExperimentalPsychology, 1969, 82, 472-481.

Inhelder, B., & Piaget, J. The growth of logical thinkingfrom childhood to adolescence. New York: Basic Books,1958.

Ittelson, W. H. Environment perception and contempo-rary perceptual theory. In W. H. Ittelson (Ed.), En-vironment and cognition. New York: Seminar Press,1973.

Izard, C. E. Human emotions. New York: PlenumPress, 1977.

Izard, C. E. On the development of emotions and emo-tion-cognition relationship in infancy. In M. Lewis &L. Rosenblum (Eds.), The development of affect. NewYork: Plenum Press, 1978.

Izard, C. E. Emotions as motivations: An evolutionary-developmental perspective. In R. Dienstbier (Ed.),Nebraska Symposium on Motivation (Vol. 27). Lin-coln: University of Nebraska Press, 1979.

Johnston, J. C., & McClelland, J. L. Perception of let-ters: Seek not and ye shall find. Science, 1974, 184,1192-1194,

Joreskog, K. G., & Sb'rbom, D. Statistical models andmethods for analysis of longitudinal data. In D. J. Aig-ner & A. S. Goldberger (Eds.), Latent variables insocio-economic models. Amsterdam: North-Holland,1977.

Kahneman, D., & Tversky, A. Prospect theory: An anal-. ysis of decision under risk. Econometrica, 1979, 47,

263-291.Kcenan, J. M., & Bailett, S. D. Memory for personally

and socially significant events. In R. S. Nickerson(Ed.), Attention and performance VIII. Hillsdale, N.J.:Erlbaum, 1979.

Kerst, S. M., & Howard, J. H., Jr. Mental comparisonsfor ordered information on abstract and concrete di-mensions. Memory and Cognition, 1977, 5, 227-234.

Kintsch, W. The representation of meaning in memory.Hillsdale, N.J.: Erlbaum, 1974.

Klahr, D. Decision making in a complex environment:The use of similarity judgments to predict preferences,Management Science, 1969, 15, 595-618.

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 173

Kosslyn, S. M., & Pomerantz, J. R. Imagery, proposi-tions, and the form of internal representations. Cogni-tive Psychology, 1977, 9, 52-76.

Kunst-Wilson, W. R., & Zajonc, R. B. Affective dis-crimination of stimuli that cannot be recognized. Sci-ence, 1980, 207, 5S7-SS8.

Lachman, R., Lachman, J. L., & Butterfield, E. C. Cog-nitive psychology and information processing. Hills-dale, N.J.: Erlbaum, 1979.

Lang, P. J. A bio-informational theory of emotionalimagery. Psychophysiology, 1979, 16, 495-512.

Langer, S. K. Mind: An essay on human feeling (Vol. 1).Baltimore, .Md.: Johns Hopkins University Press, 1967.

Ley, R. G., & Bryden, M. P. Hemispheric differences inprocessing emotions and faces. Brain and Language,1979, 7, 127-138.

Handler, G. Mind and emotion. New York: Wiley, 1975.Mandler, G., Handler, J. M., Kremen, I., & Sholiton,

R. D. The response to threat: Relations among verbaland physiological indices. Psychological Monographs,1961, 75(9, Whole No. 513).

Markus, H. Self-schemata and processing of informationabout the self. Journal of Personality and Social Psy-chology, 1977, 35, 63-78.

Marshall, J. C., & Newcombe, F. Syntactic and semanticerrors in paralexia. Neuropsychologia, 1966, 4, 169-176.

Matlin, M. W. Response competition, recognition, andaffect. Journal of Personality and Social Psychology,1971, 19, 295-300.

Meltzoff, A. N., & Moore, M. K. Imitation of facial andmanual gestures by human neonates. Science, 1977, 198,75-78.

Miller, G. A., & Johnson-Laird, P. N. Language andperception. Cambridge, Mass.: The Belknap Press ofHarvard University Press, 1976.

Milner, B. Visual recognition and recall after righttemporal-lobe excision in man. Neuropsychologia, 1968,6, 191-209.

Milord, J. T. Aesthetic aspects of faces: A (somewhat)phenomenological analysis using multidimensional scal-ing methods. Journal of Personality and Social Psy-chology, 1978, .3(5, 205-216.

Moore, E. F., & Shannon, C. E. Reliable circuits usingless reliable relays. Part I. Journal of the FranklinInstitute, 1956, 262, 191-208.

Moreland, R. L., & Zajonc, R. B. Is stimulus recognitiona necessary condition for the occurrence of exposureeffects? Journal of Personality and Social Psychology,1977, 35, 191-199.

Moreland, R. L., & Zajonc, R. B. Exposure effects maynot depend on stimulus recognition. Journal of Per-sonality and Social Psychology, 1979, 37, 1085-1089.

Moscovitch, M., Scullion, D., & Christie, D. Early versuslate stage of processing and their relation to functionalhemispheric asymmetries in face recognition. Journal ofExperimental Psychology: Human Perception and Per-formance, 1976, 2, 401-416.

Nakashima, T. Contribution to the study of the affectiveprocesses. American Journal of Psychology, 1909, 20,157-193. (a)

Nakashima, T. Time-relations of the affective process.Psychological Review, 1909, 16, 303-339. (b)

Neisser, U. Cognitive psychology. Englewood Cliffs,N.J.: Prentice-Hall, 1967.

Neisser, U. Cognition and reality. San Francisco: Free-man, 1976.

Nisbett, R. E., & Wilson, T. D. Telling more than wecan know: Verbal reports on mental processes. Psycho-logical Review, 1977, 84, 231-259.

Norman, D. A., & Rumelhart, D. E. Explorations in cog-nition. San Francisco: Freeman, 1975.

O'Hare, D. Individual differences in perceived similarityand preference for visual art: A multidimensional scalinganalysis. Perception & Psychophysics, 1976, 20, 445-452.

Osgood, C. E. Studies on the generality of affectivemeaning systems. American Psychologist, 1962, 17, 10-28.

Osgood, C. E. On the whys and wherefores of E, P, andA. Journal of Personality and Social Psychology, 1969,12, 194-199.

Osgood, C. E., & Tannenbaum, P. H, The principle ofcongruity in the prediction of attitude change. Psy-chological Review, 1955, 62, 42-55.

Paivio, A. Perceptual comparisons through the mind'seye. Memory and Cognition, 1975, 3, 635-647.

Paivio, A. Dual coding: Theoretical issues and empiricalevidence. In J. M. Scandura & C. J. Brainerd (Eds.),Structural/process models of complex human behavior.Leiden, The Netherlands: Nordhoff , 1978. (a)

Paivio, A. Images, propositions, and knowledge. In J.M. Nicholas (Ed.), Images, perception, and knowledge.The Western Ontario Series in Philosophy of Science(No. 8). Dordrecht, The Netherlands: Reidel, 1978. (b)

Paivio, A. Mental comparisons involving abstract attri-butes. Memory and Cognition, 1978, 3, 199-208. (c)

Patterson, K. E., & Baddeley, A. D. When face recogni-tion fails. Journal of Experimental Psychology: Hu-man Learning and Memory, 1977, 3, 406-417.

Posner, M. L, & Snydcr, C. R. R. Attention and cognitivecontrol. In R. L. Solso (Ed.), Information processingand cognition: The Loyola Symposium. Hillsdale,N.J.:Erlbaum, 1975. (a)

Posner, M. L, & Snyder, C. R. R. Facilitation and in-hibition in the processing of signals. In P. M. A.Rabbitt & S. Dornic (Eds.), Attention and performanceV. New York: Academic Press, 1975. (b)

Pratt, C. L., & Sackett, G. P. Selection of partners as afunction of peer contact during rearing. Science, 1967,155, 1133-1135.

Premack, D. Intelligence in ape and man. Hillsdale,N.J.: Erlbaum, 1976.

Proctor, R. W. The relationship of frequency judgmentsto recognition: Facilitation of recognition and compari-son to recognition-confidence judgments. Journal ofExperimental Psychology: Human Learning and Mem-ory, 1977, 3, 679-689.

Proctor, R. W., & Ambler, B, A. Effects of rehearsalstrategy on memory for spacing and frequency. Jour-nal of Experimental Psychology: Human Learning andMemory, 1975, 1, 640-647.

Pylyshyn, Z. W. What the mind's eye tells the mind'sbrain: A critique of mental imagery. PsychologicalBulletin, 1973, 80, 1-24.

Quandt, R. E. A probabilistic theory of consumer be-havior. Quarterly Journal of Economics, 1956, 70, 507-536.

Riskey, D. R. Verbal memory process in impressionformation. Journal of Experimental Psychology: Hu-man Learning and Memory, 1979, 5, 271-281.

Rogers, T. B., Kuipcr, N. A., & Kirker, W. S. Self-refer-ence and the encoding of personal information. Journalof Personality and Social Psychology, 1977, 35, 677-688.

Ross, L., Lepper, M. R., & Hubbard, M. Perseverance inself-perception and social perception: Biased attribu-tional processes in the debriefing paradigm. Journal ofPersonality and Social Psychology, 1975, 32, 880-892.

Rubin, Z. Liking and loving. New York: Holt, Rine-hart & Winston, 1973.

174 • FEBRUARY 1980 • AMERICAN PSYCHOLOGIST

Sadalla, E. K,, & Lo.ftness, S. Emotional images as medi-ators, in one-trial paired-associates learning. Journal ofExperimental Psychology, 1972, 95, 295-298.

Safer, N. A., & Leventhal, H. Ear differences in evaluat-ing emotional tones of voice and verbal content. Jour-nal of Experimental Psychology: Human Perception andPerformance, 1977, 3, 75-82.

Sartre, J. P. Existentialism. New York: PhilosophicalLibrary, 1947.

Schachter, S,, & Singer, J. Cognitive, social, and physio-logical determinants of emotional state. PsychologicalReview, 1962, 65, 379-399.

Schank, R. C., & Abelson, R. P. Scripts, plans, goals, andunderstanding. Hillsdale, N.J.: Erlbaum, 1977.

Scherer, K. R., Koivumaki, J., & Rosenthal, R, Minimalcues in the vocal, communication of affect: Judging emo-tions from content-masked speech. Journal of Psycho-linguistic Research, 1972, 1, 269-285.

Schneider, D. J., Hastorf, A. H., & Ellsworth, P. C. Per-son perception. Reading, Mass.; Addison-Wesley, 1979.

Schwartz, G. E., Davidson, R. J., & Maer, F. Righthemisphere lateralization for emotion in the humanbrain: Interactions with cognition. Science, 1975, 190,286-288.

Schwartz, G. E., Fair, P. L., Salt, P., Mandel, M. R., &Klerman, G. L. Facial muscle patterning to affectiveimagery in depressed and nondepressed subjects. Sci-ence, 1976, 192, 489-491.

Shepard, R. N. The mental image. American Psycholo-gist, 1978, 33, 125-137.

Shevrin, H., & Fritzler, D. E. Visual evoked responsecorrelates of unconscious mental process. Science, 1968,161, 295-298.

Shevrin, H., Smith, W. H., & Fritzler, D. E. Averageevoked response and verbal correlates of unconsciousmental processes. Psychophysiology, 1971, S, 149-162.

Shiffrin, R. M., & Schneider, W. Controlled and auto-matic human information processing: II. Perceptuallearning, automatic attending, and a general theory.Psychological Review, 1977, 84, 127-190.

Smoke, W. H., & Zajonc, R. B. On the reliability ofgroup judgments and decisions. In J. Criswell, H. Solo-mon, & P. Suppes (Eds.), Mathematic methods in smallgroup process. Stanford, Calif.: Stanford UniversityPress, 1962. .

Strnad, B. N., & Mueller, J, H. Levels of processing in

facial recognition memory. Bulletin of the Psyc/ionomicSociety, 1977, 9, 17-18.

Suberi, M., & McKeever, W. F. Differential right hemi-spheric memory storage of emotional and non-emotionalfaces. Nemopsychologia, 1977, 15, 757-768.

Titchener, E. B. A textbook of psychology. New York:Macmillan, 1910.

Tomkins, S. S. Affect, imagery, consciousness: Vol. 1.The positive affects. New York: Springer, 1962.

Tomkins, S. S. Affect, imagery, consciousness: Vol. 2.The negative affects. New York: Springer, 1963.

Trevarthen, C. B. Two mechanisms of vision in primates.Psychologische Forschimg, 1968, 31, 299-337.

Tulving, E., & Donaldson, VV. Organization oj memory.New York: Academic Press, 1972.

Vitz, P. C. Preferences for rates of information presentedby sequences of tones. Journal of Experimental Psy-chology, 1964, 68, 176-183.

Warrington, E. K., & Ackroyd, C. The effect of orientingtasks on recognition memory. Memory and Cognition,1975, 3, 140-142.

Wells, J. E. Strength theory and judgments of recencyand frequency. Journal of Verbal Learning and VerbalBehavior, 1974, 13, 378-392.

Wilson, W. R. Unobtrusive induction of positive attitudes.Unpublished doctoral dissertation, University of Michi-gan, 1975.

Wilson, W. R. Feeling more than we can know: Expo-sure effects without learning. Journal of Personality andSocial Psychology, 1979, 37, 811-821.

Wundt, W. Grundriss der Psychologic. Leipzig: WilhelmEngelmann, 1905.

Wundt, W. Outlines of psychology. Leipzig: WilhelmEnglemann, 1907.

Yntema, D. B., & Torgerson, W. S. Man-computer co-operation in decision requiring common sense. IRETransactions of the Professional Group on Human Fac-tors in Electronics 1961, Vol. HRE-2, No. 1, 20-26.

Zajonc, R. B. Cognitive structure and cognitive tuning.Unpublished doctoral dissertation, University of Michi-gan, 1955.

Zajonc, R. B. Attitudinal effects of mere exposure. Jour-nal of Personality and Social Psychology Monograph,1968, 9 ( 2 , Part 2, 1-28).

Zajonc, R. B., & Smoke, W. H. Redundancy in task as-signment and group performance. Psychometrika, 1959,24, 361-369.

AMERICAN PSYCHOLOGIST • FEBRUARY 1980 • 175