Additivity of Non Conscious Affect

8/8/2019 Additivity of Non Conscious Affect

1/14

ATTITUDES AND SOCIAL COGNITION

Additivity of Nonconscious Affect: Com bined Effects ofPriming and E xposureSheila T. MurphyUniversity of Southern California Jennifer L. MonahanUniversity of Georgia

R. B. ZajoncStanford UniversityAffect deriving from 2 independent sourcesrepeated exposure and affective primingwas in-duced, and the com bined effects were exam ined. In each of 4 studies, participan ts werefirstshown72 Chinese ideographs in which the frequency of exposure was varied (0, 1, or 3). In the secondphase partic ipants rated ideographs th at were prime d either positively, negatively, or not at all. Th e4 studies were identical except that the exposure durationsuboptimal (4 m s) or optimal (1 s)ofboth the initial exposure phase and the subsequent priming phase was orthogonally varied. Additiv-ity of affect wasobtained only when affective priming was suboptimal, suggesting that nonconsciousaffect is diffuse. Affect whose sou rce was apparent was more constrained. Interestingly, increases inliking generated throug h repe ated exposures did no t differ as a function of exposure duration.

Systematic data on nonconscious' affective processes are scarceand scattered. Although the concept figures prominently in thepsychoanalytic literature (e.g., Fenichel, 1945, pp . 161-167,238-240), its role as an explanatory construct in general experimentalpsychology has suffered critique, rejection, and disdain. Yet thereare a number of experimental results that are difficult to interpretwithout postulating some form of nonconscious affect. These in-clude the phenomena of perceptual defense (Bootzin & Natsoulas,1965; Bruner & Postman, 1948), perceptual vigilance (Blum,1954; Pratto & John, 1991), subliminal perception (Shevrin,1990), mood effects on memory (Bower, 1981), autonomic dis-crimination without awareness (Corteen & Wood, 1972; Lazarus& McCleary, 1951; McGinnies, 1949), growth of positive affectproduced by subliminal repeated exposure (Bornstein & D'Agos-tino, 1992; Kunst-Wilson & Zajonc, 1980), unconscious affectivepriming (Kitayama, 1991; Murphy & Zajonc, 1987, 1988, 1993;Niedenthal, 1987,1990), and others. Although many of hese phe-nomena were dismissed as artifacts (D ixon, 1981; Erdelyi, 1974;Eriksen, 1963; Goldiamond, 1958; Zajonc, 1962), for most, anexplanation more parsimonious than the activation of noncon-scious affect has not been successfully defended.

Intuition concurs. We have daily experiences in which we findourselves cheerful or depressed without having the slightest clue

This research was supported by National Science Foundation GrantBNS-8919734. We are grateful to Hazel Markus and Robert Bornsteinfor their helpful co mm ents.Correspondence concerning this article should be addressed to SheilaT. Murphy, Annenberg School of Com munic ation, University of South-ern California, 3502 South H oover St., Los Angeles, California 90089-0281.

as to the origins of these states. "Free-floating anxiety," surely aform of nonconscious affect, afflicts millions (Beck, 1976), yetits workings are far from explicated. There are also reports of"free-floating joy," a much less worrisome and therefore evenless researched phenomenon (Isen, 1990).Affect is nonconscious when the person is not aware of itssource, of its target, or both. In phobias, for example, the patientis aware of the target (e.g., spiders) and of the feeling evoked,although the fear's origins, embedded in the pa tient's reinforce-ment history, may not be accessible to awareness. In cases offree-floating anxiety, the origin and target of the anxiety areboth un recognized by the patient (Beck, 1976 ), but the feelingstate itself figures prominently in consciousness. It is possiblefor affect to gain influence over behavior or mental processeswith even less access to awareness. Without conscious access tosource, to target, or to the feeling state itself, affective influencesmay be revealed only indirectly (Reingold & Merikle, 1988,

1 Some authors use the term unconscious interchangeably with non-conscious. Because the former term may imply repression, we prefer th elatter. Moreover, Freud (19 15) often co ntradicted himself when speak-ing of the role of consciousness in em otional and affective reactions. Atone point, he stated the following:It is surely the essence of emotion that we should be aware of it, i.e.,that it should become known to consciousness. Thu s the possibilityof the attribute of unconsciousness would be completely excludedas far as emotions, feelings, and affects are concerned, (p . 177)

Yet in the same context he remarks, "But in psycho-analytic practicewe are accustomed to speak of unconscious love, hate, anger, etc., andfind it impossible to avoid the strange conjunction, 'unconscious con-sciousness of guilt,' or a paradoxical 'unconsciou s anxiety " ' (p p. 177178).Journal of Personality and Social Psychology, 1995, Vol. 69, No. 4, 5 89- 602Copyright 1995 by the American Psychological Association, Inc. 0022-3514/95/S3.00

589


2/14

590 S. MURPHY, J. MONAHAN, AND R. ZAJONC1991; Schacter, 1987), such as in word associations or slips ofthe tongue. Qualitative differences among these types of non-conscious affect are of interest in and of themselves. However,as a point of departure into the investigation of nonconsciousaffect, the current work focuses on affect of which the individua lis not aware of the source, that is, its eliciting stimuluscondition.Part of the confusion surrounding the study of nonconsciousaffect is that, with few exceptions (e.g., Kihlstrom, 1987 ), the-oretical reviews of the literature have sought to absorb affectinto cognitive theories and th us trea t nonconscious affect andnonconscious cognition as representing the same process. Howmight affective and cognitive processes differ? The roots of thedistinction between cognition and emotion reach to G reek epis-temology of the 5th century BC. Protarchus observed that wecan readily say of cognitions (i.e., knowledge) that they are ei-ther true or false. But the same ca nnot be said of emo tions. Onecan certainly conceal or distort the expression of an emotion,such as anger, and thus falsify its outward manifestation. But wecannot say of the experience of our anger that it is either true orfalse.Moreover, cognitions are dedicated. They have a specific ad-dress, target, or referent. They are always about something. Thiscontent specificity may apply to both conscious and noncon-scious cognitions.2 In lexical priming, for example (Marcel,1983a, 1983b; Meyer & Schwaneveldt, 1971), suboptimalprimes influence only judgm ents of words that a re semanticallyrelated. The prime NURSE reduces the response time of thelexical decision for the word D OCTOR , b ut it has n o influenceover the response time to the word M UFFL ER. Such semanticpriming effects are explained by appealing to semantic or cog-nitive network models that focus on spreading activation fromthe node representing the prime to the node representing thetarget (den Heyer, Briand, & Smith, 1985; Neely, 1977). If thetarget is semantically related to the p rime , search is facilitatedbecause the target node has been activated with the prime aspart of the entire network. In short, cognition, even when non -conscious, appears to be dedicated and content specific, influ-encing only related concepts.

In contrast, nonconscious affect need not be dedicated; itneed not be about a particular target. In the extreme case, non-conscious affect is more like liquid. It can d isperse, scatter, per-meate, com bine, fuse, blend, spill over, and become attached tototally unrelated stimuli.3 In a recent series of studies, Murphyand Zajonc (1987, 1988, 1993) presented Chinese ideographsthat were preceded by 4-msec suboptimal4 exposures to a vari-ety of primes. In certain conditions the suboptimal prim es con-sisted of faces expressing positive or negative emotions. Thesenonconscious affective primes were capable of inducing affectthat became displaced onto the Chinese ideographs. When pre-ceded by a positive facial expression, the ideographs werejudged more positively than when preceded by a negative facialexpression, even though neither expression nor even the pres-ence of an image of a face was accessible to particip ants' aware-ness.5 In contrast to the affective facial primes, suboptimalpriming using nonaffective primes, such as geometric shapes,failed to pro duce related shifts in pa rticipan ts' judgments of theideographs. For example, symm etric primes presented subopti-malJy did not induce participants to view subsequent targetideographs as more symmetric. Thus, the content of cognitive

prim ing at suboptimal viewing conditions did n ot diffuse freelyonto u nrelated stimuli. Affect and cognition, therefore, appearto have different time trajectories and distinct content con-straints (Murphy & Zajonc, 1987, 1988, 1993 Figures 2 and 3;Zajonc, 1980), suggesting that they are separable processes.The explanation of these effects appealed to the affective pri-macy hypothesis, which holds that affective reactions can beevoked with minimal sensory input and virtually no cognitiveprocessing and that they can be evoked earlier than cognitiveresponses. It was presumed th at because affect is primary andbecause of its diffuse quality, nonconscious affect deriving fromthe facial primes could "spill over" onto the target Chineseideographs.It is interesting to note tha t the identical affective primes pre-sented for 1,000 msec, and therefore available to consciousawareness, did not significantly alter ratings of the ideographs.These findings suggest tha t while nonconscious affect may bediffuse in n ature and capable of attaching itself even to un re-lated targets, conscious affect, because it contains significantcognitive components, may be more constrained.In addition to the tendency to subsume affect into cognitiveprocesses, previous research has also failed to observe the dis-tinction between reactions that are based on nonconsciousaffect and those based on more conscious affective evaluations.Nonconscious affect is diffuse, and its target is at best ambigu-ous. As cognitive processes enter into the affective reaction,elaboration or appraisal of these reactions accumulates(Ellsworth & Smith, 1988), and affect gains in specificity. Affectderiving from cognitive appraisal (Ellsworth & Smith, 1988;Lazarus, 1982) has a specific source and target. It is for thisreason that when the faces expressing affect in the experimentsby Murphy and Z ajonc (199 3) were presented for a full second,cognitive factors contribu ted to and d iluted th e overall affectivereaction. Because of increased cognitive elaboration and the re-

2 In the case of nonconscious cognitive or perceptual influences (e.g.,Bargn, Bond, Lombardi, & Tota, 1986; Bargh & Pietromonaco, 1982;Greenwald, Klinger, & Lui, 1989; Higgins, Rholes, & Jones, 1977; Lew-icki, 1986; Marcel, 1983a, 1983b; Merikle, 1982; Meyer & Schwane-veldt, 1971; Srull & Wyer, 1980; Uleman & Bargh, 1989), the effectsappear to be con tent specific.3 These metaphors may prove particularly apt in light of recent stud-ies linking mood and other massive states such as sleep to nonsynapticneurotransmission (B ach-y-Rita, 1993). In the course of this process,neurotransmitters and other neuroactive substances such as dopamine,norep inephrin e, serotonin, and acetylcholine are diffused through theextracellular fluid and reach extrasynaptic receptors (DiChiara, 1990;Ungerstadt, 1984). Extrasynaptic transmission has been demonstratednot only for massive states but for acute events such as responses tosensory stimuli. For example, nitric oxide, commonly referred to aslaughing gas, is diffused in this man ner and ha s a remarkably fast onset(a m atter of a few seconds) and decay (Garthwaite, 199 !). Such extra-synaptic transmission also takes place in the a utonomic nervous systemand, consequently, has direct implications for affective excitation(Stjarne, 1986, 1989).4 The term suboptimal was used by Murphy and Zajonc (1993) toindicate tha t stimulus presentation did not afford access to the stimulussufficient for recognition. The term subliminal was not used in the ear-lier and the present research because individual limens were not as-sessed for each p articipa nt.5 It is important to note in this context that the prime and the targetwere semantically and categorically unrela ted.


3/14

CONSCIOUS AND NONCONSCIOUS AFFECT 591sultant specificity or dedication, appraised affect may be moreresistant to being displaced or com bined. For instance, the factthat one is disappointed over a rejected article does no t readilymerge with on e's pleasure over an election outcom e. If one ofthese sources of affect is not accessible to awareness, however,fusion of the two affective reactio ns is more likely. Schwartz andClore( 1983), for example, found that when survey respondentswere asked about their well-being, they gave more positive an-swers on sunny days. When participan ts were made aware of theweather, the effect disappeared. Thus, conscious affect, to theextent it involves cognitive appraisal, may be content specific.As Merikle (1992) noted, the distinction between consciousand nonconscious perceptual processes is much more interest-ing and significant if access to awareness leads to qualitativelydifferent consequences than if nonconscious processes aremerely weaker versions of conscious processes. To test the hy-pothesis that nonconscious affect is diffuse whereas affect whosesource we are consciously aware of is more constrained, affectfrom two independent and unrelated sources could be induced,and the m anner in w hich the two effects com bine could be ex-amined. To do so, the present research relied on two paradigms:the affective priming paradigm discussed earlier and the growthof positive affect produced by repeated exposure (Kunst-Wil-son & Zajonc, 1980).

Research has consistently shown that when frequency ofexposure to a particular stimulus is increased, the stimulus isbetter liked. This "m ere e xposure " effect was initially found forstimuli fully accessible to aw areness and has been demonstratedin a variety of contexts using a wide assortment of stimuli, pop-ulations, and procedures (Bornstein, 1989; Harrison, 1977).6Early theorists explained this growth in affect with repeatedexposures as being based on feelings of subjective familiarity orrecognition. In this theoretical framework, an individual con-fronted with a familiar object experiences a "glow of warm th, asense of ownership, a feeling of intim acy" (Tichener, 1906, p.411). More recent theorizing, however, challenges this notionthat subjective familiarity underlies the mere exposure phe-nomenon . Evidence has accumulated showing that affect whichthe valence derives from repeated exposures can be obtainedwith stimulus exposures that are not accessible to the partici-pan ts' awareness (Barch as & Perlaki, 1986; Bonnano & Stilling,1986; Bornstein, Leone, & Galley, 1987; Kunst-Wilson & Z a-jonc, 1980; Mandler, Nakam ura, & Van Zandt, 1987; Seamon,Brody,&Kauff, 1983a, 1983b ). In these studies the initial stim-ulus exposures were so degraded th at subsequent tests revealedno more than chance discrimination between old and new stim-uli. Yet despite this lack of conscious recognition, the stimuliincreased in attractiveness as a function of the actual numberof degraded exposures. In fact, the growth in preference withrepeated exposures is as true for degraded stimuli (Kun st-Wil-son & Zajonc, 1980; Seamon, B rody, & Kauff, 1983a, 1983b)as it is for optimally presented stimuli (M atlin, 1971; Moreland& Zajonc, 1977, 1979), and in one comparison, the exposureeffect was found to be stronger for the degraded condition(Bornstein, 1987). Thus, several studies have found that sub-jective familiarity is no t a necessary precondition for the growthof affect associated with exposure (Kunst-Wilson & Zajonc,1980; Matlin, 1971; Wilson, 1979). Instead, it is the person'sobjective history of encounters that influences the gain in posi-tive affect toward th e object exposed.

Why is it the case that the mere exposu re effect has been dem-onstrated at both suboptimal and optimal durations, whereasaffective priming is effective suboptimally and relatively in-effective at optimal durations? Recall that in the aforemen-tioned priming study (M urphy & Zajonc, 1993), affect did notinfluence evaluations when participants became aware of asource of affect other than the target ideographs, namely theoptimal facial prim es. In the mere exposure paradigm , however,the source of the positive affect generated seems to be equallyobscure at both suboptimal and optimal exposure durations. Instudies involving m ere exposure, individuals are unaw are of therelationship between frequency of exposure and affect and vir-tually never refer to their repeated experience with the stimuliin explaining their preferences. In much the same way that in-creasing the length of time of exposure to a word in a foreignlanguage does not clarify its meaning, increasing the length oftime people are exposed to a novel stimuli does not lead them toinfer the relationship between repeated exposures and positiveaffect. Hence, in the mere exposure paradigm, the source ofthe affect typically remains unavailable regardless of exposureduration.

Combining priming and exposure within the same experi-mental paradigm and com paring their effects un der optimal andsuboptimal stimulus durations may shed some light on the in-terplay of affect, cognition , and awareness. The presen t researchattempted to do just that by crossing stimulus accessibilityoptimal and suboptimalw ith these two pa radigms of affectiveinfluencerepeated exposure and affective priming in a se-ries of four studies. Each study included an initial exposurephase in which m ere exposure was induced by varying the fre-quency of exposure to novel stimuli, Chinese ideographs. In asecond phase, participants rated ideographs that were precededby positive, negative, or no affective prim es. Extrapolating fromthe previous experimental literature, a number of predictionscan be made. Because exposure effects work under both optimaland suboptimal conditions, there should be a main effect forexposure in all four studies. According to the study by Murphyand Zajonc (1993), however, affective priming should beeffective only under suboptimal conditions.

Our prima ry hypothesis, however, concerns the additivity ofnonconscious affect. If nonconscious affect were diffuse, wewould expect suboptimal affective priming to com bine with th epositive affect generated from repeated exposures. Moreover,there is no reason to believe that the contribution from an initial6 Exposure effects have been found for ideographs (Mo reland & Za-

jonc, 1977; Zajonc, 1968), letters (Nuttin, 1987), nonsense words(Matlin, 1971), odors (Balogh & Porter, 1986; Davis & Porter, 1991),flavor (Newlin & Pretorius, 1991), colors (Franchina, 1991), food(Rogers & Hill, 1989), geometricfigures Fink, Monahan, & Kaplow-itz, 1989; Seamon, Marsh, & Brody, 1984), photographs of faces(Morinaga & M atsumura, 1987), actual persons (Moreland & Beach,1992) and many other stimuli. Populations found to be responsive toexposure effects include undergraduates (Bornstein, 1989), nationalsof 12 different coun tries (N uttin , 1987), sons of alcoholics (Newlin &Pretorius, 1991), amnesiacs (Williams, 1990), dieters (Rogers & Hill,1989), human neonates (Davis & Porter, 1991), chick embryos (Grier,Counter, & Shearer, 1967; Rajecki, 1972, 1973), chicks (Franchin a,1991), ducklings (Lickliter & G ottlieb, 1986), as well as several otherspecies. Readers might refer to Bornstein (1989) or Harrison (1977)for more extensive reviews of the mere exposure litera ture.


4/14


5/14

CONSCIOUS AND NONCONSCIOUS AFFECT 593making some snap judgments of a series of Chinese ideographs. First,however, to provide a basis for comparison, they would be shown across-section of ideographs similar to tho se they would later be asked tojudge.Phase 1: Repeated exposures. During the initial exposure phase,participants viewed 72 ideographs. In Studies 1 and 3, each ideographwas shown for 4 msec, imm ediately followed by a backward m ask con-sisting of a slide with white, grey, and black d ots. In Studies 2 and 4, thesame ideographs were shown for 1,000 msec each. In all studies therewas a 1,500-msec interval between successive exposures.Of the 72 exposures, 24 were filler ideographs, each of which wasshown only once and was not included in th e second, affective primin gphase of the experiment. The sole purpose of thefiller deographs in theinitial exposure phase was to increase th e overall pool of ideographs sothat participants would be less likely to notice repetitions. The remain-ing 48 exposures served to indu ce the m ere exposure effect by varyingfrequency of presentation . These 48 trials consisted of 24 different ideo-graphs divided into two groups (A and B) of 12 ideographs each.As frequency of exposure was manipu lated in Phase 1, it was coun-terbalanced. Specifically, to ensure th at there was no systematic differ-ence between ideographs selected to be shown only once as opposedto those shown three times, in the initial exposure phase frequency ofexposure to sets A and B was counterbalanced. Thus, half of the p artic-ipants within each study were presented with set A ideographs threetimes in the initial exposure phase (36 presentations) and set B ideo-graphs only once (12 presen tations). The other half of the participantswere presented with each of the set B ideographs three times and eachof the set A ideographs only once.Phase 2: Affective priming. Following the exposure phase, partici-pants rated a series of 48 ideographs. O f the 48 ideographs rated, 12 hadbeen shown three times during the initial exposure period, 12 had beenshown once, and 24 had not been presented previously. Thus, partici-pants rated 24 ideographs they had seen previously (the ideographs in-cluded in sets A and B) as well as 24 ideographs they had not encoun-tered previously (12 ideographs of the zero-exposure set C and 12 fillerideographs). "New" ideographs shown only in the rating phase re-mained the same for all participants. Precautions were taken to ensurethat new ideographs did not differ in term s of likability or complexityfrom those in sets A or B. Indeed, care was taken to exclude ideographswith extreme liking ratings and to include 6 relatively simple and 6 rel-atively complex ideographs at each level of exposure (zero, one, andthree).8Within each exposure condition (zero, one, or three exposures), fourideographs were primed with positive affective primes, four wereprimed with negative primes, and four had no primes. As affectivepriming was manipulated in Phase 2, valence of affective prime wascounterbalanced. Specifically, to ensure that there was no systematicassociation between particular ideographs and particular primes, thevalence of the prime associated w ith any given ideograph was counte r-balanced during this second rating phase. In other words, any givenideograph in sets A, B, and C was preceded by a negative prime for onethird of the participants, a positive prime for another third, and noprime for the final third. All participants rated the ideographs in thesame order. The firstand last three ideographs in the sequence were filler ideographs shown without primes.In Studies 1 and 2 , the affective prim es were presented suboptimallyusing a backward pattern-masking technique, in which the prime (aface) was presented for 4 msec , followed im mediately by the presenta-tion of a target stimulus (an ideograph) for 1,000 msec. Participantswere not informed that the ideographs were preceded by suboptimalprimes. The rating series began with three practice trials. By contrast,in Studies 3 and 4 the same apparatus and procedure were used exceptthe facial prim es were presented optimally for 1,000 msec each. To pro-vide a rationale for viewing the facial primes during the rating phase,participants were given additional instructions prior to rating theideographs:

While you are rating the set of Chinese Ideographs, you may noticea series of pictures of people expressing different forms of emotionon the screen. You do not need t o rate these pictures as subjects inother conditions w ill do so. Remember, your job is to rate only theChinese Ideographs.

Dependent MeasuresIn Phase 2 of the study, the target ideographs were each presentedonly once. Participants made two judgments about each ideograph: lik-ing and recognition. Participan ts werefirstasked to rate how much theyliked a particular ideograph on a scale from 1 to 5 where 1 meant thatthey did not like the ideograph at all and 5 meant they liked the ideo-graph quite a bit. Next, as an indicator of recognition, participa nts wereasked whether they thought the particular ideograph had been shownduring thefirstseries. They were instructed to press a key marked OLDif they thought they had seen the ideograph in the first phase and topress a key marked NEW if they felt the ideograph had not been shownpreviously.9

Manipulation Check: Forced-Ch oice Test of AwarenessA number of criticisms have been leveled at experiments reportingpriming effects obtained with stimuli presented under degraded condi-tions. For the most part, these critics justifiably question w hether thereis a total absence of conscious detection or identification (Holender,1986). To ensure that the degraded 4-msec presentation of the affectiveprimes was, in fact, below conscious awareness, participa nts were givena forced-choice test of awareness following the 48 trials, as suggestedby Eriksen (1 980 ). At this point, participants were informed that theywould be given a series of trials in which faces were presented briefly,immediately followed by an ideograph. A s in the experime nt itself, thecomputer em itted a "beep," signaling a 4-msec presentation of a prime(a face) that was immediately followed by a 1,000-msec presentation ofa backward mask (an ideograph). Participants were then presentedwith two test faces: an image of the actual prime on one side of the

screen and of an alternate face, or foil, on the other side of the screen.The faces were of the same gender and expressed the same emotion.10Participants were then asked which of the two faces they thought wasthe prim e. Th e rationale underlying this forced-choice test is that if par-ticipants truly cannot detect the prime, they should do no better thanchance at recognizing it. Each participan t took pa rt in 12 such forced-

8 A pretest was conducted in which an additional 30 participantsrated a preliminary pool of ideographs with respect to liking andcomplexity.9 The decision to use a dichotomous recognition judgment (OLD-NEW ) was guided by an interest in replicating the paradigm used inprevious studies (i.e., Moreland & Zajonc, 1979 ). It should be noted,however, that a 5-pointscale similar to tha t used in the liking judgmentwould have permitted a greater degree of sensitivity and c omparability(Reingold & Merikle, 1988).10 It is important to note that the emotional expression of the primeand foil was kept constant. In a study by M urphy (199 0), participantsin a similar forced-choice test of awareness were unable to discriminatebetween two faces (the actual suboptimal prime and a foil) when thefaces expressed the same emotion (i.e., both smiling) but were able toidentify the prim e accurately when the two faces expressed hedonicallyopposing emotions (one smiling and one angry). In short, it appearsthat participants are able to process nonconscious affective cues thatallow them to perform at a level greater than chance on this task. Be-cause the purpose of this manipulation check was to test for subjectiveawareness of the suboptimal p rimes, however, the hedonic valence of theprim e and foil was held constant.


6/14

594 S. MURPHY, J. MONAHAN, AND R. ZAJONCchoice trials. Participants were subsequently debriefed and thanke d fortheir participation.

ResultsManipulation Check: Forced-Choice Test of Awareness

On the forced-choice test of awareness, participants scoredan average of 5.80 correct out of a possible 12, which is notsignificantly different from a chance score of 6. This result de m-onstrates that a 4-msec exposures, participants were unable todistinguish between a facial prime presented 1 s earlier and anovel face.Analyses

These analyses examine the effects of repeated stimulusexposure and affective priming on the liking rating of the targetChinese ideographs. The 36 liking means from all four da ta sets(N = 140) were entered into a 2 (Exposure Duration: optimalvs. suboptimal) X 2 (Priming Duration: optimal vs.suboptimal) X 3 (Num ber of Exposures: zero, one, three) X 3(Valence of Prime: negative, no prim e, positive) mixed analysisof variance (ANOVA) with the first two factors between-sub-jects and the last two factors within-subjects.11 Duration of therepeated exposures and primes are between-subjects factors,whereas the frequency of exposures and type of prime arewithin-subject factors. There was no significant effect caused bywhich ideograph set (A or B) was shown three tim es as opposedto only once in the initial exposure ph ase; therefore, this coun-terbalancing was ignored in the subsequent analysis. Similarly,there appeared to be no relationship between the pairing of anyparticular ideograph with a particular valence of prime; thus,the counterbalancing of valence of prime in Phase 2 was like-wise excluded from the following analysis.Duration of repeated exposure.12 Across the four studies,the main effect for stimulus duration during the initialexposures phase was significant, (M =3 .11 optimal, M = 2.95suboptimal), F( 1,136) = 4.45, p = .037. A closer exam inationof the means, however, indicates that the effect for duration oc-curred only in the one-exposure condition. In the one-exposurecondition, ideographs shown optimally were liked marginallybetter than ideographs shown suboptimally (M s of 3.06 and2.89, respectively), t( 138) = 1.90,p < .06. There were no othersignificant differences within stimulus duration.Duration of prime. There was no between-subjects effect forduration of prime, F(l, 136) = .34, p = .56. As subsequentanalyses demonstrated, however, the duration of prime in-

teracted with valence of prime. No other effects were reliable.Number of exposures. Across all four studies, the within-subject factor of frequency of exposure (zero, one, three) onliking was significant (zero exposure M 2.78, one exposureM = 2.97, three exposures M = 3.34), F(2, 272) = 77.83, p


7/14

CONSCIOUS AND NONCONSCIOUS AFFECT 595

1.0"0.8-0.6-0.40.20 . 0 H

-0.2-0.4--0.6--0.8--1.0

O Positive PrimeA No Prime Negative Prime

tb

INumber of E xposures

Figure 1. Study 1: Liking ratings as a function of suboptimal exposure and suboptimal affective prime .The liking ratings reported are subtracted from the average liking rating of 3. Items with shared subscriptsare no t significantly different at /? < .016.

The data revealed a main effect for frequency of exposure(zero exposure M = 2.72, one exposure M = 2.93, threeexposures M = 3.34, F{2,78) = 23.37, p < .001 ) and for affec-tive priming (negative prime M = 2.69, no prime M = 3.02,positive prime M = 3.27), F( 2,78) = 30.02, p < .001. Figure 1shows mean liking ratings of the ideographs as a function offrequency of exposure and affective priming. In Figure 1, theintercept differences among the three curves reflect the effectsof nonconscious priming , whereas the slopes of the curves rep-resent the mere exposure effect. As shown in Figure 1, positivepriming adds a constant to the affect generated by all threeexposure frequencies, whereas negative priming subtracts aconstant from the affective ratings in these three frequencies.Across exposure conditions, ideographs shown with positiveprimes prod uced significantly higher liking ratings than thoseshown with no prime, which, in turn, produced significantlyhigher liking ratings than ideographs shown with negativeprimes. The interaction between priming and exposure was no tsignificant, F(4, 156) = 1.70, ns. These results indicate, there-fore, that when two sources of affect are nonconscious, affectcombines additively, the three exposure slopes being parallel.

Optimal exposure-suboptimal primes. Figure 2 showsmean liking ratings of the ideographs as a function of repeatedoptimal exposure and suboptimal affective priming . These datareveal main effects for both repeated exposure (zero exposureM = 2.72, one exposure M = 3.00, three exposuresM = 3.38),F(2, 78) = 35.38, p < .001, and for affective priming (negativeprime M = 2.61, no prime M = 3.07, positive prim e M = 3.42),F( 2,78) = 5 5.16, p < .001. In Figure 2, the intercept differencesamong the three curves reflect the effects of nonconscious prim-

ing, whereas the slopes of the curves represent the mereexposure effect.As is evident in Figure 2 , the mean s increased as a functionof frequency of exposure. In addition, across frequencies, ideo-graphs shown with positive suboptimal primes produced sig-nificantly higher liking ratings than those shown w ith no p rim e,which, in turn , produce d significantly higher liking ratings thanideographs shown with negative prim es. T he Exp osure X Prim-ing interaction was significant, F(4, 156) = 4.83, p < . 0 1 . nThe source of this interaction is to be found among the prim ingmeans in the zero-exposure condition. Among ideographs notincluded in th e initial exposure p hase, the m ean difference (M= .06) between those shown with no prime and those shownwith a happy prime was not significant. However, at oneexposure, there was a significant difference among all threepriming conditions, as depicted in Figure 2. Thus, for the one-and three-exposure conditions, the three priming curves areparallel. Furtherm ore, when th e analysis of variance was recal-culated for the one- and three-exposure conditions, eliminatingthe zero-exposure condition from the analysis, the main effectsfor exposure, F( 1,39) = 23.88, p< .001, and priming, F( 2,78)= 29.28, p < .001, remained significant, whereas the Priming XExposure interaction, F(2, 78) = 2.07, p - .13, was no longersignificant.

13 The means are as follows: three exposures (positive primes M =3.96, no prime M = 3.34, negative primes M = 2.85); one exposure(positive primes M - 3.36, no prime M = 2.99, negative primes M =2.63); zero exposure (positive primes M = 2.94, no prime M = 2.88,negative prim es Af = 2.34).


8/14

596 S. MURPHY, J. MONAHAN, AND R. ZAJONCO Positive PrimeA No Prime Negative Prime

0 t 3Number of Exposures

Figure 2. Study 2: Liking ratings as a function of optimal expos ure and suboptimal affective prim e. Theliking ratings reported a re subtracted from the average liking rating of 3. Items with shared subscripts arenot significantly different at p < .016.

The results from Studies 1 and 2 support the hypothesis thatnonconscious affective primes combine additively with re-peated exposure effects, whether the exposure effect is subopti-mal (Study 1) or optimal (Study 2 ). Studies 3 and 4 examinedthe effects of optimal affective pr imes.Suboptimal exposure-optimal primes. In Study 3, theexposure effect was suboptimal, whereas the affective primeswere shown at optimal viewing levels. Although a main effectfor exposure was obtained (zero exposure M = 2.71, oneexposure M = 2.84, three exposures M = 3.21), F(2, 58) =16.03, p < .001, there was no main effect for affective priming,F( 2,58) = .66, ns, no r was there a significant interaction effect,F( 4,116) = 1.64, n s. Even though the Chinese ideographs werepresented at suboptimal viewing conditions, liking once againincreased a s a function of frequency, with a dditional exposuresresulting in higher liking ratings, as evident in Figure 3.

Optimal exposure-optimal primes. Study 4 examinedmean liking ratings of the ideographs as a function of repeatedoptimal exposure and suboptimal affective priming. The datarevealed a main effect for frequency of exposure (ze ro exposureM= 3.01, one exposure M = 3.14, three exposures M = 3.42),F( 2, 58) = 11.08, p < .001, but the effect for affective primingwas not significant, F(2, 58) = 1.00, ns , no r was the ExposureX Priming interaction significant, F( 4,116) = 1.41, ns . In Fig-ure 4, the intercept differences am ong the th ree curves reflectthe effects ofconscious priming, whereas the slopes of the curvesrepresent the exposure effect. Although the m eans increase as afunction of exposure, as shown in Figure 4, the means did n otvary as a function of optimal affective prime.The results from Studies 3 and 4 indicate that when the affec-

tive priming stimuli were presented at optimal dura tions, therewas no diffuse affect to combine with the affect generated bymere exposure in either an additive or interactive fashion.Recognition Memory

Suboptimal stimulus durations.14 Because the Chineseideographs were presented at suboptimal viewing levels in Stud-ies 1 and 3, the data for Studies 1 and 3 (suboptimal durations)were collapsed to report one overall analysis for recognitionmem ory un der suboptimal viewing conditions. Table 1 presentsthe liking ratings of ideographs at each exposure frequency(zero, one, three) as a function of participants' perceived andactual recognition. Results of the t tests presented in Table 1suggest that the exposure effect did not depend on whether theparticipan ts themselves judged the stimulus to be old or new.15Collapsing across priming conditions, participants tended to14 The da ta for Studies 1 and 3 (suboptim al exposures) were collapsedto allow reporting of one overall analysis for recognition memory u ndersuboptimal viewing conditions because there were no significant differ-ences between the twostudies for the exposure analyses. A similar anal-ysis is reported for Studies 2 and 4 (optimal exposure). The originalanalyses, broken out by all four studies, are available from Sheila T.Murphy.15 It should be noted th at in an attempt to avoid a Type I error, weused a conservative (p < .016) significance level. This level was basedon the following logic: Because there are three t tests possible withineach level of exposure or within each level of priming, we set the alphaat .05/3 = .016.


9/14

CONSCIOUS AND NONCONSCIOUS AFFECT 597

1.010.80.60.40.20.0

-0.2-0.4-0.6--0.8

-1.0 T-0

b

O Positive PrimeA No PrimeD Neg uive Prime

be

Number of E xposuresFigure 3. Study 3: Liking ratings as a function of suboptimal exposure and optimal affective prim e. Theliking ratings reported are subtracted from the average liking rating of 3. Items with shared subscripts arenot significantly different at p < .016.

1.00.80.60.4

0.03

-0.4-0.6-0.8-1.0

O Positive PrimeA No PrimeD Negative Prime

Number of ExposuresFigure 4. Study 4: Liking ratings as a function of optimal exposure and optimal affective prim e. T he likingratings reported are subtracted from the average liking rating of 3. Items with shared subscripts are notsignificantly different at p < .016.


10/14

598 S. MURPHY, J. MONAHAN, AND R. ZAJONCgive higher liking ratings to ideographs to which they actuallywere previously exposed, independently of their perception offamiliarity. In add ition, when the ideographs were shown sub-optimally in the initial exposure ph ase, particip ants' liking rat-ings of the ideographs perceived as old were not significantlymore positive than their ratings of the ideographs perceived asnew.Regardless of the num ber of suboptimal presentations in theinitial exposure phase (zero, one, or three) the accuracy ratenever exceeded chance (48%, 40%, and 44%, respectively). Th eoverall accuracy rate was 44%. The da ta show, therefore, thatliking ratings are under the influence of the person's objectivehistory of experience with these stimuli and not the person'simpression of familiarity. This was so even if the participantswere not aware of that history. Subjective judgmen ts of famil-iarity contributed virtually nothing to liking responses. Stimulithought to be old were rated -.04 on the liking scale, whereasthose thought to be new were also near the neutral po int ( .02 ).In contrast, zero, one, and three exposures generated a clearincrease in liking scores: -. 2 3 , -.0 6 , and +.2 7, respectively.

Optimal stimulus durations. In Studies 2 and 4, the Chineseideographs were presented at optimal viewing levels. Collapsingacross priming conditions, the / tests presented in Table 2 sug-gest that pa rticipants tended to give higher liking ratings to ideo-graphs to which they had been previously exposed and thatthese judgments did not depend on whether the participantsthemselves judged the stimuli to be old or new. In other words,participants were influenced by actual frequency of exposure(zero exposure M = -.15, one exposure M = .09, threeexposures M = .39), whereas their subjective impressions offamiliarity had little influence on their judgm ents.Within each exposure frequency, liking ratings of ideographsperceived as old were not significantly different from ratings of

those perceived as new. Collapsing across both priming andexposure conditions, liking ratings of ideographs perceived asold were somewhat higher than ratings of ideographs perceivedas new (.17 vs. .02, respectively). Participants who viewed theideographs in the initial phase under optimal viewing condi-

Table 1Liking Ratings ofIdeographs as a Function ofPerceivedand Actual Suboptimal ExposurePerceivedexposure Zero One Three

Actual exposureOldNew

- . 2 4 .(437)- . 2 2 .(403)

- . 1 0 .(335)- . 0 3 .(505)24b(373).29b(467)

Percentage of correct responses48 40 44

Note. The liking ratings reported are subtracted from the average lik-ing rating of 3. The num ber of udgm ents within each cell are presentedin parentheses. This table represents the data from Studies 1 and 3, inwhich ideographs in the initial exposure phase were presented subopti-ma//y for 4 ms. Items w ith shared subsc ripts are not significantly differ-ent atp


11/14

CONSCIOUS AND NONCONSCIOUS AFFECT 599however, there was an effect for the simp le ideographs such th atthe simpler ideographs were liked significantly more when pre -sented at optimal duration levels during the initial exposurephase (Ms of 3.15 optimal and 2.93 suboptimal, F[ 1, 138] =8.97, p = .003). Stimulus duratio n of the ideographs in the ini-tial exposure phase did not interact w ith duration of priming .

Stimulus complexity X number of exposures. Stimuluscomplexity interacted with frequency of exposure, F(4,544) =4.75, p = .001, such tha t for ideographs no t previously seen inthe initial exposure phase (zero exposures), complex ideo-graphs were liked significantly more than simple ideographs(Ms of 2.93 and 2.70, respectively), t( 139) = 3.80, p < .001.However, for ideographs that appeared three times in theexposure phase, complex ideographs were liked significantlyless than simple ideographs (Ms of 3.26 and 3.42 respectively),/(139) = 3.29, p = .001. There were no significant differencesfor ideographs shown only once in the exposure phase (Ms of2.96 complex and 2.99 simp le).

DiscussionThe present studies were carried out under the premise thataffect w hose source is unavailable to awareness may have prop -erties that distinguish it from affect whose source is known.More specifically, it was proposed that nonconscious affect maybe diffuse and that affect generated from one source (repeatedexposure) might easily com bine w ith nonconscious affect elic-ited from a second source (priming). Thus, it was predictedthat affect derived from suboptimal priming, positive or nega-tive, would combine in a roughly additive fashion with mereexposure effects. This was indeed the case. Positive suboptimalpriming roughly added a constant, whereas negative suboptimalpriming subtracted a cons tant from th e positive affect generated

by mere exposure. This effect occurred regardless of whetherthe ideographs in the initial exposure phase were presented sub-optimally for 4 msec as in Study 1 or optimally for 1,000 msecas in Study 2. In sum, positive suboptimal primes augmentedwhereas negative suboptimal primes attenuated the exposureeffect.In contrast to nonconscious affect, we assumed that affectavailable to conscious awareness would recruit cognitive ap-praisal and consequently be less diffuse and less likely to com-bine with an unrelated source of affect in a simple additive fash-ion. This issue was addressed systematically in Studies 3 and 4.In Study 3, the ideographs in the initial exposure phase werepresented suboptimally, whereas the smiling and scowling facialprimes were presented optimally. Here, as in the work of M ur-phy and Zajonc (1993), the optimally presented affectiveprimes did not significantly sway judgm ents of the target ideo-graphs. It is our contention that as the duration of the primesincreased to 1 s, participants could readily identify the sourceof affect, namely the smiling and angry faces, and the associatedaffect became dedicated, thereby losing its capacity to merge.Note th at this con striction relies on awareness of the source ofaffect. To the extent that individuals were aware and perhapssuspicious of the facial prime s presented for 1 s immed iatelyprior to the ideographs being rated, the prim es should have littleor no effect. Conversely, to the extent that individuals were no taware of the positive affect generated by optimally repeatedexposures, the affect should continue undisrupted. These

hypotheses were supported in Study 4, in which bo th the ideo-graphs in the initial exposure phase and the facial primes pre-sented in the subsequent judgm ent phase were shown optimallyfor 1,000 ms.Combined Effects of Affective Priming and RepeatedExposureWhen all four studies are entered into a single analysis, a clearpattern emerges. Whereas the growth in preference with re-peated exposures proceeded regardless of stimulus duration,affective prim ing was only effective suboptimally. These diver-gent results from these two sources of affectpriming and re-peated exposurehighlight the inadequacy of relying onexposure duration alone as an index of awareness. In the presentresearch, no pa rticipant indicated being aware of any subopti-mally presented stim uli. This lack of awareness was further evi-denced by a failure on the p art of participan ts to exceed chanceon a forced-choice test of awareness. At suboptimal exposures,then, both the repeated exposures and the affective primes weresimilarly unavailable to conscious awareness. Under optimalexposure conditions, however, differences between the twosources of affect arose . M ore specifically, in the optimal prim ingconditions, participants indicated th at they were somewhat sus-picious of the smiling and angry faces presented immediatelyprior to the ideographs they were asked to judge. In conditionsinvolving optimal repeated exposures, participants were gener-ally aware of the repetitions as revealed in a n elevated, but lessthan perfect, recognition ra te. Yet despite the fact that recogni-tion accuracy was fairly high for ideographs seen for 1 s, partic-ipants rem ained unaware of the positive affect associated withthese repeated exposures.During debriefing, when participants were informed that

some ideographs were repeated and they were asked to specu-late as to possible effects of multiple rep etitions, the vast major-ity of participan ts predicted th at repeated exposures would leadto boredom and decreased liking. These naive hypothesesclearly indicated that participants were not "aware" of theaffective influence generated by the repeated exposures to thesame degree that they were aware of the potential affective in-fluence of the smiling and angry primes. T his lack of awarenessregarding the actual source of the affect, namely tha t repeatedexposures leads to increased liking, allowed the mere exposureeffect to con tinue undisrupted.The proposition that awareness of the source of affect mayconstrain its influence is particularly interesting in light of ear-lier claims that the mere exposure phenomenon can be ex-plained by the "warm glow of familiarity." B onnano and Stilling(1986), for instance, who showed that a m ore pointed recogni-tion question yielded the sam e results as liking, argued that fa-miliarity is a factor in the exposu re effect. In ascertaining famil-iarity, however, they asked their participants to "choose theshape that initially seems m ost familiar, or pops out at you" (p.407). It is possible that their particip ants, prom pted to choosethefigures hat "pop out at them," used liking as the basis formaking familiarity judgments. Like Bonnano and Stilling,Brooks and Watkins (198 9) also have asserted tha t recognitionmediates liking in the mere expo sure effect rather than the con-verse (see also Birnbaum & Mellers, 1979). Their argumenthinges on the fact that with optimal exposures, a greater "hit


12/14

600 S. MURPHY, J. MONAHAN, AND R. ZAJONCrate" was obtained for recognition than for liking. Indeed, whenexposure effects were induced optimally, as in Studies 2 and 4,perceived and objective history with the stimuli were signifi-cantly related. However, the fact that at optimal exposure levels,perceived and objective history were related does not in itselfinvalidate the proposition that liking need not depend on rec-ognition. Only if it is shown tha t pa rticipan ts' perception of fa-miliarity mediates liking, rather than their objective historywith the stimuli, would the familiarity argument hold. We donot find this relationship when the mere exp osure effect is in-duced suboptimally, as in Studies 1 and 3. Rather, when de-graded stimuli are exposed repeatedly, pa rticipan ts' liking rat-ings were more reflective of their objective history with the stim-uli than with their recognition memory. This finding isconsistent with Reingold and Merikle's (1991) criterion thatnonconscious perception is demo nstrated whenever an indirectmeasure (liking) shows greater sensitivity than a direct measure(recognition).

Properties of Nonconscious AffectAffect from two independent sourcespriming andexposurewas found to com bine in an additive fashion onlywhen the individual was unaware of the source of the affect.As the source of affect became known (e.g., optimally pre-sented facial primes), cognitive correlates became involvedand the affect became referenced. This qualitative differencehighlights one property of nonconscious affect, namely thatit is relatively diffuse and can become attached even to unre-lated stimuli.Moreover, because it is not subject to cognitive appraisal,nonconscious affect may yield stronger or less adulteratedeffects. Recall that it was argued that longer stimulus durations

may allow access to attributes of the facial primes not pre-viously available. For example, when the stimulus duration isprolonged to 1,000 msec, a smiling face may reveal features thatinduce negative affect such as thin lips or a bulbous nose. Thepositive affect produced by the sem blance of the smile may con-sequently be diminished or even reversed. Hence, at longerexposure durations further appraisals become possible, and theinitial affective reaction may be diluted.A post hoc analysis dividing the ideographs on the basis ofcomplexity supports this contention and provides further in-sight into the properties of nonconscious affect. Relatively sim-ple ideographs were liked significantly more at optimalexposure levels. This may be an indication that in more com-plex stimuli, a variety of features may trigger affect, some ofthem contradictory, whereas simple stimuli provide a morepristine canvas. This hypothesis is further supported by the factthat, in the present series of studies, simple ideographs pro-duced significantly higher liking ratings at thre e ex posures thandid complex ideographs at both optimal and suboptimalexposure levels. Together, these findings suggest that simplestimuli may hold a relative advantage in the production andtransmission of nonconscious affect.On a related note, Greenwald (199 2) argued that, at the non-conscious level, analytic capacity may be severely limited. Weconcur. Nonconscious affective processing is gross. Murphy(1990), for ex ample, was unable to find evidence of emotionalspecificity at the nonconscious level. In a forced-choice discrim-

ination paradigm, participants were exposed to 4-msec subop-timal prim es of faces expressing Ekm an's (19 72) six basic emo-tions. Participants were then shown two facesan image of theactual prime and an incorrect alternative face, or foilandasked to "guess" which of the two faces was the suboptimalprime. Participants made forced-choice discriminations be-tween all possible pairs of Ekm an's six basic emotions. Only thepositive emotion of happiness was differentiated at a levelgreater than chance from the negative emotions of anger, fear,sadness, and disgust. Participants were un able to differentiateany of these negative emotions from one another. These datawere interpreted as indicating that although primitive positiveor negative affective reactions may occur relatively early in theinformation-processing chain, even outside of conscious aware-ness, some sort of cognitive elaboration may be required to de-fine precisely which specific emo tion, such as anger or fear, willultimately emerge.

The theory offered here and by Murphy (19 90) and Murphyand Zajonc (199 3) resembles the two-factor m odel proposed 3decades ago by Schachter and Singer (196 2). B oth theories haveas their elementary concepts an affective process and a cognitiveprocess. But there is an important distinction. Schachter andSinger regarded em otion as requiring the presence of a no nspe-cific arousal that drives the person to seek an ex planation. Thisnonspecific arousal was hypothesized to recruit cognitive ap-praisal that could provide meaning to an experience. The pres-ent conceptualization, in agreement with Maslach (1979), castsdoubt on the notion of arousal that is devoid of valence. Rather,we assume that arousal mu st be either positive or negative.The present results, together with those of Murphy and Za-jonc (1987,1 988,19 93) and Murphy (1990), demonstrate thatnonconscious affect is not merely a paler version of consciousaffective processes, but rather ha s distinct dynam ic properties.

Affect derived from nonconscious sources seems to be morereadily combined and displaced than affect based on cognitiveappraisal. Moreover, nonconscious processing appears to belimited in its ability to process complex or emotionally specificinformation. Finally, in contrast to earlier theoretical modelsthat assumed undifferentiated affective arousal (e.g., Schachter& Singer, 1962), evidence suggests that even at the noncon-scious level, affect is hedonically valenced. Although these qual-itative differences are far from exhaustive, they represent a firststep toward understanding the workings of nonconscious affect.References

Bach-y-Rita, P. (19 93 ). Nonsynaptic diffusion neurotransmission(NDN ) in the brain. Neurochemistry International, 23 , 297-318.Balogh, R. D., & Porter, R. H. (1986). Olfactory preferences resultingfrom mere exposure in huma n neonates. Infant Behavior an d Devel-opment, 9, 395-401.Barchas, P. R., & Perlaki, K. M. (1986). Processing of preconsciouslyacquired information measured in hem ispheric asymmetry and se-lection accuracy. Behavioral Neuroscience, 100, 343-349.Bargh, J. A., Bond, R. A., Lombardi, W. J., & Tota, M. E. (1986). Theadditive nature of chronic and tem porary sources of construct acces-sibility. Journal ofPersonality and Social Psychology, 50, 869-878.Bargh, J. A., & Pietromonaco, P. (1982). Automatic information pro-cessing and social perception: The influence of trait inform ation p re-sented outside of conscious awareness on impression formation.Journal of Personality and Social Psychology, 43 , 437-449.


13/14

CONSCIOUS AND NONCONSCIOUS AFFECT 601Beck, A. T. (1976). Cognitive therapy and the emotional disorders. NewYork: International Universities P ress.Birnbaum, M. H., & Mellers, B. A. (19 79) . Stimulus recognition maymediate exposure effects. Journal of Personality an d Social Psychol-ogy, 37,391 -394.Blum, G. S. (195 4). An experimental reunion of psychoanalytic theory

with perceptual vigilance and defense. Journal of Abnormal and So -cial Psychology, 49 , 94-98.Bonnano, G. A., & Stilling, N. A. (1986). Preference, familiarity, andrecognition after repeated brief exposure to random geometricshapes. American Journal ofPsychology, 99 , 403-415.Bootzin, R. R., & Natsoulas, T. (19 65 ). Evidence for perceptual defenseuncontaminated by response bias. Journal of Personality and SocialPsychology, 5, 461-468.Bornstein, R. F. (1987). Subliminal mere exposure effects an d con-scious cognition: A study of attitude changes in response to stimuliperceived without awareness. Unpublished doctoral dissertation,State University of New York at Buffalo.Bornstein, R. F. (1989). Exposure and affect: Overview and meta-anal-ysis of research, 1968-1987. Psychological Bulletin, 106, 265-289.Bornstein, R. F , & D'Agostino, P. R. (19 92) . Stimulus recognition andthe mere exposure effect. Journal of Personality and Social Psychol-ogy, 63 , 545-552.Bornstein, R. F , Leone, D . R., & Galley, D. J. (198 7). The generaliz-ability of subliminal m ere exposure effects: Influence of stimuli per-ceived without awareness on social behavior. Journal of Personalityand Social Psychology, 53 , 1070-1079.Bower, G. H . (198 1). M ood and memory. American Psychologist, 36 ,129-148.Brooks, J. Q , III, & Watkins, M. J. (19 89). R ecognition memory andthe mere exposure effect. Journal of Experimental Psychology, 15,968-976.Bruner, J. S., & Postman, L. (1948). Perception under stress. Psycho-logical Review, 6, 314-323.Corteen, R. S., & Wood, B. (1972). Autonomic responses to shock-associated words in an unattended channel. Journal ofExperimentalPsychology, 94 , 308-313.Davis, L. B., & Porter, R. H. (1991). Persistent effects of early odorexposure on human neonates. Chemical Senses, 1 6, 169-174.den Heyer, K., B riand, K.., & Smith, L. (19 85 ). Automatic and strategiceffects in sema ntic prim ing: An examination of Becker's verificationmodel. Memory and Cognition, 13 , 22%-27il.DiChiara, G. (1990). In vivo brain dialysis of neurotransmitters.Trends in Pharmaceutical Science, 11 , 116-121.Dixon, N. F. (1981). Preconscious processing. New York: Wiley.Ekman, P. (19 72 ). Universals and cultural differences in facial expres-sions of emotion s. In J. Cole (Ed.), Nebraska Symposium on Motiva-tion. Lincoln: University of Nebraska P ress.Ellsworth, P. C, & Smith, C. A. (198 8). From appraisal to em otion:Differences among unpleasant feelings. Motivation and Emotion, 1 2,271-302.Erdelyi, M. H . (197 4). A new look at the New Look: Perceptual defenseand vigilance. Psychological Review, 81 , 1-25.Eriksen, C. W. (1963). Perception and personality. In J. M. Wepman &R. H. Heine (Eds.), Concepts of personality. Chicago: Aldine.Eriksen, C. W. (198 0). D iscrimination and learning without awareness:A methodological survey and evaluation. Psychological Review, 67 ,279-300.Fenichel, O. (1945). Th e psychoanalytic theory of neurosis. New \brk :Norton.Fink, E. L., Monahan, J. L., & Kaplowitz, S. A. (1989). A spatial modelof the m ere exposure effect. Communication Research, 16, 746-769.Franchina, J. J. (1991). Mere exposure to telereceptive cues facilitatesintake of a novel flavor in chickens (Gallus domesticus). Meeting ofthe M idwestern Psychological Association, Chicago, IL, 1990.

Freud, S. (1915). Th e unconscious. (Vol. XIV). London: HogarthPress.Garthwaite, J. (1 991 ). Glutam ate, nitric oxide and cell-cell signallingin the nervous system. Trends in Neuroscience, 14 , 60-67.Goldiamond, I. (1 958 ). Indicators of perception: I. Subliminal percep-tion, subception, nonconscious perception: An analysis in terms ofpsychophysical indicator methodology. Psychological Bulletin, 55,373-411.Greenwald, A. G. (1992). The New Look 3: Unconscious cognitionreclaimed. American Psychologist, 47,166-119.Greenwald, A. G., Klinger, M. R., & Lui, T. J. (1989). Unconsciousprocessing of dichoptically masked words. Memory and Cognition,17,35-41.Grier, J. B., Counter, S. A., & Shearer, W. M. (19 67). Prenatal auditoryimprinting in chickens. Science, 155, 1692-1693.Harrison, A. A. (1977). Mere exposure. In L. Berkowitz (Ed.), Ad -vances in experimental social psychology (Vol. 10, pp. 39- 83) . SanDiego, CA: Academic Press.Higgins, E. T, Rholes, W. S., & Jones, C. R. (197 7). C ategory accessi-bility and impression form ation. Journal of Experimental Social Psy-chology, 13,141-154.Holender, D. (1986). Semantic activation without conscious identifi-cation in dicho tic listening, parafoveal vision, and visual masking: Asurvey and appraisal. Behavioral an d Brain Sciences, 9, 1 -66.Isen, A. (1990). Affect and social behavior. In B. S. Moore and A. M.Isen (Eds.), New York: Cambridge Press.Kihlstrom, J. F. (1987). The cognitive unconscious. Science, 237,1445-1452.Kitayama, S. (1991 ). Impa irment of perception by positive and nega-tive affect. Cognition and Emotion, 5, 255-274.Kunst-Wilson, W. R., and Zajonc, R. B. (1980). Affective discrimina-tion of stimuli that cann ot be recognized. Science, 207, 557-558.Lazarus, R. S. (19 82). T houghts on the relationship between em otionand cognition. American Psychologist, 37, 1019-1024.Lazarus, R. S., & McCleary, R. A. (195 1). Autonomic discriminationwithout aw areness: A study of subception. Psychological Review, 58 ,113-122.Lewicki, P. (198 6). Nonconscious social information processing. SanDiego, CA: Academic Press.Lickliter, R., & Gottlieb, G. (198 6). Visually im printed m aternal pref-erence in ducklings is redirected by social interaction with siblings.Developmental Psychobiology, 19, 265-277.Mandler, G., Nakam ura, Y , & Van Zandt, B. J. S. (1987). Nonspecificeffects of exposure on stimuli that cannot be recognized. Journal ofExperimental Psychology: Learning,Memo ry, and Cognition, 13,646-648.Marcel, A. J. (198 3a). Conscious and unconscious perception: An ap-proach to the relation between visual masking and word recognition.Cognitive Psychology, 15, 197-237.Marcel, A. J. (1983 b). Conscious and unconscious perception: An ap-proach to the relation between phenomenal experience and percep-tual process. Cognitive Psychology, 15, 238-300.Maslach, C. (1979). Negative emotional biasing of unexplainedarousal. Journal ofPersonality and Social Psychology, 37, 953-969.Matlin, M. W. (1971). Response competition, recognition, and affect.Journal of Personality and Social Psychology, 19 , 295-300.McGinnies, E. M. ( 1949 ). E motionality and perceptual defense. Psy-chological Review, 56, 471-482.Merikle, P. M. (1982). Unconscious perception revisited. Perceptionan d Psychophysics, 31 , 298-301.Merikle, P. M. (1992). Perception without awareness: Critical issues.American Psychologist, 47, 792-795.Meyer, D., & Schwaneveblt, R. (19 71 ). Facilitation in recognizing pairsof words: Evidence of dependence between retrieval operations.Journal ofExperimental Psychology, 90, 227-234.Moreland, R. L., & Beach, S. R. (1992). Exposure effects in the class-


14/14

Documents

Additivity of Non Conscious Affect