The Mere Exposure Effect: Is It a Mere Case of Misattribution?

Angela Y. Lee, University of Toronto
[ to cite ]:
Angela Y. Lee (1994) ,"The Mere Exposure Effect: Is It a Mere Case of Misattribution?", in NA - Advances in Consumer Research Volume 21, eds. Chris T. Allen and Deborah Roedder John, Provo, UT : Association for Consumer Research, Pages: 270-275.

Advances in Consumer Research Volume 21, 1994      Pages 270-275

THE MERE EXPOSURE EFFECT: IS IT A MERE CASE OF MISATTRIBUTION?

Angela Y. Lee, University of Toronto

INTRODUCTION

A number of consumer studies indicate that product evaluations and brand choices are often made under minimal time and with little attention (e.g. Dickson & Sawyer, 1990). Under these conditions, choices may be made based on affective rather than cognitive responses (Baker & Lutz, 1987; MacInnis & Jaworski, 1989) towards the brand. The result of enhanced evaluation under these low levels of processing may be due to a mood congruent effect (eg., Isen, 1984), or it may be due to the mere exposure effect (Zajonc & Markus, 1982).

The focus of this paper is on the latter phenomenon, the mere exposure effect. Over 200 experiments investigating this effect have been published to-date (for a review, see Bornstein, 1989), and while there is still no general consensus as to the underlying mechanism that produces this effect, most explanations involve some form of uncertainty reduction. The purpose of this paper is to discuss the mere exposure effect in light of recent research on implicit memory and to present the results of two experiments which test a misattribution explanation of the mere exposure effect against an uncertainty reduction explanation.

THE MERE EXPOSURE EFFECT

By "mere exposure", researchers refer to a condition which "just makes the given stimulus accessible to the individual's perception" (Zajonc, 1968, p1) and the mere exposure effect is the enhancement of an attitude towards an object as a result of repeated exposures to that object.

Zajonc was one of the first psychologists interested in this effect. Subsequent research has demonstrated the effect across a wide variety of stimuli like drawings, photographs, ideographs, music, using a variety of rating procedures such as ratings of liking, pleasantness, and forced-choice preference judgements (Bornstein, 1989). A number of consumer behavior researchers became interested in the mere exposure effect when the idea that a lot of consumer learning occurs under low involvement conditions was introduced (Krugman, 1965). This subtle repetition effect has since been examined within the context of subliminal effects on brand preference (Hawkins, 1970) and political campaigns (Grush, McKeough & Ahlering, 1978).

It is, however, important to note that not all studies found the mere exposure effect. While most studies have found a monotonic relationship between repetition and evaluation, where evaluation increases with repetition, other studies have found a moderation effect where there is an inverted U-shaped relationship between exposure and affect (eg. Crandall, Montgomery, & Reese, 1973), and a few others have found a novelty effect where repeated exposure leads to a decrease in affect (eg. Cantor, 1968).

In the mere exposure paradigm, the typical experiment has two stages C a study stage and a test stage. Subjects are exposed to stimuli at different levels of repetition in the study stage, and are later asked to evaluate the stimuli amongst other similar but unpresented stimuli. The mere exposure effect is demonstrated by the evaluation enhancement observed for stimuli when they are "old" (presented in the study stage) versus "new" (not presented in the study stage).

A number of different explanations of the mere exposure effect have been proposed (for a review, see Sawyer, 1981). Most of them generally involve some form of uncertainty reduction; these include the two-factor model (Berlyne, 1970), response competition (Harrison, 1968), and expectancy arousal (Crandall, 1970). Other explanations include non-cognitive mediation (Zajonc, 1980), and misattribution (Jacoby & Kelley, 1987).

The uncertainty reduction model posits that organisms prefer stimuli that are predictable. As a stimulus becomes more familiar, it becomes increasingly predictable, less startling, and hence better liked. Berlyne (1966), for instance, argues that any stimulus has a certain amount of arousal potential. The repetition of high arousal potential stimuli will increase liking, whereas repetition of low arousal potential stimuli will lead to the moderation or novelty effect. Berlyne (1970) later adds a second factor to the model, proposing that affect toward a stimulus will become more positive as the stimulus becomes increasingly familiar, until boredom occurs and the frequency-affect curve turns downward.

Zajonc does not really offer a theory that explains the mere exposure effects, rather he argues that a separate affective response system is responsible for the effect, thereby refuting all theories positing that affect is based on cognition (Zajonc, 1980). Zajonc argues that any learning, and therefore the uncertainty reduction mechanism, requires cognition, as demonstrated by recognition. Thus empirical results indicating that enhanced affect can occur in the absence of recognition is supportive of a separate affective response system. This controversy has led to a series of studies which examine whether or not cognition plays a role in the mere exposure effect (eg. Anand, Holbrook, & Stephens, 1988; Obermiller, 1985).

Recent research by cognitive psychologists examining the distinction between explicit and implicit memory has provided a new approach to understanding the mere exposure effect. Jacoby and Kelley (1987), in particular, suggest that the enhanced affect towards a stimulus may be the by-product of a facilitation in information processing as a result of previous exposure to the stimulus.

REPETITION PRIMING AND MISATTRIBUTION

Recent research in memory distinguishes explicit memory, as characterized by conscious recollection of a past event, from implicit memory, which refers to a subsequent facilitation in performance that does not entail explicit retrieval of the experience. Numerous studies have demonstrated that prior exposure to a stimulus can facilitate subsequent performance on various tasks such as lexical decision, perceptual identification, picture naming, etc., without any reference to or conscious recollection of the episode. This facilitation effect is often referred to as 'repetition priming' (for a review, see Schacter, 1989). [In cognitive psychology, the term "priming" refers to the facilitation achieved in a task as a result of prior exposure, while in social cognition the term refers to the manipulation of prior exposure. In this paper, we use the former definition.] Jacoby suggests that the facilitation attained in various task performances is due to the fluency in the perceptual identification of the stimulus (Jacoby, Kelley, & Dywan, 1989) as a result of prior exposure. Furthermore, subjects may misattribute this perceptual fluency to sources other than the past experience, and cause them to (mis)judge that the duration of a word being presented is longer (Witherspoon & Allan, 1985), that the background noise accompanying the presentation of a sentence is lower (Jacoby, Allan, Collins & Larwill, 1988), that a statement is true, an argument valid, or a problem easy (Jacoby & Kelley, 1987), and more interesting to us, that a stimulus is more pleasant.

A number of experimental results on repetition priming that are of interest to the present research have been consistently replicated (Tulving & Schacter, 1990). Firstly, repetition priming can be obtained in the absence of recall or recognition. In addition, it has been demonstrated that there is stochastic independence between repetition priming and episodic memory performance like recall and recognition (Eich, 1984; Tulving, Schacter, & Stark, 1982; Schacter, Cooper & Delaney, 1990) in the sense that the performance of an individual on an implicit memory task does not help to predict his/her performance on an explicit memory task, and vice versa. Secondly, different independent variables are found to affect the two types of memory performance differently; in particular, functional dissociations between implicit and explicit memory have been observed in levels of processing manipulations (Jacoby & Dallas, 1981). Finally, repetition priming has also been found to be very sensitive to changes in the particular perceptual feature(s) being processed. Studies have shown that repetition priming effects are attenuated by a change in the surface information available at test (e.g. Roediger & Blaxton, 1987).

The mere exposure effect, like repetition priming, has been observed in the absence of stimuli recognition or recall in a number of studies (e.g. Mandler, Nakamura, & Van Zandt, 1987; Seamon, Brody & Kauff, 1983). The effect is also found to be more pronounced when surface feature processing rather than elaborative processing took place (Obermiller, 1985).

These similarities suggest that the mere exposure effect may be a result of repetition priming. As explained by Tulving, repetition priming is "a nonconscious form of learning that consists in the facilitation of perceptual identification of words and objects" (Tulving, in press). As learning takes place, the uncertainty reduction model may apply. Jacoby, however, goes one step further by arguing that while exposure to a stimulus promotes perceptual fluency, it is the attribution of the cause of this fluency that is critical to the evaluation of the stimulus. When asked to evaluate repeated stimuli, subjects misattribute the cause of the perceptual fluency to affect, rather than to prior exposure, producing the mere exposure effect.

The purpose of the present study is to investigate whether the mere exposure effect is a case of misattribution or of uncertainty reduction.

STUDY 1

As substantiated by research in memory, subjects exposed to a stimulus object develop perceptual fluency for that object. Extending Jacoby's misattribution explanation, subjects evaluating the stimulus are not aware that the perceptual fluency they enjoy while "sizing up" the stimulus comes from prior exposure, and misattribute the ease of processing to affect. So if subjects think they have seen the stimulus object before ("old" in terms of subjective familiarity) when they in fact have not ("new" in terms of objective familiarity), the lack of perceptual fluency may be misattributed as negative affect, resulting in a lower evaluation. On the other hand, if they think they have not seen the stimulus object before ("new" in terms of subjective familiarity) when in fact they have ("old" in terms of objective familiarity), then the perceptual fluency may be misattributed as positive affect, resulting in a higher evaluation. The evaluation thus reflects an effect mediated by subjective familiarity.

On the other hand, Berlyne and other uncertainty reduction advocates believe that the effect is mediated by learning. When the uncertainty that occurs from encountering a stimulus is reduced, subjects will experience enhanced affect. This uncertainty may come from a number of different sources, such as the uncertainty of how to react to the stimulus, the uncertainty of the task they have to do, or the uncertainty of their performance on the task. Therefore, if subjects have been repeatedly exposed to the stimulus, then the learning following exposure, in either conscious or nonconscious form, should result in some uncertainty reduction toward the stimulus. Furthermore, if subjects are informed as to whether or not they have seen the stimulus before, this should also reduce some of the uncertainty that may accompany the task of evaluating the stimulus.

By manipulating both objective familiarity and subjective familiarity, a study can be conducted to examine the processes underlying the mere exposure effect. Objective familiarity is manipulated by frequency of exposure, and subjective familiarity is manipulated by the use of explicit instructions to the subjects that they have or have not been exposed to the stimuli presented for evaluation. For ease of reference, the convenient usage of the terms "old" and "new" will refer to the two states of objective familiarity, and the terms "seen" and "not-seen" will refer to the two states of subjective familiarity.

If the mere exposure effect is entirely due to misattribution, then evaluation of the stimuli should be the same regardless of their being "old" or "new" when evaluation is not contaminated by misattribution of perceptual fluency. In other words, if subjects know whether or not they have encountered the stimuli before, then any perceptual fluency (or the lack of it) will not be misattributed, and evaluation of the stimuli will be unbiased.

H1a Evaluation will be the same under the "old/seen" condition and the "new/not-seen" condition.

When perceptual fluency occurs ("old") in the absence of subjective familiarity ("not-seen"), then misattribution causes an enhanced evaluation rating. However, when a new stimulus is encountered ("new") in the guise of subjective familiarity ("seen"), then the lack of perceptual fluency is misattributed to a less positive affect.

H1b Evaluation will be higher in the "old/not-seen" condition than in the "old/seen" condition; and lower in the "new/seen" condition than in the "new/not-seen" condition.

If, on the other hand, the mere exposure effect is due to uncertainty reduction, then evaluation should reflect a main effect of objective familiarity.

H2a Evaluation will be higher in the "old" condition than in the "new" condition.

For those subjects who are told that they have seen the stimuli before, they should feel more relaxed as a result of some uncertainty reduction towards the stimuli, resulting in higher evaluation.

H2b Evaluation will be higher in the "old/seen" condition than in the "old/not-seen" condition; and higher in the "new/seen" condition than in the "new/not-seen" condition.

The different predictions of the two models are represented as shown in Figure 1.

FIGURE 1

PREDICTIONS OF THE MISATTRIBUTION AND THE UNCERTAINTY REDUCTION MODELS

METHOD

Previous studies have measured subjective familiarity by having subjects report whether or not they had seen the stimuli while also providing evaluations of the stimuli (e.g. Anand et.al., Stephens, 1988; Obermiller, 1985; Wilson, 1979). This creates a number of problems. Firstly, assignment of subjects to this factor is not randomized. Secondly, the results may be due to taking the two measures at the same time. In this study, subjective familiarity is manipulated by informing the subjects that they had either seen or not seen the stimuli earlier and by using stimuli that can only be recognized at chance levels. Complexity of the stimuli was also manipulated, since previous research has sometimes indicated different effects for stimuli with different complexity (e.g. Bornstein, Kale, & Cornell, 1990).

Design

The experiment was a 2x2x2 factorial design with subjective familiarity ("seen" or "not-seen"), objective familiarity ("old" or "new"), and pattern density ("high" or "low") as between-subject factors.

Subjects

Sixty undergraduate students from a large urban university participated in the experiment as partial fulfilment for course credit. The subjects were randomly assigned to the different conditions.

Stimuli

The stimuli were abstract patterns generated by the program Gauss. Ten patterns were made up of 1500 randomly distributed points (high density patterns), with five patterns designated by little squares, and the other five designated by little triangles. Another ten patterns were made up of 800 randomly distributed points (low density patterns), again with five designated by little squares, and the rest by little triangles.

Two high density triangle patterns and two low density triangle patterns were randomly selected to serve as test patterns, and the rest were used as filler patterns.

Abstract patterns were selected for their lack of meaning and distinctive feature. The lack of meaning was intended to minimize within cell variance due to interpersonal differences, while the lack of distinctive feature was intended to ensure a chance recognition for the stimuli. The patterns were pretested for recognition in a forced choice task, and results demonstrated that recognition was at chance level (P(Rn) = 0.5, n = 31).

Two lists consisting of 10 patterns each were developed. List A was made up of 2 high density test patterns, 6 high density filler patterns, and 2 low density filler patterns, with squares and triangles counterbalanced. List B is made up of 2 low density test patterns, 6 low density filler patterns, and 2 high density filler patterns, squares and triangles again counterbalanced.

A total of 25 exposures were presented during the study stage: 3 exposures for each of the test patterns, and 2 or 3 exposures for each of the filler patterns. The sequence of exposures were randomized with the condition that there be no consecutive presentations of the same pattern.

Independent Variables

Objective familiarity was a between- subject manipulation. One half of the test patterns were presented to one group of subjects 3 times, and the other group of subjects 0 times. The reverse was true with the other half of the patterns.

Subjective familiarity was a between-subjects manipulation. Half of the subjects were told that the test patterns were new, and the other half were told that the patterns were old.

Pattern density was a between-subject manipulation. Half of the subjects had high density patterns as old patterns at time of rating, and the remaining half had low density patterns as old patterns at time of rating.

Dependent Measure

Evaluation was measured with a 7-point bipolar scale anchored with "Dislike Very Much" and "Like Very Much".

Procedure

Subjects were presented with a list of 25 patterns at the rate of 8s per exposure, and were asked to indicate whether the patterns were made up of squares or triangles while being exposed to each pattern. At the end of the study phase, the subjects engaged in a distractor task for five minutes. All subjects were then presented with two "old" and two "new" patterns at the rate of 10s per exposure. Half of the subjects were told that the patterns were ones that they had seen earlier, while the other half were told that the patterns were all new. The subjects were asked to evaluate the four patterns on a 7-point scale ranging from -3 to +3, anchored by 'Dislike Very Much' and 'Like Very Much'.

Results

The results were analyzed with a 2x2x2 ANOVA. The predicted main effect was not statistically significant for neither objective nor subjective familiarity. The density of the patterns was highly significant, with low density patterns preferred over the high density patterns (Meanlow = .46; Meanhigh = -.56; F(1,232) = 43.12; p < .001). The interaction between objective and subjective familiarity was, however, significant (F(1,232) = 4.51; p < .04), so was the interaction between density and objective familiarity (F(1,232) = 11.61; p < .001). The pattern of means indicated a mere exposure effect for the low density patterns, and a novelty effect for the high density patterns.

TABLE 1

MEAN RATING FOR AFFECTIVE RESPONSES BY PATTERN DENSITY, EXPOSURE, AND INSTRUCTION

In light of the interaction between objective and subjective familiarity, separate ANOVAs for old and new patterns were then conducted. With the "old" patterns, there was an insignificant main effect for subjective familiarity, and a significant main effect for density. With the "new" patterns, there was a significant main effect for density (F(1,116) = 5.04; p < .03), and a significant main effect for subjective familiarity (Meanseen = .53, Meannot-seen = 0.03; F(1,116) = 4.43; p < .04), partially supporting hypothesis H2b.

The significant interaction between density and objective familiarity also prompted separate ANOVAs for high and low density patterns to be conducted. With high density patterns, the main effect of objective familiarity was significant, but in the opposite direction of our prediction (Meanold = -0.42, Meannew = 0.02, F(1,116) = 2.87; p < .10). The main effect of subjective familiarity was not significant; however, the interaction between objective and subjective familiarity was significant (F(1,116) = 2.87; p < .10). Subsequent contrasts revealed that the evaluation under the "new/seen" condition was significantly higher than the "old/seen" condition (F(1,58) = 4.84; p < .04).

For low density patterns, the main effect of subjective familiarity was not significant, nor was the interaction between objective and subjective familiarity. But the main effect of objective familiarity was significant (F(1,116) = 10.59, p < .001), demonstrating the mere exposure effect, supporting hypothesis H2a in favour of the uncertainty reduction explanation. A comparison between evaluations in the "old/seen" condition and the "new/not-seen" condition shows that the two evaluations were significantly different (Meanold/seen = 1.23, Meannew/not-seen = 0.30, F(1,58) = 14.12; p < .001). Hypothesis H1a, which favours the misattribution explanation, was thus not supported. Evaluation in the "old/seen" condition was not significantly different from the "old/not-seen" condition (Meanold/seen = 1.23, Meanold/not-seen = 1.30, F(1,58) = 0.50; p <.90); however, evaluation in the "new/seen" condition was higher than in the "new/not-seen" condition as predicted by hypothesis H2b, although the difference was only marginally significant (Meannew/seen = .80, Meannew/not-seen = .30, F(1,58) = 2.56; p =.11). When the high density pattern data were included in the analysis, however, the comparison showed a significant subjective familiarity effect on the evaluation of "new" stimuli, as reported above.

Discussion

Clearly the misattribution explanation is not supported by the data. The results are more consistent with the uncertainty reduction explanation, although not all of the uncertainty reduction hypotheses are supported. The mere exposure effect was observed only for low density patterns under the objective familiarity "old" condition, providing evidence for enhanced evaluation following learning. Uncertainty reduction induced by informing subjects that they had seen the stimuli before did not influence evaluation of genuine "old" stimuli, but did produce enhanced evaluation for "new" stimuli. This may represent a ceiling effect achieved in terms of uncertainty reduction when all the stimuli were "old". Informing the subjects that stimuli about to be presented for evaluation were ones that they had seen before, on the other hand, reduced the uncertainty that subjects might have towards unfamiliar stimuli, resulting in an enhanced evaluation.

The novelty effect evidenced in the high density patterns was unexpected, because most studies had reported the mere exposure effect. Bornstein et. al. (1990; Exp.2) found a similar effect when he showed subjects stimuli with varying degrees of complexity. They suggested that when subjects were exposed to both interesting and uninteresting stimuli, evaluations of the two types of stimuli were made relative to each other, so any difference in their evaluations might be contrasted and accentuated, producing an overall decline in the evaluation of the less interesting stimuli. The degree of complexity was varied as a between-subject manipulation in a subsequent study (Exp.3), and the novelty effect for the less interesting stimuli was eliminated.

Since the low density patterns were rated much higher than the high density patterns, a similar stimulus contrast effect might be the reason for the novelty effect observed in the high density patterns in our study. A second study was undertaken to investigate the possibility of a stimulus contrast effect.

STUDY 2

Subjects

Nineteen undergraduate students from the same university participated voluntarily in the experiment.

Procedure

The subjects were again shown 25 patterns generated by the program Gauss, similar to the Study 1, except all the patterns were high density patterns. Subjects followed the same procedure and evaluated two old triangle patterns (at three exposures) and two new patterns (at 0 exposure) at test. Subjects were not told if the patterns were "old" or "new".

Results

A within-subject ANOVA indicated a significant main effect for objective familiarity, in the same direction as in Study 1 (Meanold = -0.42, Meannew = 0.29; p < .02.). The novelty effect persists for the high density patterns, in the absence of the more preferred stimuli.

GENERAL DISCUSSION

Developments in cognitive psychology on implicit memory and the results from various studies examining the mere exposure effect (eg., Anand & Sternthal, 1991; Obermiller, 1985) provide sufficient evidence to demonstrate a cognition-mediated mere exposure model. Research has shown that memory of an event can be expressed in terms of a conscious recollection of the event, and in terms of a facilitation in the performance of subsequent tasks. This facilitation, which is independent of recall or recognition of the event, provides evidence that cognitive processing takes place even though the event may not be consciously recollected.

Our data do not provide support for a misattribution explanation. Subjects were informed of the exposure status of the stimuli, and results still showed a preference for the "old" stimuli. It may be argued that even though subjects were aware of the status of exposure, they continued to misattribute the source of perceptual fluency to affective evaluation; although Whittlesea, Jacoby, and Girard (1990) reported a study in which the misattribution from manipulating visual clarity was eliminated when subjects were informed of the manipulation. Nonetheless, this remains a possibility that future research could address. An important lesson is that the mere exposure effect cannot be undone by knowledge of prior exposure. In other words, the consumer's enhanced evaluation for a brand as a result of repeated exposure to the product or to an ad will not be dampened by her/his knowledge that s/he has been repeatedly exposed to the product or the ad.

Furthermore, results from the two studies demonstrate that the mere exposure effect is stimulus-specific. A significant decrease in evaluation of high density patterns is observed with only three exposures, suggesting that evaluation either declined at the second exposure, or that the moderation effect set in at the frequency level of three exposures. When the repetition effect does not behave positively or uniformly across all stimuli, marketers should therefore be cautious in their selection of ad copies, packaging, logos, etc.

It may be interesting to apply the uncertainty reduction model in explaining results found in other mere exposure studies, and to compare their results with those found in the present study. A quick glance at a few studies (Anand et.al., 1988; Anand & Sternthal, 1991; Obermiller, 1985; Wilson, 1979) shows that evaluation is consistently highest for the "hits". [A foufold classification of actual exposure status and judged status yields "hit" (old item judged as old), "miss" (old item judged as new), "false alarm" (new item judged as old), and "correct rejection" (new item judged as new).] Under the objective familiarity "old" condition, "hits" are rated higher than the "misses"; whereas data in Study 1 showed that evaluations in the "old/seen" and the "old/not-seen" conditions were not different. Bear in mind that a chance (or near chance) level of recognition was manipulated in these studies, which means subjects were faced with having to make a decision that might or might not be correct. The uncertainty of their making a correct decision might add to the anxiety experienced by the subjects. Using the uncertainty reduction explanation, "hits" were evaluated higher than "misses" firstly because the uncertainty towards a stimulus was reduced through a feeling of familiarity. Secondly, in making a recognition response, subjects searched for a match of representation in memory, a "hit" represented situations when a match was successful. Based on a successful match, subjects were more certain of making a correct response, and felt more comfortable with the stimuli, resulting in a higher evaluation. In case of a "miss", subjects failed to find a match in memory, but might not be sure if a match truly did not exist in memory, or that it did, but was simply not accessible. This uncertainty was reflected in a lower rating for the stimuli compared to the situation when subjects were successful in finding a match.

Subjects in our Study 1 did not have the same dilemma. They did not have to perform the recognition task; rather, they were informed that they were going to evaluate patterns that they had either seen or not seen earlier. Subjects were asked during debriefing if they were able to recognize any of the patterns. None of them said they were able to do so. Typically, they said that although they could tell the patterns were different from each other, they were not able to recognize which ones they saw because they all looked similar. The subjects also said they believed the experimenter when told that they had either seen or not seen the stimuli to be evaluated. Consequently, it seems that the instructions and the absence of a recognition task provided the subjects with some certainty that subjects in other studies had not enjoyed. This brought the evaluation in the "old/seen" condition on par with the "old/not-seen" condition.

An alternative explanation could be that telling subjects whether or not they saw the stimuli earlier was not comparable to the feeling of subjective familiarity that subjects relied on to make their recognition responses. So the "old/seen" and "old/not-seen" cells in Study 1 were in fact comprised of both "hit" and "miss" data in the other studies. The external instruction had no effect on evaluation when perceptual fluency was present, and the equivalence in the two cells was an artifact that resulted from combining the "hit" and the "miss" responses.

A more interesting explanation for 'hits' enjoying higher evaluation than 'misses' in other studies may be an issue of the direction of causality. Instead of familiarity mediating recognition and affective responses, recognition responses may be mediated by affective responses, especially when stimuli could only be recognized at chance level. Within the context of implicit memory and uncertainty reduction, it may be argued that implicit learning produces enhanced evaluation, which in turns mediates an explicit memory response. This possibility is currently being tested.

Results from our objective familiarity "new" condition do, however, match the pattern of "false alarm" and "correct rejection" data in Wilson's (1979) study and Anand & Sternthal's (1991) study. Learning could not have taken place when stimuli had not been exposed. That explains the main effect of exposure for these two cells. Between these two conditions, however, subjects in the "false alarm" condition erroneously decided that they had found a match in memory, which made them feel more at ease with the stimulus than those subjects in the "correct rejection" condition who did not find a match, and felt more apprehensive. Subjects in our Study 1 were told that they had seen the stimuli before, even though they had not. This induced sense of familiarity C that they are dealing with "old friends" C is sufficient to boost comfort level which was reflected in their higher rating compared to those who were told that they were going to be dealing with "strangers".

In summary, the mere exposure effect is an important phenomenon to study in understanding consumer behavior, especially under low involvement situations when product evaluations and brand choices are typically influenced by affect. The present study provides additional insight in understanding this phenomenon and suggests how implicit memory may influence the consumer's decision making process.

BIBLIOGRAPHY

Anand, P., M.B. Holbrook, & D. Stephens (1988), "The formation of affective judgments: The cognitive-affective model versus the independence hypothesis," Journal of Consumer Research, 15, 386- 391.

Anand, P., & B. Sternthal (1991), "Perceptual fluency and affect without recognition," Memory & Cognition, 13(3) 293-300.

Baker, W.E., & R.J. Lutz (1987), "Levels of processing and levels of explanation," in S. Hecker & D.W. Stewart (Eds.), Nonverbal Communication in Advertising. Lexington: Lexington Books, 84-89.

Berlyne, D.E. (1966), "Curiosity and exploration," Science, 153, 25-33.

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

Bornstein, R.F. (1989), "Exposure and affect: Overview and meta- analysis of research, 1968-1987," Psychological Bulletin, 106, 265- 289.

Bornstein, R.F., A.R. Kale, & K.R. Cornell (1990), "Boredom as a limiting condition on the mere exposure effect," Journal of Personality and Social Psychology, 58, 791-800.

Cantor, G.N. (1968), "Children's 'like-dislike' ratings of familiarized and nonfamiliarized visual stimuli," Journal of Experimental Child Psychology, 651-657.

Crandall, J.E. (1970), "Predictive value and confirmability of traits as determinants of judged trait importance," Journal of Personality, 38, 77-90.

Crandall, J.E., V.E. Montgomery, & W.W. Reese (1973), "Mere exposure versus familiarity, with implications for response competition and expectancy arousal hypothesis," Journal of General Psychology, 88, 105-120.

Dickson, P.R., & A.G. Sawyer (1990), "The price knowledge and search of supermarket shoppers," Journal of Marketing, 54, 42-53.

Grush, J.E., K.L. McKeough, & R.F. Ahlering (1978), "Extrapolating laboratory exposure research to actual political elections," Journal of Personality and Social Psychology, 36, 257-270.

Harrison, A.A. (1968), "Response competition, frequency, exploratory behavior, and liking," Journal of Personality and Social Psychology, 9, 363-368.

Hawkins, D. (1970), "The effects of subliminal stimulation on drive level and brand preference," Journal of Marketing Research, 7, 322- 326.

Isen, A.M. (1984), "The influence of positive affect on decision making and cognitive organization," Advances in Consumer Research, 11, 534-537.

Jacoby, L.L., L.G. Allan, J.C. Collins, & L.K. Larwill (1988), "Memory influences subjective experience: Noise judgments," Journal of Experimental Psychology: Learning, Memory and Cognition, 14, 240-247.

Jacoby, L.L., & M. Dallas (1981), "On the relationship between autobiographical memory and perceptual learning," Journal of Experimental Psychology: General, 3, 306-340.

Jacoby, L.L., & C.M. Kelley (1987), "Unconscious influences of memory for a prior event," Personality and Social Psychology Bulletin, 13, 314-336.

Jacoby, L.L., C.M. Kelley, & J. Dywan (1989), "Memory Attributions," in H.L. Roediger & F.I.M. Craik (Eds.), Varieties of Memory and Consciousness: Essays in Honour of Endel Tulving. Hillsdale: Erlbaum.

Krugman, H. (1965), "The impact of television advertising: Learning without involvement," Public Opinion Quarterly, 29 (Fall), 349-356.

Mandler, G., Y. Nakamura, & B.J. Van Zandt (1987), "Nonspecific effects of exposure to stimuli that cannot be recognized," Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 646-648.

MacInnis, D.J., & B.J. Jaworski (1989), "Information processing from advertisements: Toward an integrative framework," Journal of Marketing, 53, 1-23.

Obermiller, C. (1985), "Varieties of mere exposure: The effects of processing style and repetition on affective response," Journal of Consumer Research, 12, 17-30.

Roediger, H.L., III, & T.A. Blaxton (1987), "Retrieval modes produce dissociations in memory for surface information," in D.S. Gorfein & R.R. Hoffman (Eds.), Memory and Cognitive Processes: The Ebbinghaus Centennial Conference (pp.349-379). Hillsdale: Erlbaum.

Sawyer, A. (1981), "Repetition, cognitive responses and persuasion," in R.E. Petty, T.M. Ostrom, & T.C. Brock (Eds.), Cognitive responses in persuasion (pp.237-261). Hillsdale: Erlbaum.

Schacter, D.L. (1989), "Implicit memory: History and current status," Journal of Experimental Psychology: Learning, Memory, and Cognition, 13, 501-518.

Schacter, D.L., L.A. Cooper, & S.M. Delaney (1990), "Implicit memory for unfamiliar objects depends on access to structural descriptions," Journal of Experimental Psychology: General, 119, 5- 24.

Seamon, J.G., N. Brody, & D.M. Kauff (1983), "Affective discrimination of stimuli that are not recognized: Effects of shadowing, masking and cerebral laterality," Journal of Experimental Psychology: Learning, Memory, and Cognition, 9, 544-555.

Tulving, E. (in press), "Conscious and nonconscious memory: What kind of a difference?" in A. Baddeley and L. Weiskrantz (Eds.), Attention: Selection, Awareness and Control. A Tribute to Donald Broadbent. New York: Oxford University Press.

Tulving, E., & D.L. Schacter (1990), "Priming and human memory systems," Science, 247, 301-305.

Tulving, E., D.L. Schacter, & H.A. Stark (1992), "Priming effects in word-fragment completion are independent of recognition memory," Journal of Experimental Psychology: Learning, Memory, and Cognition, 8, 336-342.

Wilson, W.R. (1979), "Feeling more than we can know: Exposure effects without learning," Journal of Personality and Social Psychology, 37, 811-821.

Witherspoon, D., & L.G. Allan (1985), "The effects of a prior presentation on temporal judgments in a perceptual identification task," Memory & Cognition, 13, 101-111.

Whittlesea, B.W.A., L.L. Jacoby, & K.A. Girard, K.A. (1990), "Illusions of immediate memory: Evidence of an attributional basis for feelings of familiarity and perceptual quality," Journal of Memory and Language, 29, 716-732.

Zajonc, R.B. (1968), "Attitudinal effects of mere exposure," Journal of Personality and Social Psychology Monographs, 9(2, Pt.2).

Zajonc, R.B. (1980), "Feeling and thinking: Preferences need no inferences," American Psychologist, 35, 151-175.

Zajonc, R.B., & H. Markus (1982), "Affective and cognitive factors in preferences," Journal of Consumer Research, 9, 123-131.

----------------------------------------