Mapping the Relationship Between Preattentive Processing and Attitudes

Stewart Shapiro, University of Arizona
Deborah J. MacInnis, University of Arizona
ABSTRACT - While prior research suggests that preattentively processed information can influence attitudes (Janiszewski 1988), it has not (1) provided independent evidence for the existence of preattentive processing or (2) demonstrated why preattentively processed material may influence attitudes. This study attempts to address both issues. First, implicit memory is used as a process measure to document the existence of preattentive processing and its relationship to ad affect. Second, three possible mechanisms underlying the effect of preattentive processing on affect (mere exposure, classical conditioning and mood) are investigated. The results suggest that mere exposure accounts for the preattentive processing effect. Further, results indicate that given equal levels of priming, affect change occurred only in the three exposure (not the one exposure) condition suggesting that the mere presence of an implicit memory trace is not sufficient to influence affective responses.
[ to cite ]:
Stewart Shapiro and Deborah J. MacInnis (1992) ,"Mapping the Relationship Between Preattentive Processing and Attitudes", in NA - Advances in Consumer Research Volume 19, eds. John F. Sherry, Jr. and Brian Sternthal, Provo, UT : Association for Consumer Research, Pages: 505-513.

Advances in Consumer Research Volume 19, 1992      Pages 505-513

MAPPING THE RELATIONSHIP BETWEEN PREATTENTIVE PROCESSING AND ATTITUDES

Stewart Shapiro, University of Arizona

Deborah J. MacInnis, University of Arizona

[The authors would like to thank Susan Heckler for her comments as well as Kent Nakamoto and Jonathan Frenzen for their help with an earlier draft of this paper. The helpful comments of two anonymous reviewers are also gratefully acknowledged.]

ABSTRACT -

While prior research suggests that preattentively processed information can influence attitudes (Janiszewski 1988), it has not (1) provided independent evidence for the existence of preattentive processing or (2) demonstrated why preattentively processed material may influence attitudes. This study attempts to address both issues. First, implicit memory is used as a process measure to document the existence of preattentive processing and its relationship to ad affect. Second, three possible mechanisms underlying the effect of preattentive processing on affect (mere exposure, classical conditioning and mood) are investigated. The results suggest that mere exposure accounts for the preattentive processing effect. Further, results indicate that given equal levels of priming, affect change occurred only in the three exposure (not the one exposure) condition suggesting that the mere presence of an implicit memory trace is not sufficient to influence affective responses.

INTRODUCTION

Though it has been proposed that attitude formation is a cognitive process involving focused, conscious attention (Fishbein and Ajzen 1975; Hastak and Olson 1989, Lutz 1975; 1977, Olson et al 1982), recent literature in psychology and marketing proposes that attitudes can also be formed preattentively (Dixon 1981; Janiszewski 1988; Kihlstrom 1984; 1987; Moreland and Zajonc 1977; Wilson 1979). In an advertising context, for example, Janiszewski (1988) showed that ad attitudes can be influenced when information in the ad is processed at a preattentive level, and that some relationship exists between the nature of the material to be processed (visual vs. verbal) and the dominant hemisphere in which the information is processed (right vs. left).

While interesting, much needs to be learned about preattentive processing effects on attitudes. Specifically, greater effort is needed to clarify that preattentive (vs. conscious) processing has actually taken place. Furthermore, greater effort is needed to uncover the specific processes which may explain attitude formation under the preattentive processing route. The purpose of this paper is two-fold: (1) to explore a methodology for assessing the existence of preattentive processing, specifically through the use of implicit memory, and (2) to test three possible mechanisms-- mere exposure, classical conditioning, and mood-- as processes potentially underlying the preattentive processing effect.

The paper is divided into four sections. The first describes preattentive processing, defines key constructs, and identifies factors thought to affect preattentive processing. The second section identifies potential mechanisms underlying preattentive processing effects on ad attitudes, and their possible dependence on implicit memory. An experiment designed to test each mechanism and its link to implicit memory is described in the third section. The final section describes limitations and implications of the study, as well as directions for future research.

PREATTENTIVE PROCESSING

Discussion of preattentive processing first requires differentiating conscious from nonconscious experiences. Consciousness includes both monitoring and controlling ourselves and the environment so that memories and thoughts that influence behavior and cognition are both represented in phenomenal awareness and can be voluntarily initiated and terminated (Kihlstrom 1984). Nonconsciousness involves memories and thoughts which are not represented in phenomenal awareness and which cannot be voluntarily initiated or terminated (Kihlstrom 1984). Preattentive processing refers to information not in focal attention but which receives cognitive processing (Kihlstrom 1984), and hence can be analyzed for its emotional significance and semantic meaning (Dixon 1981).

Preattentive processing has been studied using a variety of methodologies (e.g., dichotic listening and masking paradigms). In consumer research the methodology used by Janiszewski (1988) involved two factors, hemispheric processing style and contralateral conduction.

The two hemispheres of the brain have been shown to use two different processing styles (Bouma 1978; Sergent 1983). The right hemisphere is said to have a holistic style, allowing the simultaneous integration of multiple pieces of information. This hemisphere uses a template matching process, invoking a template to give meaning to incoming information (Janiszewski 1988). Because of this holistic processing style, the right hemisphere is more compatible than the left with processing pictorial stimuli. The left hemisphere is described as unit-integrative, as it combines well-learned individual units into a meaningful whole (Janiszewski 1988). Because of this unit-integrative style, the left hemisphere is more compatible than the right with the processing of verbal stimuli.

Contralateral conduction refers to the fact that stimuli observed in the right visual field are processed in the left hemisphere of the brain, while stimuli observed in the left visual field are processed in the right hemisphere. Combined with the differences in hemispheric processing styles noted above, this phenomenon implies that visual stimuli presented in the left visual field should receive greater processing than verbal stimuli presented in the left visual field. Conversely, verbal stimuli presented in the right visual field should receive greater processing than visual stimuli presented in the same field.

Janiszewski (1988) extends notions of preattentive processing, hemispheric processing style and contralateral conduction to make predictions about the effect of preattentively processed stimuli on affective responses. Subjects' attention was focused on a task (e.g.,an article) with a critical visual ad placed in subjects' peripheral vision to the right or left of the attended task (right or left visual field). This peripherally placed ad was the stimulus assumed to be preattentively processed. Following exposure to the article, subjects' attitudes toward the preattentively processed stimulus were assessed. The results indicated when the attended task was verbal, the pictorial critical ad was most preferred when it was placed in the left vs. the right visual field, even when subjects were unaware of having seen the ad.

While interesting, additional research in this area is needed to clarify that (1) preattentive processing did indeed take place and (2) what processes account for the observed effects. We propose a methodology (implicit memory) for assessing the existence of preattentive processing, and propose three potential processes as underlying the existence of the preattentive processing-affect relationship.

THEORY AND HYPOTHESES

Implicit Memory - A Process Check for Preattentive Processing

One problem with past research on preattentive processing in an advertising context is that the existence of preattentive processing is determined by its effects. However, using affect change as both an indicator of preattentive processing and as an outcome of such processing is problematic. Hence independent evidence for the existence of such processing is necessary to demonstrate the preattentive processing-affect relationship. By decoupling the measure of preattentive processing from that of its effects, it becomes possible to directly demonstrate that preattentively processed stimuli do influence affect formation processes. One indicator of preattentive processing is assessing whether stimuli supposedly processed at a preattentive level leave an implicit memory trace. As the implicit memory literature is just now burgeoning in psychology, and is unlikely to be familiar to most readers, a brief overview is in order.

Implicit Memory

Implicit memory refers to the expression of previously encoded information without conscious or deliberate recollection of prior exposure to such information (Schacter 1987). If information has been processed preattentively (e.g., the word Stripe), for example, this information is likely to be primed in memory, and hence more accessible, even though the individual may not tie the primed information with the preattentive processing task. Often implicit memory is measured through indirect memory tasks which, unlike direct memory tasks, do not ask subjects to refer back to recently presented information when completing the memory task. In a stem completion task, for example, the subject may be presented with the stem STR and asked to complete the word. If the word STRIPE has been primed in memory (through preattentive or other processes), one would expect a greater number of primed subjects to complete the above stem with the word STRIPE than with some other word.

Placement Effects, Implicit Memory, and Affect

Because not all stimuli are preattentively processed, evidence for preattentive processing effects should be demonstrated only in the conditions that favor preattentive processing. It is thus proposed that subjects in exposure conditions thought to be conducive to preattentive processing will demonstrate significantly greater priming for the exposure content than subjects in exposure conditions less conducive to preattentive processing. As indicated earlier, contralateral organization effects propose that the nature of the stimulus (visual or verbal) and its placement (left or right visual field/right or left hemisphere) produce effects that are more or less compatible with preattentive processing. Thus, we would expect greater preattentive processing (and greater implicit memory) when pictorial information is placed in the left vs. right visual field, and when verbal information is placed in the right vs. left visual field. As this study uses only visual stimuli to scale down the study size, we expect that:

H1: Subjects exposed to pictorial information in the left visual field will demonstrate significantly greater implicit memory traces for the visual information than subjects exposed to visual information in the right visual field.

If preattentive processing does promote affect, one would expect to replicate Janiszewski's (1988) finding of the relationship between placement effects and ad affect. Hence:

H2: Subjects exposed to pictorial information in the left visual field will demonstrate significantly more positive affect for the visual information than subjects exposed to visual information in the right visual field.

While the above hypotheses are geared toward establishing a direct linkage between exposure conditions, preattentive processing and affect, they do not indicate the process by which preattentively processed material results in more positive affect. Three potential processes by which this effect may occur (mere exposure, classical conditioning, and mood) are explored below.

Mere Exposure

Mere exposure theory asserts that a positive relationship can be found between the level of stimulus exposure and preference for the stimulus (Moreland and Zajonc 1977; Wilson 1979). Formally, the theory states that the higher one's objective familiarity with a stimulus, the greater one's preference for it. Note that mere exposure theory is concerned with objective, not subjective familiarity.

Uncertainty reduction, a proposed explanation for the mere exposure effect, asserts that cognitive elaboration during encoding reduces uncertainty, resulting in greater familiarity with, and hence greater positive affect toward the stimulus (Obermiller 1985). Suggesting cognitive elaboration during encoding does not eliminate the possibility that this process is done preattentively, for higher levels of mental processes involved in judgment and problem solving can all occur outside of phenomenal awareness (Kihlstrom 1987). If ad attitudes of preattended stimuli are affected in a manner suggested by the mere exposure theory, research which systematically manipulates exposure levels should find increasingly positive ad attitudes for preattentively processed ads as exposure increases.

If mere exposure does operate as a theoretical explanation for attitude effects observed from preattentively processed stimuli, pictorial ads placed in the left visual field should be most affected by an increase in exposure levels relative to pictorial ads placed in the right visual field. Specifically, uncertainty reduction (and hence mere familiarity) should occur only for those ads that are preattentively processed. Hence mere exposure theory posits that:

H3: When the stimulus is pictorial, ad attitudes will increase with increasing exposure rates, but only when the ad is placed in the left (vs. the right) visual field.

Classical Conditioning

Classical conditioning involves exposing subjects to a positively or negatively valenced unconditioned stimulus such as music (Gorn 1982) or visual images (Stuart et al 1987) along with or just preceding a neutrally valenced stimulus such as the ad or brand. Classical conditioning asserts that the pairing of a positively (negatively) valenced unconditioned stimulus with a neutrally valenced target ad (conditioned stimulus) will lead to positive (negative) affective response toward the target ad.

If ad attitudes of preattended stimuli are affected in a manner suggested by classical conditioning, research which systematically manipulates the valence of the attended task should find an interaction between stimulus valence (positive vs. negative) and placement (left vs. right visual field). Specifically, if the relationship between preattentive processing and affect is driven by classical conditioning, one would expect that:

H4: When the stimulus is pictorial, ad attitudes are most favorable when the attended task is positively valenced and the ad is placed in the left visual field, and least favorable when the attended task is negatively valenced and the ad is placed in the left visual field.

Though conditioning can occur with the first paired CS-US exposure (Stuart et al 1987), it is hypothesized that greater exposures should strengthen the link between the CS and US, resulting in more extreme affect with greater levels of exposure. Valence and exposure levels should also interact with hemispheric compatibility. Specifically:

H5: When the attended task is negatively (positively) valenced, ad attitudes will be more negative (positive) with increasing exposure rates, but only when the stimulus is processed in the compatible hemisphere.

The use of the negatively valenced attended task helps differentiate classical conditioning from mere exposure. While both classical conditioning and mere exposure predict increases in ad attitudes when positive stimuli are used for the attended task, classical conditioning posits negative attitudes with repeated exposures for the negatively valenced attended task, while mere exposure posits positive attitudes with repeated exposures for the negatively valenced attended task.

Distinguishing Classical Conditioning from Mood

It may be argued that the effects of classical conditioning could be attributed not to the involuntary response conditioned by the CS-US pairing, but rather to the mood created by the unconditioned stimulus. Mood effects literature suggests that a positively (negatively) valenced stimulus (e.g., a happy/sad article) may put the reader in a positive (negative) mood, which would in turn, increase the accessibility of related thoughts in memory (Goldberg and Gorn 1987). When asked to rate their preference for an otherwise neutral stimulus (ad), the activated mood would prime likened material in memory, and the person would likely rate the stimulus (ad) in a mood congruent direction.

Having subjects rate ads that were both present and not present in the experimental context should enable an assessment of which theory is at work. Mood effects theory would predict that all ads evaluated after a positive (negative) mood induction would be rated higher (lower) regardless of whether they had been previously seen. In contrast, classical conditioning would predict effects only for those ads previously paired with the positively (negatively) valenced stimulus. Thus classical conditioning theory also predicts that:

H6: Consumers exposed to the positively (negatively) valenced unconditioned stimulus will have more (less) favorable ad attitudes only for those ads previously present in the exposure set.

FIGURE 1

THE CRITICAL OBJECT

METHOD

Stimuli

Two types of stimuli were developed for use in the study: preattentively processed stimuli and attended stimuli. For use in this study, preattentively processed stimuli had to be pictorial in nature, adaptive to an indirect memory test, and neutral in valence. The attended stimuli needed to differ in valence (positive vs. negative). Pretests designed to develop these stimuli are described below.

Preattended stimuli. Six dot patterns were created and assigned to 40 subjects (N=40 per dot pattern), who were instructed to draw lines to connect the dots in each pattern in any way they desired. This pretest was required to obtain baseline figures for the indirect memory test. The baseline figure represents the percentage of subjects who connected the dots to form the critical object used as the preattended stimuli.

Twenty four objects were formed by connecting the dots in each of the 6 dot patterns in four different ways. Thirty subjects were randomly assigned eight of the 24 objects (N=10 per object) and asked to evaluate each on four, nine-point bipolar items: unattractive/attractive, unappealing/appealing, unlikable/likable, unpleasant/pleasant (alpha=.95). T-test comparisons indicated that 8 of the 24 objects were rated as having affective responses no different from the neutral rating of 5 (p>.05). Of those 8, two had acceptable baseline levels of dot completions to form the object [18.42% (7/38) and 15.00% (6/40)]. The object with the higher baseline level was randomly chosen as the critical preattentive stimulus. This stimulus is shown in Figure 1.

Attended stimuli. Seven newspaper articles, each 330 words in length, were taken from the Sunday edition of the Chicago Tribune. Seventy undergraduate subjects participating in a study ostensibly on processing style, were randomly assigned to one of the seven articles (N=10 per article). Subjects were told to read the article as quickly as possible while at the same time reading for detail. Following completion of the article, each subject evaluated the article on a set of nine-point bipolar items: unpleasant/pleasant, unlikable/likable, unenjoyable/enjoyable (alpha=.94). Based on these pretest results, several modifications in the articles, and further pretesting, articles evoking strong positive and strong negative affective responses were identified. Subjects also indicated on a ten point scale the extent to which the article was very difficult to understand/very easy to understand and very difficult to follow/very easy to follow (alpha=.94). None of the articles differed in difficulty level, and each was found to be relatively easy to read (X=8.71).

Subjects

Subjects were sixty three undergraduate students who participated in the experiment for course credit (N=44) or as part of a chance to win a $50.00 cash prize lottery (N=19). Two subjects were suspicious of the true purpose of the experiment and were discarded. Due to time constraints, the remaining subjects were randomly assigned to only the 6 cells critical to the predictions made in this study, allowing at least 10 subjects per each of the 6 conditions. Subjects were run in groups ranging in size from 3 to 18. Experimental sessions lasted approximately 40 minutes. When possible, all 6 conditions were represented during each experimental session.

TABLE 1

INDIRECT MEMORY TEST RESULTS

Design and Procedure

A 2 x 2 x 2 between subjects experimental design was used, with valence of the attended task (positive vs. negative), exposure level ( 1 vs 3), and placement of the critical object (right vs. left visual field) as the independent variables. Data for two of the experimental conditions (see Table 1) are still in the process of being collected.

Subjects were informed that the purpose of the study was to examine how individuals' processing style affected their memory for the details of previously read newspaper articles. Subjects were told they would be exposed to a total of four short newspaper articles. Subjects were informed that they would be given an unspecified time limit to read the four articles, and that undivided attention to the articles would allow them to finish all articles before the time limit expired. Subjects were instructed to start with the article on the first page of the newspaper, and proceed to the next article until all four articles marked with an orange dot were read. An ostensible time limit was imposed to focus attention on the article and away from the critical object. In reality, subjects were not under any time constraint, and all were allowed to finish reading all four articles.

The stimulus material subjects received contained four double-sided mock newspaper pages. The first page of the newspaper was a no-treatment page. The following three pages contained the positively or negatively valenced articles. In the one exposure condition, the critical object was placed alongside (right or left) of only one of the articles. Assignment of the critical object to the specific article was random. In the three exposure condition, the critical object was placed alongside (right or left) of each of the three articles following the first no treatment page. For a more detailed description of the newspaper layout see Janiszewski (1988).

After reading the last article, subjects completed a questionnaire which ostensibly measured processing style. They then completed the implicit memory task. As with the pretest, subjects were given the 6 dot patterns-- one per page-- and were asked to connect the dots in any way they felt fit. One of the dot patterns represented the critical stimulus. The order of dot patterns was randomized across subjects. To ensure that performance on the dot completion task was attributable to implicit vs. explicit memory, subjects were asked several questions that assessed their hypotheses about the purpose of the dot completion task, and their awareness of the relationship between the newspaper and the dot completion task (Bowers and Schacter 1990). Subjects who made no mention of the newspaper or objects within the newspaper were classified as test unaware. Only their responses were retained to test the implicit memory hypotheses.

The third part of the questionnaire measured subjects' affect toward the critical object, two other objects present in the newspaper, and three objects not present in the newspaper but rated as neutral in the pretests. After rating all six objects, subjects were instructed to re-examine each object and indicate whether they recalled having seen it in the newspaper. Order of objects was randomized across subjects. The final part of the questionnaire assessed affect toward and perceived difficulty of the three newspaper articles.

Dependent Measures

Implicit memory was assessed by comparing the percentage of subjects who were test unaware and who formed the critical object in the dot completion task with the baseline percentage who formed the critical object in the pretest. A similar dot completion measure was successfully used to show priming of novel nonverbal patterns in a study by Musen and Treisman (1990).

Evaluation of the critical object, the noncritical objects in the newspaper, and the noncritical objects not in the newspaper was assessed by an averaged composite measure (alpha=.95) of subjects' ratings of the object as unappealing/ appealing, unattractive/attractive, unlikable/likable, unpleasant/pleasant. Each item was assessed on a nine-point scale.

The valence of the attended task was assessed by three nine-point bipolar items: unpleasant/pleasant, unlikable/likable, unenjoyable/enjoyable. Items were averaged to form a composite index (alpha= .95). Perceived difficulty of the articles was assessed by taking the average of two ten-point bipolar items: very difficult to understand/very easy to understand, and very difficult to follow/very easy to follow (alpha= .86).

RESULTS

Manipulation Checks Results

To check on the success of the valence of the attended task manipulation, the average ratings for the positively valenced attended task (X=7.98) and the negatively valenced attended task (X=2.82) were compared to the neutral score of 5. One-tailed t-tests indicated the valence of each task was significantly different from the neutral score (both p< .001). These results confirm the success of the experimental manipulation. Additionally, perceived difficulty of the articles was not found to be any different between the positively and negatively valenced conditions (p > .1).

Data on handedness and gender were also collected and used as covariates for all ANOVA analyses reported in the results section. As neither variable was found to be significant in any of the results, only the simple ANOVAs are reported here.

Recognition Test Results

Differences in recognition of the critical object across the two placement or the two exposure conditions would be problematic as such differences might suggest that conscious processing had taken place in one condition but not the other. To examine whether such differences existed, two nonparametric tests for comparison of two proportions were conducted using the recognition measure as the dependent variable. The recognition levels were not different when the stimulus was placed in the left (20%) vs. the right (19.35%) visual field. Nor were recognition levels different in the one (15%) vs. three exposure (28.6%) conditions (both p's > .10). Thus any observed differences in the results should not be attributable to differences in recognition levels (conscious processing) across the various treatments. Additionally, a test for a positive influence of recognition on attitude toward the object was not significant (p > .1).

Test of the Hypotheses

H1 predicted that subjects exposed to the critical object in the left visual field would demonstrate significantly greater implicit memory for the critical object than subjects exposed to the same object in the right visual field. To test this hypothesis, a nonparametric test for comparison of two proportions was conducted for each of the six experimental cells. The results showed greater implicit memory for the positive/one exposure/right visual field and negative/three exposure/right visual field conditions (see Table 1). These two conditions did not differ from one another in their level of implicit memory. The results indicate that contrary to H1, implicit memory for visual stimuli is greater when they are placed in the right, not the left visual field. These results do not support H1.

H2 predicted more positive affect for the critical object when it was placed in the left vs. the right visual field. Mean affect ratings for the critical object are shown in Table 2. An ANOVA on affect toward the critical object revealed neither a main effect for placement (F=1.53, p>.10), nor a main effect for valence (F<1, p >.10). However, a placement by exposure interaction was observed (F=4.06, p<.05) with highest preferences in the negative/three exposure/right visual field condition. The observed interaction runs counter to H2 which predicted the object would be most preferred when placed in the left visual field. Thus, though a placement effect did exist it (1) was opposite the placement condition predicted and (2) was contingent on numbers of exposures.

H3 predicted that if the observed results are attributable to mere exposure, attitude toward the critical object should increase with increasing exposure, but only when the critical stimulus is placed in the left visual field. While the placement effect is reversed, as noted above, the proposed interaction did emerge. Preferences did increase with exposure, and the impact of exposure on preferences seemed to depend on where the object was placed. To gain greater evidence for the existence of the mere exposure mechanism, further analysis was conducted by cell. Specifically, a one-tailed test indicated that the evaluation of the critical object in the negative/three exposure/right visual field condition was significantly greater than the neutral rating of 5. Thus, under certain conditions, preattentive processing of pictorial stimuli can lead to increased affective responses in a way generally consistent with mere exposure theory.

H4 and H5 predicted that if the observed results are attributable to classical conditioning, evaluations of the critical object should be most positive (negative) when the attended task is positively (negatively) valenced, particularly when the critical stimulus was processed in the appropriate visual field, and presented multiple times. However, the results revealed no main effects for valence of the attended task, nor interactions of attended task valence with either placement, exposure, or both. In fact, the critical object showed highest affect ratings in the three exposure/negatively valenced attended task condition. These findings cast doubt on classical conditioning as the mechanism responsible for preattentive processing effects.

TABLE 2

MEAN AFFECT RATINGS OF CRITICAL OBJECT AND ANOVAS

The results also cast doubt on mood as an explanation for preattentive processing effects on affect. As noted earlier, if the valence of the attended task influenced subjects' moods, mood should transfer to evaluations of both seen and previously unseen objects. However, no effects involving valence of the attended task emerged. Furthermore, the mean score affect ratings of the three unseen objects (5.46) was not found to differ significantly from the neutral rating of 5 (p >.10). Additionally, the affect rating of the critical object in the positively valanced conditions (X=5.19) was not found to be significantly different from the affect ratings of the unseen objects in these same conditions (X=5.35,p>.10). However, the affect ratings of the critical object in the negatively valenced conditions (X=5.67) was found to be significantly different from the ratings of the unseen objects in these same conditions (X=5.06,p<.05).

To ascertain whether an implicit memory trace represents a sufficient condition for a change in affective response, we examined the impact of priming (e.g., implicit memory vs. no implicit memory) on evaluations of the critical object across the two exposure conditions. This interaction (see Figure 2) was significant, and indicates that an implicit memory trace for a stimulus is not sufficient to influence affective response. Specifically, the implicit memory trace translates into more positive affective responses only after several exposures.

FIGURE 2

ATTITUDE TOWARD THE OBJECT PRIMING VS NO PRIMING

DISCUSSION

This study observed several interesting effects regarding preattentive processing and affect. First, using implicit memory measures, the results indicated that some sort of preattentive processing did occur, and that those stimuli processed preattentively did influence affective responses in a manner predicted by mere exposure theory, but not classical conditioning or mood. However, the study did not find evidence for the hemispheric processing placement effects found in past literature. In fact, the placement effect was opposite that hypothesized. Two possible explanations could account for this reversed placement effect.

The first explanation is measurement based. Blaxton (1989) suggests that performance on an implicit memory test is facilitated by the degree to which the processing style used during study matches that used during test. The implicit memory test used in this study (dot completion) is very similar to a fragment completion task which is thought to utilize more of a data driven processing style than a conceptually driven processing style (Blaxton 1989). Therefore, performance on the dot completion task would be greater under those conditions favoring data driven versus conceptually driven processing during encoding. Because the left hemisphere is associated with a more analytic (data driven) processing mode and the right hemisphere is associated with a more holistic (conceptually driven) processing mode, the dot completion task was partial to the condition in which the stimulus was processed in the left hemisphere (i.e., when placed in the right visual field).

This explanation is weakened, however, when one considers the fact that there was an increase in evaluation of the stimulus only when it was placed in the right visual field and therefore processed in the left hemisphere. Unlike the dot completion task, an evaluation task is conceptually based and therefore one's performance on it would be facilitated if encoding utilized a more conceptually based processing style as well. Using the same logic as before, performance on the evaluation task should have been facilitated if the stimulus was placed in the left visual field and therefore processed in the right hemisphere. As noted above, however, just the opposite was found.

The second explanation for the reversed placement effect is stimulus based. It is very common to divide stimuli into verbal and nonverbal categories when studying hemispheric asymmetries (Sergent 1983). Unfortunately, this categorization may be too simplistic, as there are many other stimulus based characteristics that could lead to a left or right hemispheric advantage. For example Pitblado (1979) examined hemispheric asymmetries, studying object identification using random dot stereograms. Under certain conditions he was able to show that the left hemisphere was more efficient when large dots were used whereas the right hemisphere was more efficient when small dots were used. Similarly, Sergent (1982e) found a left hemisphere advantage in visual search for large letters and a right hemisphere advantage for visual search for small letters. Other hemispheric reversals from the typical verbal/visual paradigm have been found under conditions of differing exposure durations, luminance, eccentricity and degradation (Sergent 1983). In the same vein, it is quite possible that while the stimulus used in this study appeared to be visual in nature, due to other characteristics it was in fact partial to the left versus right hemisphere.

The study also indicated that the mere presence of an implicit memory trace is not sufficient for a change in affective response. Specifically, even though priming was no different between the positive/one exposure/right visual field condition and the negative/three exposure/right visual field condition, affect change occurred only for the latter condition. It may be that the strength, rather than the mere presence of the implicit memory trace determines whether there is a change in affective response. Repeated exposures (and perhaps other stimulus factors) may enrich or strengthen the implicit memory trace. Assessing the veridicality of this hypothesis, and identifying factors which may foster stronger implicit memory traces thus constitutes an important direction for future research.

The above findings and interpretations must, however, be considered in light of the study's limitations. A major limitation of the present study is the use of a simple geometric shape for use as the preattended stimulus. It is quite possible that a "real" pictorial ad (or a verbal ad) would lead to different results. It should be recalled, however, that the geometric figure was used in this study so that the implicit memory-affect relationship could be assessed. Additional study using stimuli more representative of those used in advertising should follow the present research. Furthermore, the study's conclusions should be held as tentative, pending further data collection to complete the experimental design.

REFERENCES

Blaxton, Teresa A. (1989), "Investigating Dissociations Among Memory Measures: Support for a Transfer-Appropriate Processing Framework", Journal of Experimental Psychology: Learning, Memory and Cognition, 15, 657-668.

Bouma, H. (1978), "Visual Search and Reading: Eye Movements and Functional Visual Field: A Tutorial Review," in Attention and Performance, vol. 7, ed. J. Requin.

Bowers, Jeffrey S. and Daniel Schacter (1990), "Implicit Memory and Test Awareness," Journal of Experimental Psychology: Learning, Memory and Cognition, 16, 404-416.

Dixon, Norman F. (1981), Preconscious Processing, New York: Wiley.

Fishbein, Martin and Icek Ajzen (1975), Beliefs, Attitudes, Intentions, and Behavior: An Introduction to Theory and Research, Reading, Mass.: Addison-Wesley.

Gorn, Gerald J. (1982), "The Effects of Music in Advertising on Choice Behavior: A Classical Conditioning Approach," Journal of Marketing, 46 (Winter), 94-101.

Goldberg, Marvin E. and Gerald J. Gorn (1987), "Happy and Sad TV Programs: How They Affect Reactions to Commercials," Journal of Consumer Research, 14 (December), 387-403.

Hastak, Manoj and Jerry C. Olson (1989), "Assessing the Role of Brand-Related Cognitive Responses as Mediators of Communication Effects on Cognitive Structure," Journal of Consumer Research, 15 (March), 444-456.

Janiszewski, Chris (1988), "Preconscious Processing Effects: The Independence of Attitude Formation and Conscious Thought," Journal of Consumer Research, 15 (September), 199-209.

Kihlstrom, John F. (1984), "Conscious, Subconscious, Unconscious: A Cognitive Perspective," in The Unconscious Reconsidered, eds. Kenith S. Bowers and Donald Meichenbaum, New York: Wiley.

Kihlstrom, John F. (1987), "The Cognitive Unconscious," Science, 237 (September), 1445-1452.

Lutz, Richard J. (1975), "Changing Brand Attitudes Through Modification of Cognitive Structure," Journal of Consumer Research, 1 (March), 49-59.

Lutz, Richard J. (1977), "An Experimental Investigation of Causal Relations Among Cognitions, Affect, and Behavioral Intention," Journal of Consumer Research, 3 (March), 197-208.

Musen, Gail and Anne Treisman (1990), "Implicit and Explicit Memory for Visual Patterns," Journal of Experimental Psychology: Learning, Memory and Cognition, 16, 127-137.

Moreland, Richard L. and Robert B. Zajonc (1977), "Is Stimulus Recognition a Necessary Condition for the Occurrence of Exposure Effects?," Journal of Personality and Social Psychology, 35 (April), 191-199.

Obermiller, Carl (1985), "Varieties of Mere Exposure: The Effects of Processing Style and Repetition on Affective Response," Journal of Consumer Research, 12 (June), 17-30.

Olson, Jerry C., Daniel R. Toy, and Philip A. Dover (1982), "Do Cognitive Responses Mediate the Effects of Advertising Content on Cognitive Structure," Journal of Consumer Research, 9 (December), 245-262.

Pitblado, C.B. (1979), "Cerebral Asymmetries in Random-Dot-Stereopsis: Reversal of Direction With Changes in Dot Size," Perception, 8, 683-690.

Schacter, Daniel L. (1987), "Implicit Memory: History and Current Status," Journal of Experimental Psychology: Learning, Memory and Cognition, 13, 501-518.

Sergent, Justine (1983), "Role of the Input in Visual Hemispheric Asymmetries," Psychological Bulletin, 93, 481-512.

Sergent, Justine (1982e), The Cerebral Balance of Power: Confrontation or Cooperation?," Journal of Experimental Psychology:Human Perception and Performance, 8, 253-272.

Stuart, Elnora W., Terence A. Shrimp and Randall W. Engle (1987), "Classical Conditioning of Consumer Attitudes: Four Experiments in an Advertising Context," Journal of Consumer Research, 14 (December), 334-349.

Wilson, William R. (1979), "Feeling More Than We Can Know: Exposure Effects Without Learning," Journal of Personality and Social Psychology, 37 (June), 811-821.

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