Você está na página 1de 47

THE IMPACT OF EMOTIONAL ADVERTISING APPEALS ON CONSUMER IMPLICIT AND EXPLICIT MEMORY: AN ACCESSIBILITY/DIAGNOSTICITY PERSPECTIVE

Patti Williams The Wharton School University of Pennsylvania

May 2000

Rough working draft. Please do not quote without authors permission.

Patti Williams is Assistant Professor of Marketing at the Wharton School, University of Pennsylvania, 1400 Steinberg/Dietrich Hall, Philadelphia, PA 19104 Thanks to Carol Scott, Jennifer Aaker, Bob Bjork, Gavan Fitzsimons, Debbie MacInnis and Don Morrison for their generous donations of time and support to my dissertation research, upon which this paper is based. Special thanks to Jennifer for her insight and editing skills in creating this version of the paper. This research was funded in part by the Procter and Gamble Marketing Innovation Fund. Particular thanks to Chris Allen for his efforts in facilitating my relationship with P & G.

Abstract Emotional advertisements have a substantial impact on consumer attitudes, as well as upon purchase intentions. However, research on the influence of emotional appeals on memory has been somewhat mixed, with some researchers asserting that they result in poor consumer memory, while others argue that if tested properly, they have a substantial impact. The current research addresses these mixed results by relying on an accessibility/diagnosticity framework to explore the effect of emotions on consumer implicit and explicit memory. Explicit memory performance is characterized as relying upon both the accessibility of memory traces and their relative diagnosticity in contrast with other inputs. In contrast, implicit memory performance is primarily driven by accessibility alone. Results from two experiments are supportive of this perspective. The first experiment demonstrates that overall emotional advertising appeals have a bigger impact on implicit versus explicit memory performance, though explicit memory performance is enhanced after exposure to an intense emotional appeal. The second experiment demonstrates that the diagnosticity of emotional appeals can be enhanced, and that such enhancement leads to better explicit memory performance under conditions of high involvement. In contrast, emotional diagnosticity is shown to have no effect on consumer implicit memory.

INTRODUCTION Emotional or feeling advertising appeals have received considerable attention over the past decade in consumer behavior research. Past research has focused on the specific types of emotional appeals frequently used (Stayman, Aaker and Bruzzone 1989), as well as the effects of emotional appeals on consumer advertisement and brand attitudes, as well as purchase intentions (e.g., Burke and Edell 1987, Batra and Ray 1986). However, research examining the impact of emotional advertising appeals on more cognitivebased effects such as consumer memory has been somewhat mixed; some researchers have shown that the use of emotion of persuasion appeals results in poor consumer memory (e.g., Zielske 1982), while others argue that if tested properly, emotions may have a substantial effect on memory (e.g., Friestad and Thorson 1993). As a result, the impact of emotions on memory in persuasion contexts is not clearly understood. The objective of this research is to address these mixed results by relying on an accessibilitydiagnosticity model (Feldman and Lynch 1987) to explore the impact of emotions on both implicit and explicit memory. Thus far, research has focused upon measuring explicit memory for emotional advertisements, ignoring their potential impact upon consumers implicit memory (Schacter 1987). However, while emotional traces stored in memory may be accessible, they may not be particularly diagnostic when consumers engage in the effortful, strategic searches of memory necessary for explicit memory performance, and thus may be outshone by other cues (Smith 1988). In contrast, however, these traces may have a substantial impact upon implicit memory, which requires no assessments of diagnosticity, instead relying entirely upon accessibility. Moreover, researchers have argued that emotional experiences might often be more implicit or unconscious in nature (Kihlstrom 1993), which suggests, in accordance with the encoding specificity principle, that investigations of the effects of such ads on implicit memory may shed light on previously conflicting results by tapping into the unconscious effects of emotions on memory. This paper reports the outcome of two experiments designed to investigate the potential impact of emotional advertising appeals upon consumer implicit versus explicit memory. The first experiment explores the impact of advertising appeals of varying degrees of emotional intensity upon both implicit and explicit memory. Experiment 2 extends these results, investigating the degree to which the diagnosticity of intense

emotional experiences can be heightened, via relevance of an emotional appeal to the advertised product, thus improving consumer explicit memory for emotional persuasion appeals, particularly under conditions of high involvement. ADVERTISING APPEALS AND THE USE OF EMOTIONi As a result of calls to give greater attention to emotional and experiential aspects of consumer behavior (Zajonc 1980, Holbrook and Hirschman 1982), the importance of affective responses to advertisements, and their impact upon both attitudes and choice processes has become increasingly clear. A large number of advertisements can be characterized by their emotional aspects (Stayman, Aaker and Bruzzone 1989), and emotional responses are central to consumers perceptions of and reactions to advertisements (Aaker and Bruzzone 1981). Much of this research has focused on psychological theories of affective experience to determine the types of emotions that may be evoked in persuasion appeals. For example, one common finding in both the basic emotion literature which focuses on emotional responses to general stimuli (e.g., Izard 1977; Mehrabian and Russell 1974; Plutchik 1980) and the emotion literature in consumer behavior which focuses on emotional responses to advertising appeals Batra and Ray (1986; Burke and Edell 1989; Edell and Burke 1987; Mehrabian and Russell 1974), is that three general types of emotional responses exist. Two of the emotional responses appear to be positive (one typified by more arousing emotions such as surprise, elation and joy, and the other by more soothing emotions such as warmth, hope and gentleness), while the third is negative. Based on these specific emotional responses, researchers have focused on the types of consequences that they yield. Batra and Ray (1986), for example, expanded the traditional coding of thought protocols to include not only cognitive responses to appeals (Wright 1973), but emotional responses. Importantly, these emotional responses accounted for significant levels of variance in advertisement attitudes, over and above that provided by the traditional cognitive responses. Further, such emotional responses can also directly impact brand attitudes and purchase intentions (e.g., Stayman and Aaker 1988, Edell and Burke 1987, Burke and Edell 1989).

Importantly, in this literature stream, a distinction has been made between emotions depicted in the advertisement and those actually felt by the consumer viewing the advertisement (Aaker and Stayman 1989). While depicted and felt emotions often coincide, there are also conditions in which they are discrete, such as in the case of upbeat emotional responses (Stout, Homer and Liu 1990). In contrast, depicted and felt emotions tend to be highly related in the case of warm emotional responses (e.g., depicted relaxed emotions are highly correlated with felt relaxed emotions; Burke and Edell 1987). This variation in the level of correspondence between the constructs may have to do with differences in arousal. Relative to warm emotional responses, upbeat emotional responses tend to be more highly arousing. In such conditions, the cognitive appraisal processes may be more likely to be distinct from the actual emotional experience (Lazarus 1982) . Conversely, in conditions where the emotional experience is characterized by low levels of arousal, the cognitive appraisal processes may be quite similar to the actual experience. In this research, we focus on conditions where depicted and felt emotions highly correspond, but the construct of most interest is felt or experienced emotions. The Impact of Emotional Advertising Appeals on Consumer Memory While previous work has demonstrated the importance of emotional responses upon advertising effectiveness as conceptualized by attitudes and purchase intentions, advertising practitioners are often interested in other measures of advertising effectiveness, such as recall (Lynch and Srull 1982, Krishnan and Chakravarti 1993). In this domain, the impact of feeling advertisements is much less clear. A number of researchers have found that emotional advertisements often do not perform well on measures such as day-after recall (Zielske 1982), suggesting that emotional responses evoked by advertisements are poor retrieval cues compared to cognitive responses, and resulting in widespread practitioner belief that emotional commercials do poorly in standard memory tests (Berger 1981). In contrast, others have shown that emotional advertisements do have a recall advantage, at least under retrieval conditions which encourage search of episodic memory (Friestad and Thorson 1986, 1993, Thorson and Friestad 1989, Thorson and Page 1988). Asserting that advertisements are encoded into episodic memory (i.e., the mental storage of personal experiences and their

spatial and temporal context), with a trace that can be strengthened via the experience of emotional arousal, they show that the typical semantic retrieval cues (i.e. product category cues) contained in the customary cued recall measures are generally inappropriate, and lead to poor recall performance for emotional appeals. In contrast, free recall, an episodic memory task, as well as the use of executional or experiential cues (i.e., Recall the ad that featured the grandfather playing with his grandson.) appear to lead to faster, higher recall for emotional advertisements (Friestad and Thorson 1986, 1993). However, even in this body of work, the impact of emotional advertising appeals on consumer memory has been somewhat unreliable (Page, Thorson and Heide 1990). The apparently inconsistent impact of emotional advertisements upon consumer memory implies that the role of feelings in memory may not be appropriately conceptualized. Nearly all of the previous work on this topic has conceptualized emotional responses stored in memory as available to the conscious or intentional retrieval processes tapped by traditional, explicit measures of memory. However, examination of a wide variety of studies investigating emotional memory suggests that while emotional responses are encoded into memory, and thus potentially accessible for explicit retrieval, they may not always be considered diagnostic in such explicit searches (Feldman and Lynch 1988). In addition, it may be that feelings evoked in response to advertisements are much less conscious, and thus more implicit in nature (Zajonc 1980, Kihlstrom 1993). If true, this would suggest that one key to disentangling these inconsistent results may lie in determining the relative impact of emotional appeals on implicit memory relative to explicit memory. IMPLICIT MEMORY AND EMOTION Comparing Implicit and Explicit Memory Information processing models of human memory have traditionally relied upon memory models and tasks which presuppose that individuals have conscious access to the contents of a long-term memory store (Lynch and Srull 1982). Thus free recall, cued recall and recognition have been the predominant methods used to assess memory performance. Each of these tasks makes direct reference to, and indeed requires, conscious recollection of a specific learning episode during a particular time period. For example, to answer these types of questions about advertisements previously seen, a respondent would be required to

consciously think back over those appeals seen in the target time period and develop an intentional, strategic search strategy to retrieve the memory traces associated with the relevant appeals. Over the past decade, however, a large literature has grown which demonstrates that much information stored in memory is not available to such conscious retrieval, but is instead more implicit in nature, and is thus accessed unconsciously or automatically rather than consciously or strategically (Schacter 1987). This finding has led to a distinction between the traditional conceptualizations of memory, classified as explicit memory, and the newer conceptualizations, implicit memory. Explicit memory refers to the conscious awareness of material and an intention to remember it, and is typically measured by free or cued recall and recognition tasks. In contrast, implicit memory refers to the effects of a previous learning episode that are expressed without awareness or intention to remember, and is typically measured by tasks such as stem or word fragment completion and category associate generation (Bjork and Richardson-Klavehn 1988, Schacter 1987), as well as via preference judgments (e.g., Kunst-Wilson and Zajonc 1980).ii One point merits noting, however; the relative infancy of this research domain has led to a proliferation of terms to describe these memory processes (e.g, unaware-aware, unconscious-conscious, intuitive-analytic, direct-indirect, procedural-declarative, automatic-controlled), as well as a difficulty in separating the conceptual meaning of the constructs from their measurement (for a review, see Bjork and Richardson-Klavehn 1988). In the present research, the terms, implicit and explicit, are used for the memory processes, while the measures for each will be referred to as direct and indirect, respectively. Specifically, direct tests are those in which the instructions at the time of test make reference to a target event in the personal history of the subject, while indirect tests refer only to the task at hand, making no reference to prior events.iii Implicit memory effects were first observed in experiments with amnesics, who often could not remember specific events but who nonetheless exhibited behavioral or other changes consistent with the outcomes of these specific encounters, such as the ability to perform certain new skills despite no recollection of the episodes during which these skills were learned (Schacter 1987). Identification of this phenomenon led to studies with normal participants, primarily focusing on repetition priming in various perceptual tasks. These experiments demonstrate that exposure to a target can impact subsequent tasks

involving those targets, such as word fragment or word stem completion (e.g., Graf, Mandler and Haden 1982), word identification (e.g., Jacoby and Dallas 1981) and lexical decisions (e.g., Scarborough, Gerard and Cortese 1979). The presentation of the target stimulus is said to prime the perceptual representation of that stimulus in memory, regardless of the participants ability to recall the previous presentation. Importantly, manipulations long known to impact explicit recall, such as elaborative processing, have no effect upon implicit memory at the perceptual level (Jacoby and Dallas 1981). In fact, the traditional levelsof-processing hierarchy of effects (Craik and Lockhart 1972) is reversed for perceptual implicit memory, with low-levels of processing resulting in greater implicit perceptual memory benefits, while high-levels of elaborative processing result in little or no benefits. Other dissociations between implicit and explicit memory have been identified as well, including the attenuation of priming effects (Jacoby and Dallas 1981), minimal impact of interference in priming (Graf and Schacter 1987), and stochastic independence between performance on implicit and explicit tests (Eich 1984). Further, researchers have expanded the domain of implicit memory phenomena to include conceptual-level effects. For example, while perceptually-based processing facilitates word-fragment completion, elaborative processing increases the degree to which correct answers are given in response to general knowledge questions (Roediger, Srinivas and Weldon 1989). Similarly, elaborative processing of word targets increase the accessibility of category words associated with that trait (Smith and Branscombe 1989). Thus, implicit memory appears to exist for both perceptual and for conceptual information, where the latter is typically measured by the priming of general knowledge and /or category membership. Recently, consideration of implicit memory has had broader impact outside the cognitive psychology realm. For example, the role of implicit memory in social psychological phenomena has been examined by focusing on the implicit use of stereotypes (Banaji and Greenwald 1994), the biased interpretation of ambiguous trait information (Bargh and Pietromonaco 1982), and the impact of mood on evaluations (Schwarz and Clore 1983). Importantly, this research stream has also expanded the repertoire of indirect measures to include response latencies, with the implication that the use of implicit, unintentional memory should be faster than more strategic, explicit use of memory (Fazio et al, 1986, Bargh et al, 1992).

In the consumer behavior domain, implicit memory effects have also been gaining attention. For example, Nedungadi (1990) used both fragment completion and category listing tasks to assess brand priming effects. Krishnan and Chakravarti (1993) highlighted the potential importance of implicit memory on brand equity. Others have examined the impact of mere exposure to advertising stimuli to enhance brand evaluations without recall of the original presentation (Lee 1995), and have replicated the effects of processing type, word-frequency, name awareness and repetition and levels of processing found in the cognitive psychology literature for advertising based brand recall (Lee 1995, Krishnan and Shapiro 1996). However, while this stream has made early progress in understanding the effects of implicit memory in the cognitive realm of consumer behavior, very little is known about its potential effects on the processing of emotions and the more general effects of emotional appeals. In this research, we attempt to address this gap in an attempt to provide theoretical progress toward understanding the impact of emotions on memory. The premise put forth is that the mixed memory results found in the emotion and persuasion literature may lie in diagnosticity of the emotional cues in aiding memory. The Accessibility and Diagnosticity Framework and its the Implications for Implicit Memory The accessibility/diagnosticity framework (Feldman and Lynch 1988; Lynch, Marmorstein and Weigold 1988) hypothesizes that information contained in memory is only used to the extent that it is relatively more accessible in memory than is other information, and is perceived as more diagnostic than information that is equally accessible. Accessibility is defined as the degree to which information can be retrieved from memory, while diagnosticity refers to the perception that a single piece of information available in memory is adequate to perform a task. In tests of cued recall or recognition, highly diagnostic cues (e.g., product category membership cues) are given to respondents as part of the memory test itself (e.g., Do you recall any advertisements for toothpaste?). In such cases, it is unlikely that participants intentionally search for or use the relatively weaker emotional cues contained in long-term memory (Smith 1988, Bower 1981). In contrast, in free recall, participants must create their own intentional search strategies, making the use of emotional cues more likely. However, even with free recall as the measure, mixed results often occur, suggesting that even in this type of task, participants very often find their way to other, self-

generated cues that may be more diagnostic than those presented by emotions (Tobias, Kihlstrom and Schacter 1992). Examining the impact of more rational, cognitive advertisements on consumer memory adds support for this perspective. The use of explicit memory is an intentional, strategic process, facilitated by elaborative processing at the time of encoding. A variety of experiments have found that while cognitive appeals result in higher recall under conditions of high elaboration, emotional appeals, which tend to be characterized as peripheral in nature and low in diagnosticity compared with rational appeals, do not (Cacioppo and Petty 1982). As use of explicit memory is in itself involving, it is not surprising that those elements of a message deemed most diagnostic under conditions of high involvement also lead to higher levels of explicit memory. Indeed, principles of encoding specificity and transfer-appropriate-processing (Roediger 1990) suggest that this overlap between processing of central cues under conditions of high involvement and higher explicit memory for advertisements containing those cues is natural. Similarly, the lack of diagnosticity of the w eaker emotional cues is not surprising. As they tend to not be processed as elaborately, they are unlikely to serve as diagnostic cues under the conditions which most favor explicit memory performance. This might be interpreted to imply that emotional advertisements would result in enhanced explicit memory under conditions of low involvement. However, even then, emotional cues may be relatively weak compared to product category or other cues also available, as even peripheral or heuristic cues vary in perceived reliability (Eagly and Chaiken 1993). Emotional cues are, however, likely to be accessible in memory if emotional responses to an advertisement or other stimulus have occurred. An over-reliance on explicit memory effects may thus lead to under-estimation of the potential impact of emotional appeals on consumer memory. The examination of the impact of emotional appeals on implicit memory may yield greater insights, as importantly, the use of implicit memory does not call for assessments of diagnosticity. Such assessments necessarily implicate intentionality and a strategic search of and use of memory, and are thus highly relevant to tests of explicit memory. The use of implicit memory, however, relies solely upon relative accessibility of existing knowledge structures in memory. If a knowledge structure is primed, reflecting implicit memory for a previous exposure, that heightened accessibility will reveal itself on indirect tests (Graf and Mandler 1984).

Emotional information encoded into mem ory may heighten the accessibility of relevant emotional nodes, and via spreading of activation, other information linked to it. Thus indirect tests of memory may reveal effects of emotional experience, whereas with direct tests may be overridden by considerations of diagnosticity, reducing the impact of emotional information. Consistent with this perspective, the impact of feelings on judgment is believed to decrease as the amount or salience of competing information increases (Clore, Schwarz and Conway 1994). Similarly, affect impacts evaluations of unfamiliar brands, when no other relevant information is available, but not that of familiar brands, when presumably consumers have other relevant information to consider (Srull 1983). Likewise, Ellis (1985) has found that mood effects on memory are most likely to occur when processing is impoverished or incidental, that is, when the availability of other cues potentially more diagnostic is low. Further, Friestad and Thorson (1993) have shown that consumers are most likely to show effects of emotional advertisements on memory under conditions of free recall than when product category cues are used to facilitate memory. Again, this product category information is likely to outshine the weaker emotional information encoded with the advertisement during the original presentation. However, executionally-cued recall strategies do enhance memory for these emotional advertisements, suggesting that these types of questions make emotional cues more diagnostic than do the product category cued recall questions. The Implicit Nature of Emotional Experience Besides issues of diagnosticity, a number of researchers have asserted that emotional responses may be relatively implicit in nature, further supporting the importance of investigating their potential memory effects within the domain of implicit rather than explicit memory. Emotional reactions are often assumed to reflect an underlying implicit appraisal process by which events are evaluated (Clore, Schwarz and Conway 1994, for a review). Thus, emotions may be an intrinsic, yet relatively unnoticed feature of an environment or episode. The view in cognitive psychology is often that emotions are part of the environmental context, and thus only contiguous to conscious experience rather than an integral part of it. Nonetheless they are incorporated, along with other aspects of the situation, into a unitary memory representation (cf. Macaulay, Ryan and Eich 1993). Thus, many emotional reactions may not be consciously mediated, but may occur

spontaneously and implicitly in reaction to some stimulus. If emotional reactions are indeed more implicit in nature, both the encoding specificity principle and theories of transfer appropriate processing would suggest that their effects on memory should be conceptualized implicitly as well (Roediger 1980). Zajonc (1980) has made persuasive arguments that much affect requires no cognition, prior to the consciousness. For example, research on mere exposure has directly linked implicit memory to affect, though the focus has been on the affective outcomes of implicit memory effects, rather than the implicit use of emotional information encoded into memory. This work has demonstrated that both liminal and subliminal exposure to a target stimulus can result in more positive evaluations of that stimulus, regardless of participants awareness of the stimulus, demonstrating that implicit memory can be a more sensitive gauge of past experience than explicit memory (Kunst-Wilson and Zajonc 1980). Kihlstrom (1993) furthers this argument, suggesting that both emotions and cognitions often occur unconsciously, and that neither are likely to receive the activation necessary to bring them to consciousness unless they are somehow connected with the self, and thus worthy of additional allocation of mental resources. While emotional associations between a brand and its advertising may be accessible under conditions of explicit memory, they are likely to be relatively weak cues, compared to other experimenter-generated cues in tests of cued recall or recognition, and perhaps even to other subject-generated cues under tests of free recall. As a result of this impaired diagnosticity, examination of implicit memory may result in a more accurate representation of the impact that emotional advertisements can have on long-term memory. Moreover, emotional experience may often be relatively implicit in nature, occurring prior to conscious awareness and cognitive activity, making the application of tests of implicit memory even more relevant to the investigation of these effects. H1: Emotional advertising appeals will lead to enhanced consumer implicit (versus explicit) memory. Clearly, however, consciously experienced emotional reactions are frequent, and evidence for emotional impact on memory has been found. Thus, it is important to consider the factors that account for the differences among emotional experiences. For example, intense affective experiences are more likely to be recalled explicitly than are mild ones, even after a 24 hour delay (Hardin and Banaji 1990). Intense affect

presumably heightens an individuals arousal, increasing attention to and elaboration of those experiences. Thus, arousal at encoding produces better memory for related information after delay, while match in arousal at encoding and retrieval produces better memory for affectively neutral information (Clark et al, 1983). These results support the hypothesis that increased intensity of emotional experience at the time of encoding will promote the arousal conditions that contribute to conscious elaboration and thereby lead to explicit memory benefits. However, emotional experiences are likely to enter consciousness under conditions of adequate intensity. This intensity is likely to draw attention to the emotional experience, thereby prompting greater elaboration upon the experience and encouraging the creation of a memory trace that will be stronger and thus perhaps more diagnostic under conditions of explicit memory use. Thus, very intense emotional appeals will have a greater impact upon explicit memory, while mild emotional appeals will not. In addition, intensity of emotion should enhance the accessibility of the emotional trace, thus benefiting implicit memory as well. However, it is expected that increased intensity will have a greater impact upon explicit memory than upon implicit memory, as the former will benefit not only from an increase in accessibility per se but from an increase in elaboration as well. H2: Intense (versus mild) emotional advertisements will enhance both consumer explicit and implicit memory, with a greater effect upon explicit than upon implicit memory. While the presentation of emotional advertisements is expected to enhance the accessibility of the target brand name featured in the advertisements, it may at the same time lessen the relative accessibility of other brands in the same category. For example, Alba and Chattopadhyay (1986) found that increasing the salience (accessibility) of one brand in memory reduced the accessibility of alternative brands, including those that would otherwise be candidates for potential purchase. Thus, as the accessibility of the target brand is increased via exposure to emotional advertisements, the accessibility of other major brands in the category is expected to decrease. H3: As the accessibility of one brand name in memory increases, the accessibility of other brands in the category will decrease.

EXPERIMENT 1 Overview Experiment 1 is a 2 (Emotional Strength: Intense versus Mild) by 2 (Memory Task: Indirect versus Direct) full-factorial between subjects design. The indirect measure of memory used is a conceptual priming task while the direct measure is a cued recall question. As the various memory measures are sensitive to prior measurement (Schacter, Bowers and Booker 1989), this between subjects collection is necessary. The primary dependent variable is individual participants successful memory performance on the memory task they are asked to complete. In addition, brand and advertisement attitude measures and thought and feeling protocols are also taken. Method Undergraduate students (N = 149; 85 female) from an East Coast University required to complete market research experiments for class credit participated in this experiment. They were first asked to complete demographic questions and then to review a mock magazine entitled Grad Life, ostensibly aimed at graduate students. Participants were told this was a magazine developed by a group of MBA entrepreneurial students considering launching it as a publication in the near future. Before making that launch decision, however, the entrepreneurs wanted feedback from current and future graduate students about its content and general interest in such a publication. Subjects were asked to review all aspects of the magazine including the content, advertisements, layout and design and to provide their opinions of it in a questionnaire that would follow. The mock magazine consisted of ten pages: A cover page and table of contents, three pages of editorial copy (on the subjects of health, weekend getaways and web site reviews), and five pages of advertising. The target advertisement was repeated twiceiv, always on the fifth and ninth pages. In addition to the target advertisement, the magazine also contained a full-page appeal for American Express Student Card services, a half-page ad for MCI Long Distance services and a one-half page ad for Hewlett Packard computers. To avoid drawing undue attention to the target advertisement, the ads for American Express was also repeated twice. The entire magazine was printed in full color and was encased in plastic sheet covers.

Immediately after reviewing the magazine, participants completed a short questionnaire consistent with the cover story including several questions about the articles that appeared in the magazine, as well as potential additional topics ostensibly under consideration for future issues. Subjects then completed a 15minute unrelated filler task. Immediately following that task was a single sheet of paper which included either the implicit or explicit memory task (described below) and probed for suspicion regarding the connection between that task and the previously viewed magazine. Unless specifically mentioned in the memory task, this page included no references to the Grad Life magazine study. After completing another 10-minute unrelated filler task, subjects were given a final survey that asked additional questions consistent with the experimental cover story. In addition, it assessed target brand and advertisement attitudes, collected openended thought and feeling protocols and included a manipulation check on emotional reactions to the target appeal. Independent Variables Emotional Strength. Warmth, defined as a positive, mild, volatile emotion, involving physiological arousal and precipitated by experiencing directly or vicariously a love, family or friendship relationship (Aaker, Stayman and Hagerty 1986), was chosen as the emotion of focus for several reasons. First, in the emotion literature, dimensions resembling warmth are often predominant (e.g., Batra and Ray 1986, Burke and Edell 1987). Moreover, warm advertising appeals are common, accounting in some studies for nearly 20% of all appeals (Stayman, Aaker and Bruzzone 1989) and warm emotional responses have been found to be commonly evoked among advertising viewers (Burke and Edell 1987, Batra and Ray 1986). In addition, felt and depicted warm emotional responses tend to be highly correlated, thereby reducing this difference as a potential confound (Stout, Homer and Liu 1990). To reduce potential confounds and noise, print advertisements were used. As a result the primary manipulation of warmth in the advertisements was color photographs. Examination of the warmth literature, as well as a pretest with 50 participants provided guidance in stimuli creation. Warmth was executionally created taking a photograph of a young couple seated together in a warm, sun-washed location. Intensity of the evoked emotion was manipulated through three factors

(Vanden Abeele and Machlachlan 1994): (1) eye contact between the two models (looking at each other versus looking at the camera), (2) their physical closeness (close together versus farther apart), and (3) color tones in the photos (orange and yellow tones versus blue and green tones). A second pretest (N = 18, 9 female) was then conducted to identify one relatively intense and one mildly warm photograph from the set of eight photos. Participants were asked to indicate the degree to which they experienced a particular emotion while viewing each photograph (where 1 = not at all; 7 = very intensely). Drawn from Burke and Edell (1987), 52 emotions were measured (12 warm emotions, Cronbachs alphawarm = .84; 26 upbeat emotions Cronbachs alphaupbeat = .97; 14 negative emotions, Cronbachs alphanegative = .87). The results indicated our photographs evoked significantly more warm than upbeat and negative felt emotions (ps < .005). The two photos chosen were one that evoked the most intense feelings of warmth (m = 5.76) and another that evoked mild warm responses (m = 3.27, p < .001). Next, the advertisements were created. Film was chosen because it is a familiar product category, and one that makes the use of the photographic manipulation relevant. Further, it is classified as neither a strongly feeling product nor a strongly thinking product (m = 2.93 out of 7), and as moderately involving (m = 4.95 out of 7; Ratchford 1987). Based on pretest results v, advertising copy was created, focusing on weak attributes possessed by the film. In this way, some copy is provided for participants to read and later answer questions about, but the copy maintains the status of the ad as one that is primarily emotional with relatively little rational content. The film advertisement, which was for an unfamiliar yet real brand, Agfa film,vi featured a headline, the photograph, copy, a tagline and a logo. Memory Measures. Two memory measures, one indirect (conceptual priming) and one direct (cued recall) were collected between subjects. Conceptual priming was assessed via generation of category membership, where the measure of interest is inclusion of Agfa in the list of film category members (As quickly as possible, please list all the brands of film you can think of.). Inclusion of the dominant brands, Kodak and Fuji, was also measured, consistent with Hypothesis 3. Explicit memory was assessed via a direct memory measure in the form of a cued recall question which asked participants to recall the brand of film they saw advertised in the magazine (The magazine you viewed earlier, Grad Life, featured an ad for a brand

of film. Please write the name of that brand of film in the space below.). In addition to brand name memory, all participants were asked to recall as much as they could from the target advertisement in an openended question. Finally, it should be noted that, in addition to the primary data collection, a second group of participants (N = 35) was asked to complete the conceptual priming measure, without exposure to the target advertisement. This group thus served as the baseline standard against which those exposed to the target are compared. Nine percent (3 subjects) of this control group included Agfa on their list of film brands. Dependent Variables Successful Memory Performance. The primary dependent variable in this study is successful memory performance. Provision of the target brand name, Agfa, in response to the memory question asked in each condition will be counted as a success. Brand and Advertisement Attitudes. Participants were asked to complete a series of measures to assess their attitudes toward both the target advertisement and the target brand. Four questions measured both brand and advertisement attitudes (where 1 = bad, not at all likable, negative, unfavorable; 7 = good, likable, positive, favorable). An average of these responses led to a four-item advertisement attitude index (Cronbachs alpha = .95) and a four-item brand attitude index (Cronbachs alpha = .96). Cognitive and Emotional Responses. Subjects were asked to describe any thoughts or feelings they had during exposure to the target ads. The order of the thought and feeling protocol questions was counterbalanced between participants. Two independent raters categorized cognitive and emotional responses to the target advertisements. Both coders were trained in cognitive and emotional response analysis and were given a series of examples of each thought and feeling type. Both raters coded a common set of examples similarly and were encouraged to ask any questions of clarification during subsequent coding (Brislin 1980). Adopted from Wright (1973), cognitive responses were coded in terms of positive (i.e, The ad was sweet.), negative (i.e., It was very uninviting.) and neutral thoughts (i.e., The ad was in color.) about the advertisement and about the brand (i.e., positive: I thought I should buy that film next time; negative:

Agfa is a stupid name for a brand; neutral: Ive never heard of that brand of film.). Irrelevant thoughts (thoughts that were theoretically meaningless, i.e., Film developing costs too much.) were also coded. Drawing on Batra and Ray (1986) and Burke and Edell (1987), emotional responses were coded into three categories: Warm emotional responses (as defined above, e.g.,, The ad was sweet and made me feel emotional.), upbeat emotional responses (Feelings of excitement, energy or active joy, e.g., It gave me a strong feeling of happiness.) and negative emotional responses (Feelings of unhappiness, sadness, anger, or boredom, e.g.,, The picture of the loving couple made me feel lonely.). Other unrelated emotional responses were also coded (e.g., The ad made me feel confused:). Inter-rater agreement was high (90%). When disagreements occurred, the raters were instructed to discuss their differences until a consensus was reached, a process which required less than a minute of discussion. Results Typical memory experiments in the cognitive psychology literature test memory for a battery of items, and thus the dependent variable of interest is percentage of successful memory performance on the total list of potential items. In our experiments, however, memory is tested for a single brand name, and the dependent measure is thus successful or unsuccessful memory performance for the single item on each type of memory task. This dependent variable is thus a dichotomous variable, and as a result, the standard ANOVA analysis is inappropriate. Instead, a more general linear model provides the basis for the analysis, which permits the flexibility to incorporate both binary and continuous dependent variables. This type of analysis allows for a clear distinction between independent and dependent variables and is a natural extension of the usual ANOVA approach for continuous data, thus allowing for the easy exploration of both main and interaction effects. vii All hypotheses were tested in a 2 (Emotional Intensity: Mild vs. Strong) by 2 (Memory Task: Indirect vs. Direct) between subjects analysis. Manipulation Checks. A manipulation check on the emotions experienced in response to the target advertisements indicates that both the mild and emotional advertising appeals evoked w arm emotions as intended. Participants completed a ten-item emotional response scale based upon Burke and Edell (1987) that asked them to indicate the degree to which they experienced five upbeat (joyfulness, energetic, inspired,

excited, happy; Cronbachs alpha = .91) and five warm (warm, sentimental, moved, emotional, hopeful; Cronbachs alpha = .93) emotions in response to the target advertisement (where 1 = did not experience emotion at all, 7 = experienced the emotion very strongly). The average upbeat response was subtracted from the average warm response to assess the degree to which the two appeals evoked more warm than upbeat emotions. Both appeals evoked significantly more warm than upbeat feelings (ps < .05). Further, a one-way ANOVA (df = 1, 148) on the impact of intensity of emotional appeal indicates that the intense versus the mild emotional appeal evoked significantly greater feelings of warmth (m = 5.44 vs. 3.73; F = 14.82, p < .01), as intended. As noted previously, 9% of a control group not exposed to the target appeals included the target brand, Agfa, in response to the conceptual priming indirect memory task. In this experiment, 24% of subjects who received this task included Agfa in response to this question. Thus a significant amount of priming occurred over the base rate (Chi-Square = 132.57, p < .01), indicating significant implicit memory benefits resulting from the exposure to the target emotional appeal. Subjects in both memory conditions were asked to indicate what they thought the purpose of the memory question they completed might be. Consistent with Schacter, Bowers and Booker (1989), this question served as a check on the degree to which those subjects in the indirect memory task condition were aware of the link between the memory questions and their previous exposure to the target advertisement, and thus to what extent they may have relied upon conscious recollection of that appeal when completing the task. Five subjects in the indirect task condition correctly related the task to the advertisement they had previously seen and were eliminated from the data set.viii As the cued recall question directly referred back to the target appeals, it is not surprising that every subject in the direct task condition made the connection between the two tasks. Memory Performance. Overall, performance on the indirect memory task (m = .64) outstripped that on the direct memory task (m = .57), supporting Hypothesis 1 (Chi-square = 2.64, p < .05). Analysis also indicates a significant main effect of emotional intensity (mintense = .73, mmild = .50; Chi-square = 10.21, p < .01) on memory performance. This is qualified, in partial support of Hypothesis 2, by a significant interaction

(Chi-square = 9.68, p < .01) between emotional intensity and memory task such that while intensity of emotional experience significantly benefited performance on the direct memory task (mmild = .34, mintense = .79), it had no significant effect on the indirect memory task (mmild = .60, mintense = .67; see Figure 1). These results indicate that intensity of emotional experience had a significant effect on explicit memory, as presumably it is brought into consciousness and thus is more available for subsequent intentional retrieval. Intensity does not appear to have a significant effect upon implicit memory, however, though the means on the indirect memory task are directionally consistent with a greater accessibility of the more intense emotional experience in memory. ---------------------------------------Insert Figure 1 about here. ---------------------------------------Finally, Hypothesis 3 examined the extent to which priming the target brand name might inhibit implicit memory performance for the dominant brands in the category. To measure this effect, successful memory performance in the implicit memory conditions for both Kodak and Fuji, the major film brands in the U.S. was coded. Analysis indicates that with greater emotional strength of the target appeal, Kodak and Fuji became less accessible for participants, supporting Hypothesis 3. Significant main effects of emotional strength were found in analyses of the memory performance for both brands (Kodak Chi-square = 9.22, p < .01; Fuji Chi-square = 10.01, p < .01). While all participants recalled Kodak after viewing the mild emotional appeal, just 74% listed Kodak after exposure to the intense appeal. Similarly, 96% included Fuji after the mild appeal, while 63% did so after the intense appeal. Attitudes. Participant attitudes toward the advertisement and the brand were analyzed via a one-way ANOVA (Emotional Strength: Intense versus Mild; df = 1, 148). Participants expressed more favorable attitudes toward the intense emotional appeal (m = 4.48) compared to the mild emotional appeal (m = 3.84, F = 5.86, p< .02). As shown in Table 1, there were no differences in attitudes toward the brand (F < 1).

---------------------------------------Insert Table 1 about here. ---------------------------------------Cognitive and Emotional Responses. Advertisement effects on subjects thoughts and feelings were analyzed via a one-way ANOVA (Emotional Strength: Intense versus Mild; df = 1, 148). No significant effects were found on brand thoughts (Fs < 1) or on total thoughts (F < 1) generated. Consistent with the attitude results reported above, however, participants expressed significantly more negative thoughts toward the mild (m = .49) than the intense (m = .17) emotional appeal (F = 5.45, p < .05). There were no differences in positive or neutral thoughts about the advertisement (Fs < 1). While there were no significant differences in the total number of feelings experienced in response to the two appeals (mmild = .70, mintense = .96, F = 2.21, p < .14), participants did experience significantly more warm feelings in response to the intense emotional appeal (m = .70) compared to the mild appeal (m = .38, F = 4.16, p < .05). No differences were found for upbeat or negative emotional responses to the advertisements. Discussion Results of Experiment 1 indicate that emotional advertising appeals in general maybe most likely to impact implicit memory to a greater degree than explicit memory. Overall implicit memory, measured at the conceptual-level, for the target brand was better than explicit memory performance after exposure to the emotional advertising appeal. However, the degree to which emotional appeals do impact explicit memory can be enhanced through increased emotional intensity. Intensity of emotion, however, appears not to impact implicit memory, contrary to expectations. This is somewhat surprising as intensity of arousal was expected to heighten the degree of accessibility of the emotion in memory, thereby impacting implicit as well as explicit memory. The results also suggest that significant indirect memory benefits can accrue to a brand featured in an emotional appeal. When participants were exposed to the intense emotional appeal for Agfa, they were subsequently less likely to mention the two dominant brands in the film category in response to the conceptual priming, category membership listing-task. Such a decrement in performance could be highly

beneficial for a smaller brand in a product category, even when memory performance for its own brand name does not improve. In actual choice situations, if subjects are less likely to generate the names of the category leaders, they may be more likely to turn to the lesser-known members of the category. Overall, the results of this study strongly suggest that a relatively intense emotional appeal is generally superior to a milder appeal. Exposure to the intense appeal resulted in fewer negative thoughts, more overall feeling responses, higher attitudes toward the advertisement, better explicit memory performance and better indirect implicit memory performance. Much advertising that is intended to be emotional might be perceived as only mildly so by many consumers. Based on these results, an advertising manager contemplating the creation of emotional advertising appeals would certainly want to ensure that the appeal was perceived by target consumers as highly emotional in order to reap the bulk of the benefits that emotional appeals can provide to a brand. EXPERIMENT 2 Hypotheses and Overview This theoretical premise of this research focused on the relative poor diagnosticity of emotional experience arising from advertisements on consumer memory. While Experiment 1 offers in sights into the relationship between emotional appeals and implicit versus explicit memory, it does not bear directly on the questions of emotional diagnosticity. The key premise of the current research is that under certain circumstances, the perceived diagnosticity of emotional experiences can be enhanced, leading to enhanced explicit memory performance. Therefore, Experiment 2 examines the degree to which that diagnosicity can be manipulated, and the impact of that manipulation upon measures of consumer implicit and explicit memory. Considerations of diagnosticity implicate involvement as well. Thus the impact that differential processing, resulting from varying degrees of involvement, has upon implicit and explicit memory is first discussed. The dual process literature has consistently demonstrated that involvement has a direct impact upon the nature of encoding operations (e.g., Eagly and Chaiken 1993). While low levels of involvement at encoding lead to shallow processing and a peripheral or heuristic route to persuasion, higher levels of

involvement promote deep, semantic level processing and a central or systematic route (Cacioppo and Petty 1982). Further, more elaborative processing increases explicit memory performance compared to shallower levels (Craik and Lockhart 1972). As implicit memory phenomena have received more attention, levels of processing effects have also been investigated and have shown that elaborative processing has effects upon tests of conceptual implicit memory but no effect upon perceptual implicit memory (Jacoby and Dallas 1981, Graf and Mandler 1984). In fact, it is this observed dissociation which led to the widespread recognition of the two types of implicit memory (Roediger 1990). Rather than focus upon involvement-based differences in processing, cognitive psychologists have limited either the ability or the opportunity of participants to process information (Eich 1984). However, involvement driven processing effects are expected to lead to the same effects. This is especially the case as involvement instructions often encourage participants to focus on perceptual features of a stimulus in low involvement conditions (i.e. looking for typos or spelling errors) versus encouraging more semantic level processing in high involvement conditions (carefully reading and thinking about the message). The overlap between these instructions and the nature of the material assessed in direct and conceptual indirect versus perceptual indirect memory tests promotes the type of transfer-appropriate processing likely to result in differential effects across the various types of memory being assessed (Roediger, Weldon and Challis 1989). H4: High (versus low) involvement will enhance explicit memory and conceptual (versus perceptual) implicit memory. The dual process models of persuasion have also conceived of affective elements in a persuasive appeal as peripheral in nature, impactful only under conditions of low involvement and peripheral or heuristic processing (Eagly and Chaiken 1993). Under central or systematic processing, affect is typically seen as a non-diagnostic cue. While such a perspective adds additional support for considering the role of affect on implicit memory, it is possible for affect to play a more central role, being processed more systematically and thus impacting explicit memory performance in some situations. Intensity of affective experience, investigated in Experiment 1 is one of those situations, as the heightened arousal believed to accompany intense affect is believed to alter the amount of issue relevant thinking (Cacioppo and Petty 1982), though in some conditions intense affect may inhibit processing more cognitive processing (Park and Young 1986). In

addition, however, the perceived relevance of the affect to a persuasive message will enhance the likelihood that it is processed in an elaborative manner (Isen 1989). For example, relevant affective states should be more diagnostic than irrelevant affective states under conditions of high involvement, and thus are more likely to be accessed in an explicit search of memory contents (Feldman and Lynch 1988) H5: High (versus low) diagnosticity of the emotional appeal will enhance consumer explicit versus conceptual and perceptual implicit memory, particularly under conditions of high involvement. To test this hypotheses, the design used is a 2 (Involvement: High versus Low) by 2 (Emotional Diagnosticity: High versus Low) by 3 (Memory Task: Perceptual Priming, Conceptual Priming versus Cued Recall) full factorial between subjects design. Method Undergraduate students at UCLA (N = 223, 100 female) were recruited to participate in a series of marketing research studies in return for payment of $10. They were first asked to complete several demographic questions, and then to review the Grad Life mock-magazine used in the previous experiment, with two changes. First, while all the advertisements featured in the magazine were still presented in full color, the editorial pages in the magazine were printed in black and white.ix Second, the target ads featured in the magazine were changed to manipulate the diagnosticity of the emotional advertising appeal to the product category, as is described in more detail below. Respondents were given the same cover story used in Experiment 1 regarding the nature of the mock-magazine. After reading through the magazine, participants completed the questionnaire consistent with the cover story, asking a series of questions about the articles they read and about potential topics for articles in the magazine. Participants then spent approximately 20 minutes completing two unrelated filler tasks. Following this was a one-page questionnaire featuring memory questions consistent with one of the three memory conditions. These will be described in more detail below. This page did not carry a headline or other tag that identified it as a continuation of the previous research associated with the Grad Life magazine, unless it was referred to directly in the memory question, as in Experiment 1. In addition, a suspicion check was included to assess whether participants connected the memory question to the previously viewed magazine. Next,

participants completed another short filler task (approximately 5 minutes) before receiving the final questionnaire, which was identical to the one used in Experiment 1. Independent Variables Emotional Diagnosticity. The diagnosticity of the emotion to the advertised brand was manipulated by varying the product category featured in the target ads. First, eight product categories were pretested to determine their appropriateness for the target stimuli. These categories were chosen based upon their relative positions on the FCB think/feel product grid (Ratchford 1987): Four categories rating as high thinking products (motor oil, batteries, health insurance, and credit cards); four categories rated as high feeling products (wine, greeting cards, perfume/cologne, gift box chocolates). All categories pretested were moderately involving (means range from 4.5 to 5.0, where 1 = not at all involving, 7 = very involving) in the FCB grid. Pretest participants (N = 54; 29 female) were asked to indicate the degree to which a decision to purchase in the category was based upon two feeling or thinking items (see Ratchford 1987). In addition, warm feeling brands were also specified so that the warm emotional manipulation would be highly relevant to the product category. Therefore, the participants were also asked to indicate the degree to which a series of ten emotions was descriptive of each product category (where 1 = not at all, 7 = very much), five were warm items (Cronbachs alphawarm = .94) and five were upbeat items (Cronbachs alphaupbeat = .92). The results of this pretest showed that batteries (mTF = 2.69) were the most thinking product category, while gift box chocolate (m TF = 6.40) was the most feeling product category. In addition, gift box chocolate was also perceived as significantly more warm (m = 6.15) than upbeat (m = 5.61; p < .05). As in Experiment 1, real, but relatively unfamiliar (Whitmans Chocolates and Rayovac Batteries) brands were used in the advertisements. x Also as in Experiment 1, copy was created focusing on the unimportant to minimize the rational content, thus making them primarily emotional appeals.xi A final pretest was conducted to create the advertisements. Each ad featured the intensely warm photograph used in Experiment 1, as well as a headline reading, The Ultimate in Warmth and Serenity Whitmans Chocolates (Rayovac Batteries). Below the photograph were the weak attributes, and the respective company logo. Participants (N = 20, 10 female) were asked to indicate the degree to which each

ad, the order of which was counterbalanced, was diagnostic with respect to product category by completing a 3-item scale regarding the diagnosticity of the appeal for making a purchase decision in the category (Isen 1989; where 1 = not at all relevant/not at all fitting/ not at all appropriate, 7 = very relevant/very fitting/very appropriate; Cronbachs alpha = .96), as well as to indicate their attitudes toward the advertisement on a fouritem scale (1 = bad/not at all likeable/negative/unfavorable; 7 = good/likeable/positive/favorable; Cronbachs alpha = .95). The results indicated higher diagnosticity ratings for the Whitmans chocolate ad (m = 4.07) than for the Rayovac batteries ad (m = 2.02, F = 3.96, p < .05), as intendedxii. Involvement . Consistent with much of the literature on the involvement construct, motivation to engage in deep or shallow processing was manipulated by changing the personal relevance of the magazine under evaluation. In the high involvement condition, subjects were given instructions indicating they are likely to serve as the target audience for the magazine and encouraging them to form a thoughtful impression of it: Attached you will find a mock-up version of a new magazine that some MBA students from the Anderson School are considering launching, starting with a Southern California edition and then expanding nationwide. Please review the magazine and form an evaluation of it. As a participant in this survey, your opinion is extremely important, and will be analyzed individually by the publishers of this magazine. Thus, your individual opinion will weigh heavily in the eventual decision to introduce this magazine in Southern California. Accordingly, please take your time to read the magazine, both the articles and the advertisements it contains, and form a careful impression of it. After you have formed a well-thought out impression of the magazine, you will be asked to answer some questions regarding your opinion of it. In contrast, subjects in the low involvement condition received instructions downplaying the relevance of the magazine to them and encouraging them to form a quick impression of it. Attached you will find a mock-up version of a new magazine that some MBA students from the Wharton School are considering launching, starting with an East Coast edition and then expanding nationwide. Please review the magazine and form an evaluation of it. As a participant in this large-scale survey, your opinion will be averaged with those of other participants, and will be analyzed at the aggregate level. Thus, an individual opinion will not weigh too much in the eventual decision to introduce this magazine to the East Coast. Accordingly, it is not necessary for you to take much time reading the magazine or its advertisements. Forming a quick impression of magazine, both the articles and the advertisements it contains, will suffice. After forming a quick impression, you will be asked to answer some questions regarding your opinion of it. Memory Tasks. All subjects participating in the experiment were asked memory questions regarding both the product category to which they were exposed, as well as the product category they did not

see in their version of the Grad Life magazine. This allowed the non-exposure group to serve as a control for the exposure group, providing a baseline measure of memory performance over which priming due to exposure can be assessed in the indirect task conditions. The memory tasks in this experiment were identical to those used in Experiment 1, with the addition of a perceptual priming indirect task. As before, participants in the direct task memory condition received a cued recall question (In the magazine, Grad Life, which you reviewed earlier, you saw an ad for a brand of BATTERIES (GIFT BOX CHOCOLATE). Please write down the name of that brand of BATTERIES (GIFT BOX CHOCOLATE) in the space below.) Participants in the indirect memory task conditions received either a conceptual priming instruction (As quickly as possible, please list all the brands of BATTERIES (GIFT BOX CHOCOLATE) you can think of in the space below.), or a perceptual priming instruction. In the latter task, subjects were given 10 brand stems, consisting of the first letter of the brand name followed by a blank (i.e., R ), and were

asked to complete the stems with the first brand name that came to mind. The list of stems included a stem beginning with R to measure the perceptual priming associated with viewing the target advertisement for Rayovac, as well as a stem beginning with W to measure the priming associated with the Whitmans appeal. Though Experiment 1 did not include a measure of perceptual priming, a significant literature suggests that in addition to the dissociation found in implicit memory performance at the conceptual and perceptual levels due to differences in involvement at processing, investigation of perceptual priming may also provide insight into the impact of emotional stimuli on implicit memory. Emotion has long been considered capable of enhancing the perceptual readiness of a perceiver. The primary goal of the New Look paradigm in perception was to assess the role that emotional states might have on the basic processes of attention and perception (Kitayama and Niedenthal 1994). The researchers associated with this paradigm argued that the emotional meaning of a stimulus could be responded to before the stimulus was consciously perceived. In particular, physical attributes of positive stimuli were expected to result in perceptual enhancement as compared to those of negative stimuli. The accompanying emotional reaction was in turn believed to determine the nature of the resulting conscious percept of the stimulus.

Affect has also been shown to influence the perception of a stimulus, by focusing and narrowing attention to the relevant perceptual code (Kitayama 1990). Affect associated with a target is produced by the preconscious activation of the target, and is therefore induced prior to conscious, attentive processing. It then in turn influences the subsequent conscious processing. For example, affective voice tone has been found to enhance word recognition, even before the verbal content is fully analyzed and comprehended (Kitayama and Howard 1994). Thus, affect appears to influence what has been termed implicit perception reflecting a data-driven process by which the affective elements of a stimulus impact subsequent processing. Similarly, Niedenthal (1990) shows that emotional expressions can be perceived implicitly, to influence subsequent judgments in an emotionally congruent fashion. However, affect can also exert a top-down, or conceptually-driven force as well, in that words or other stimuli congruent with the emotional state of an individual are more accurately perceived than are those which are incongruent with the current emotional state (Niedenthal and Showers 1991). Bower (1981) discusses the potential for emotionally-congruent words to pop-out at the perceiver. Thus, affective stimuli appear to operate at both a perceptual and a semantic level, influencing both what is perceived as well as priming associated ideas, thoughts and images (Kitayama and Howard 1994). Importantly, this work justifies thinking about the role of emotions on both perceptual and conceptual aspects of implicit memory. Dependent Variables The identical set of dependent measures used in Experiment 1 were used in Experiment 2, although slight differences in the reliability indices in the attitude measures existed: Aad (Cronbachs alpha = .93) and Ab (Cronbachs alpha = .85). In terms of emotional and cognitive responses to the appeals, thoughts were again categorized into positive, negative and neutral thoughts about both the advertisement and about the brand. In addition to those categories used in the first experiment, however, thoughts on the diagnosticity (i.e., You want to buy chocolates that are romantic and express your emotion for the person.) or non-diagnosticity (i.e, The couple and their love seemed really unrelated to the function of batteries.) of the advertisement to the product category were also coded (Isen 1989), as were irrelevant thoughts (thoughts that were theoretically meaningless, i.e., My brother buys a lot of batteries.). Feeling responses were coded exactly as in the first

experiment, into categories of warm, upbeat, negative and other feelings. Inter-rater agreement was high (93%), and disagreements were resolved by discussion. Results The hypotheses were tested using a 3 (Memory Task Type) by 2 (Emotional Relevance) by 2 (Involvement) between subjects analysis. Manipulation Checks. A manipulation check on the emotions experienced in response to the target advertisements indicates that both appeals evoked warm emotions as intended. Participants indicated the degree to which they experienced each of the five upbeat emotions (Cronbachs alpha = .87) and five warm emotions (Cronbachs alpha = .92) in response to the target advertisements on the 7-point scale summarized in Experiment 1. After subtracting the average upbeat response from the average warm response, both appeals evoked significantly more warm than upbeat emotions (ps < .05). Interestingly, however, a two-way ANOVA (df = 3, 218) on the impact of relevance and involvement on emotional responses found that the diagnostic emotional appeal (Whitmans Chocolates) evoked significantly more warm emotional responses (m = 5.04) than did the non-diagnostic emotional appeal (Rayovac Batteries, m = 4.30, F = 7.05, p < .01). No other effects were significant (Fs < 1). None of these variables had a significant impact upon upbeat emotional responses (Fs < 1). This suggests that the diagnosticity of the emotional appeal to the product category also plays a role in the degree to which emotions are experienced in response to advertisements. As in Experiment 2, subjects in all memory conditions were asked to indicate what they thought the purpose of the memory question they completed might be. Three subjects in the conceptual priming memory condition, and one in the perceptual priming condition correctly related the memory task to the advertisements they had seen in the Grad Life magazine. These subjects were eliminated from the data set. Participants completed conceptual and perceptual implicit memory tasks both for the brand to which they were exposed as well as for the non-exposed brand. Thus, those subjects who viewed the Whitmans Chocolates advertisement acted as a control group for the subjects in the Rayovac advertisement condition. Based upon the completion rates of the control groups for each brand, significant amounts of priming occurred due to exposure for both brands in the exposure groups. In the non-exposure groups, 23% of

subjects listed Whitmans Chocolates in response to the conceptual priming question, while 3% provided it in response to the perceptual priming question, in contrast to 73% and 32% respectively in the exposure group (Chi-squareconceptual priming = 18.09, p < .01, Chi-squareperceptual priming = 10.78, p < .01). Similarly, 5% of non-exposure subjects listed Rayovac in response to the conceptual priming question, and 7% completed the perceptual priming question stem with it, while 65% and 26% provided it in the exposure group (Chi-squareconceptual priming = 29.59, p < .01, Chi-squareperceptual priming = 4.56, p < .03). Both of these results indicate that there was significant priming of each brand over the pertinent base rate control group due to exposure to the target appeals. Subjects in the high involvement conditions produced significantly more thoughts in response to the target advertisements (m = 1.58) than did subjects in the low involvement conditions (m = 1.30), suggesting that the involvement manipulation worked as intended (F = 5.37, p < .03). Memory Performance. As in Experiment 1, subjects were deemed to have been either successful or unsuccessful in providing the target brand name. Analysis using a general linear model revealed a significant effect of memory condition on memory performance (Chi-square = 22.22, p < .01). This effect was driven by better memory performance in the explicit (m = .51) and conceptual-level indirect task (m = .67) conditions compared to performance by subjects in the perceptual indirect task conditions (m = .28). Overall, subjects in the perceptual task conditions had difficulty with completing the stems with the target. Emotional diagnosticity was also found to exert a main effect on memory across all conditions (Chi-square = 5.62, p < .02), with higher overall performance occurring with high diagnosticity (m = .57) compared with low diagnosticity (m = .43). No significant interaction between memory task condition and emotional diagnosticity was found (Chi-square = 1.83, p < .40). Hypothesis 4 predicts that involvement at the time of processing will result in better explicit and conceptual level (versus perceptual) implicit memory performance. There was a main effect of involvement (Chi-square = 12.57, p < .01), such that high involvement (m = .61) led to better memory performance compared to low involvement (m = .39). Further, this effect was qualified by a significant two-way interaction (Chi-square = 7.55, p < .03) between involvement and memory condition; both explicit and conceptual level implicit memory performance were better under high involvement (mdirect = .68 mconceptual indirect =

.81) than under low involvement (mdirect = .34 mconceptual indirect = .54). In contrast, perceptual-level implicit memory performance was identical under both high and low involvement (m = .28 in both conditions), supporting Hypothesis 4 and past findings in the cognitive psychology literature (Bjork and RichardsonKlavehn 1988). Emotional diagnosticity and involvement were also found to interact significantly upon memory performance (Chi-square = 3.79, p < .05) such that the best memory performance occurred in the high emotional diagnosticity and high involvement conditions (m = .73), while the low emotional diagnosticity and low involvement condition accounted for the worst performance (m = .37). See Figure 2. ---------------------------------------Insert Figure 2 about here. ---------------------------------------Finally, in support of Hypothesis 5 a marginal three-way interaction between emotional diagnosticity, memory condition and involvement was found (Chi-square = 4.18, p < .12). Consistent with predictions, explicit memory performance is much better with deeper processing at encoding and high (m = .94) versus low (m = .43) emotional diagnosticity. When the emotional appeal was relevant to the product category, the emotional cue in memory appears to cross the diagnosticity threshold as predicted, resulting in significantly better memory performance (F = 11.60, p < .01). Similarly, conceptual implicit memory was also best when both involvement and emotional diagnosticity are high (m = .95), and drops when diagnosticity was low (m = .80), however this difference is not significant (F < 1 ) in a planned contrast. Thus, as predicted, while involvement was critical for conceptual level implicit memory performance (mhigh involvement = .87, mlow involvement = .54, planned contrast F = 9.53, p < .01), emotional diagnosticity had no significant effect on this type of implicit memory. This is an important dissociation between explicit memory performance and implicit memory. Under conditions of low involvement, memory performance in each condition was flat (all Fs < 1) across the diagnosticity manipulations, suggesting that such diagnosticity is only likely to impact memory performance when effortful processing is engaged in at the time of encoding. This is not surprising as diagnosticity is expected to only be relevant when meaning-based processing occurs. In sum, while the overall

three-way interaction is not significant, this analysis provides support for Hypothesis 5 as the pattern of results is as predicted. Attitudes. Attitude results were analyzed via a 2 (Emotional Diagnosticity: High versus Low) by 2 (Involvement: High versus Low) between subjects ANOVA (df = 3, 206). Results indicate that participants have higher attitudes toward the advertisement featuring the diagnostic emotional appeal (F = 8.40, p < .01, m = 3.78) compared to the non-diagnostic emotional appeal (m = 2.58). No other significant effects on attitudes toward the advertisement were found. Similarly, there was a significant effect of diagnosticity (F = 7.26, p < .01) on attitudes toward the brand such that attitudes were higher after exposure to the diagnostic (m = 3.37) versus non-diagnostic (m = 2.99) emotional appeal. See Table 2. ---------------------------------------Insert Table 2 about here. ---------------------------------------Cognitive and Emotional Responses. The results of an ANOVA on thoughts toward the advertisement indicate that subjects produced more positive thoughts in response to the diagnostic emotional advertisement (m = .22, F = 8.35, p < .01) than to the non-diagnostic advertisement (m = .07). In addition, participants produced more thoughts regarding the non-diagnosticity of the emotional appeal to the product category after exposure to the non-diagnostic emotional appeal (m = .49, F = 14.12, p < .01) than to the diagnostic emotional appeal (m = .22). No other significant effects were found. Analysis of the feelings generated in response to the two ads finds that participants experienced more warm feelings after seeing the diagnostic (m = .47) than the non-diagnostic emotional appeal (m = .16, F = 14.15, p < .01). In addition, a main effect of involvement on warm feelings was found; high versus low involved subjects experienced more warm feelings (m = .42 vs. .21; F = 6.48, p < .02). Further, subjects experienced more overall feelings after exposure to the diagnostic appeal (m = .95) than to the non-diagnostic appeal (m = .71, F = 6.38, p < .02), as well as a main effect of involvement, such that more total feelings were experienced under high (m = .98) than under low (m = .69) involvement (F = 9.62, p < .01). No other significant effects on feelings were found.

Discussion Previous research has suggested that emotional advertising appeals do not perform well on the typical explicit memory tasks often conducted in copy-testing. This research has argued that this poor performance is primarily due to the fact that emotional advertising appeals often do not meet the diagnosticity hurdle necessary for successful explicit memory performance. Experiment 2 was conducted to determine whether a diagnostic emotional appeal could lead to successful explicit memory performance, particularly under conditions of high involvement, when diagnostic aspects of an appeal are likely to be processed in detail. Results show that diagnosticity of the emotional advertising appeal clearly has a beneficial impact upon explicit memory under conditions of high involvement. Participants had excellent explicit memory performance for the diagnostic emotional appeal, but not for the non-diagnostic emotional appeal when highly involved in processing the original message. Low involvement subjects, in contrast, were equally likely to recall both the diagnostic and the non-diagnostic emotional appeal, and their overall performance levels were lower than those of the high involvement subjects. These results strongly suggest that the previously mixed results regarding the impact of emotional advertising appeals on explicit memory may be due to the variations in the diagnosticity of the emotion to the product categories being studied. This aspect of the advertisements previously studied has not received much attention, however, results of this research argue that if explicit memory performance and a thoughtful decision process are important for brand consideration or purchase, the diagnosticity of the emotional appeal to the product category purchase is important. Further, as predicted, emotional diagnosticity does not appear to have a significant influence upon consumer implicit memory, either at the conceptual or perceptual levels, offering support for the position that implicit memory performance relies primarily upon accessibility of knowledge contained in memory rather than the perceived diagnosticity of that information. Diagnostic emotional advertising appeals are clearly important for evoking emotional responses and for producing positive attitudes toward the appeal as well as explicit memory. While emotions are evoked in response to a non-diagnostic emotional appeal, they do not appear to be as strong and are accompanied by

less positive attitudes as subjects produce fewer positive thoughts about the advertisement and more thoughts about its irrelevance for the product category. Thus diagnosticity of the emotional appeal is important not only for memory performance but for attitudes and cognitive and emotional responses toward an appeal, as well. GENERAL DISCUSSION While emotional advertising appeals have not traditionally performed well on memory tests, the present research has argued that they can in fact have a substantial impact upon consumer memory. Relying upon an accessibility/diagnosticity perspective, this research has argued that while emotional appeals may not always be likely to impact explicit memory, they may be likely to consistently impact consumer implicit memory. Results from both experiments support this premise. Results of Experiment 1 indicate that in general emotional advertising appeals may be likely to have more impact upon implicit versus explicit memory. However, results of this experiment and that of Experiment 2 also suggest circumstances under which emotional appeals may be likely to impact explicit memory, specifically under conditions of heightened emotional intensity (Experiment 1) or emotional diagnosticity (Experiment 2) of the appeal. The similarity of the results on both types of memory of these two manipulations suggests that perhaps emotional intensity impacts perceptions of diagnositicy in much the same manner as does appropriateness of the emotion to the product category featured in the appeal. It seems adaptive that intense emotional stimuli would capture more processing resources and perhaps be perceived as more relevant or diagnostic pieces of information than milder emotional stimuli, purely because of their ability to inspire intense emotional responses. And it is logical that we would want to pay more heed to those stimuli in our environment which provoke stronger reactions. However, in Experiment 2, diagnosticity as manipulated by relevance to the product category only influenced explicit memory under conditions of high involvement at the time of processing. No involvement manipulation was employed in Experiment 1, making it difficult to draw final conclusions about the ability of intensity to heighten perceptions of diagnosticity. Perhaps intensity heightens both involvement and diagnosticity or perhaps it is merely that the effects of

arousal can mimic the effects of diagnosticity under some circumstances. Both potentialities suggest interesting areas for future research. The present research offers several significant contributions to the consumer behavior and psychology literature. This is the first research to explicitly relate emotions to implicit and explicit memory performance using an integrated framework from which specific predictions can be derived. Moreover, this framework offers explanations for previously mixed results regarding the impact of emotional advertising appeals on consumer explicit memory. Finally, while a number of researchers have discussed the degree to which emotions can be made more relevant or more central to processing (i.e., Isen 1989), this represents the first attempt to specifically manipulate factors related to emotional diagnosticity in order to assess their impact on a variety of dependent variables of interest to consumer researchers. In addition to heightened explicit memory performance, the present results show that greater emotional diagnosticity also leads to more positive attitudes and thoughts toward the advertisement and the brand, and more emotional responses to the appeal. The bulk of these positive effects argue for the importance of considering emotional diagnosticity for both marketing academics and practitioners. Given such a conclusion, however, it is interesting to consider whether it is possible, with creative and well-executed advertisements, to make emotions diagnostic even in situations where it might at first seem unlikely. For example, recently a brand of outboard motors executed a series of warm, fuzzy advertisements for its product. While a priori outboard motors appear to be a more thinking than feeling category, these appeals featured vignettes of families sharing special times in the outdoors together as a result of their purchase of the product, thereby making the emotions evoked by the appeal more relevant to the ultimate purchase of the product. Perhaps there might be differences in the perceived diagnosticity of an emotional appeal across target markets. While serious sportsmen may not purchase an outboard motor for emotional reasons, perhaps families looking to enjoy the outdoors together are more likely to do so. This suggestion is in accordance with research which finds that perceptions of the diagnosticity of information may vary according to expertise with the domain of interest (cf. Alba and Hutchinson 1987).

This example also suggests that the perceived diagnosticity of emotions in advertisements may vary depending upon the ultimate goal of a marketing campaign. Perhaps the non-diagnostic emotional appeal described above might not be successfully linked in memory with the target outboard motor brand featured in the appeal, and thus perhaps not contribute to greater sales for that brand per se. However, experienced emotions might pique interest in the category overall, resulting in information search and potential purchase from within the group of brands that make up the category. Further, the emotion featured in an advertisement may make the category more relevant to traditional non-users. This suggests that a market expansion goal might more successfully rely on the use of what would otherwise be considered nondiagnostic emotions leading to explicit memory benefits than would a goal of increased market share for a specific brand. Finally, while this research provides a unique contribution in the marketing literature by highlighting the role of emotional appeals on implicit memory under a variety of important conditions, there are a number of limitations associated with the research which afford areas for future research. For example, this research focuses on the assessment of memory via brand name recall and category membership generation. These are important measures, often of interest to advertising practitioners, but they offer a relatively limited perspective on the impact of advertising on consumer memory. In large part, this limitation is due to the nature of the implicit memory measures that have been developed and studied to date. These implicit memory measures are highly relevant for work conducted in cognitive psychology experiments that expose subjects to word lists or other context free stimuli, but in the richer world of advertising testing, they are rather impoverished. Moreover, they offer a unique disadvantage to the study of major brands in product categories, as these brands are likely to be highly accessible prior to experimentation, leading to an inability to produce significant priming effects above these base levels. These issues make it important to think of new measures that are appropriate to assess implicit memory in the domain of consumer behavior. In addition, if the impact of emotional advertising upon implicit memory is to be a continued area of investigation, it may be necessary to tailor the development of new measures to the investigation of emotions in particular. While the theory reviewed above suggests that the measures used in these studies are

appropriate, they certainly are not very emotional in nature. Perhaps more emotional measures (akin to the execution-related explicit measures identified by Friestad and Thorson 1993) would show even greater effects (Krishnan and Chakravarti 1993). This research also only examines the impact of one type of emotional response, warmth, upon implicit and explicit memory. However, other types of emotional responses may impact memory differently. For example, more upbeat emotional responses, characterized by higher levels of individual arousal, may have a greater impact upon explicit memory. Moreover, the greater distinction between felt and depicted upbeat emotions (Stout, Homer and Liu 1990) may have implications for memory effects as well. Perhaps emotional appeals which successfully depict upbeat emotions but do not actually inspire upbeat feelings in consumers will primarily impact implicit memory, in contrast to the potential explicit memory effects of those which successfully generate upbeat feeling responses. Finally, as mentioned previously, the experiments in this research do not compare the impact of emotional advertising appeals with non-emotional, or rational appeals. However, this comparison is critical for truly comparing the impact of emotional advertising appeals on implicit and explicit memory performance. While the emotional appeals tested in this set of experiments perform generally in the manner expected with respect to implicit and explicit memory, additional insight is needed with respect to the impact that a non-emotional appeal might have upon the same memory measures.

Figure 1 Experiment 1: Impact of Emotional Strength on Direct (Explicit) and Indirect (Implicit) Measures of Memory

0.9 0.8 Memory Performance 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Mild Emotional Strength Intense 0.6 0.34 0.79 0.68 Conceptual Priming (IMPLICIT) Cued Recall (EXPLICIT)

Table 1 Experiment 1: Impact of Emotional Strength of Attitudes, Cognitive and Emotional Responses

Emotional Strength Mild Dependent Variable Attitudes Attitude toward the Ad Attitude toward the Brand Cognitive Responses Positive Brand Negative Brand Neutral Brand Positive Advertisement Negative Advertisement Neutral Advertisement Total Emotional Responses Warm Upbeat Negative Total Intense

4.30 3.75 .04 .10 .08 .06 .49 .08 1.13 .38 .30 .02 .70

4.20 4.50 .08 .11 .17 .06 .17 .17 .96 .70 .23 .02 .96

Figure 2 Experiment 2: Impact of Emotional Diagnosticity and Involvement on Direct (Explicit) and Conceptual and Perceptual Indirect (Implicit) Measures of Memory

High Involvement
Memory Performance 1 0.8 0.6 0.4 0.2 0 High Diagnosticity Low Diagnosticity Emotional Diagnosticity
0.43 0.3 0.33 0.95 0.94 0.8

Direct Conceptual Indirect Perceptual Indirect

Low Involvement
Memory Performance 1 0.8 0.6 0.4 0.2 0 High Diagnosticity Low Diagnosticity Emotional Diagnosticity
0.54 0.33 0.26 0.53 0.35 0.24

Direct Conceptual Implicit Perceptual Implicit

Table 2 Experiment 3: Impact of Emotional Diagnosticity and Involvement on Attitudes, Cognitive and Emotional Responses

Involvement High Emotional Diagnosticity High Dependent Variable Attitudes Attitude toward the Ad Attitude toward the Brand Cognitive Responses Positive Advertisement Negative Advertisement Neutral Advertisement Positive Brand Negative Brand Neutral Brand Diagnosticity Non-diagnosticity Total Emotional Responses Warm Upbeat Negative Total 3.74 3.22 .29 .55 .23 .04 .06 .08 .00 .21 1.62 .62 .49 .00 1.17 Low 3.82 3.50 .06 .42 .23 .06 .04 .09 .02 .47 1.52 .33 .51 .02 .75 Low Emotional Diagnosticity High 2.56 2.99 .15 .44 .23 .07 .03 .03 .00 .23 .23 .38 .03 .79 Low 2.59 2.99 .08 .42 .15 .06 .04 .06 .00 .52 1.23 .10 .44 .00 .63

REFERENCES Aaker, David A. and Donald E. Bruzzone (1981), Viewer Perceptions of Prime-Time Television Advertising, Journal of Advertising Research, 21 (October) 15-23. ------------, Douglas M. Stayman and Michael R. Hagerty (1986), Warmth in Advertising: Measurement, Impact and Sequence Effects, Journal of Consumer Research, 12 (March) 365-381. Alba, Joseph W. And Amitava Chattopadhyay (1986), Salience Effects in Brand Recall, Journal of Marketing Research, Vol. 23 (November) 363-369. --------------- and J. Wesley Hutchinson (1987), Dimensions of Consumer Expertise, Journal of Consumer Research, 13 (March), 411-454. Anderson, John R. (1983), The Architecture of Cognition, Cambridge, MA: Harvard University Press. Banaji, Mahzarin R. and Anthony G. Greenwald (1994), Implicit Stereotyping and Unconscious Prejudice, in The Psychology of Prejudice, The Ontario Symposium, eds. Mark P. Zanna and Jerry M. Olson, Vol. 7, Hillsdale, NJ: Lawrence Erlbaum Associates, 55-76. Bargh, John A. and P. Pietromonaco (1982), Automatic Information Processing and Social Perception: The Influence of Trait Information Presented Outside Conscious Awareness on Impression Formation, Journal of Personality and Social Psychology, 43, 437-449. Batra, Rajeev and Michael Ray (1986), Affective Responses Mediating Acceptance of Advertising, Journal of Consumer Research, 13 (September), 234-239. Berger, David (1981), A Retrospective: FCB Recall Study, Advertising Age, October 26, 5-36. Bjork, Robert A. and Alan Richardson-Klavehn (1988), Measures of Memory, Annual Review of Psychology, 39, 475-543. Bower, Gordon H. (1981), Mood and Memory, American Psychologist , 36 (February) 129-148. Burke, Marian Chapman and Julie A. Edell (1989), The Impact of Feelings on Ad-Based Affect and Cognition, Journal of Marketing Research, Vol. XXVI (February), 69-83. Cacioppo, John T. and Richard E. Petty (1989), The Elaboration Likelihood Model: The Role of Affect and Affect-Laden Information Processing in Persuasion, in Cognitive and Affective Responses to Advertising , eds. P. Cafferata and A. Tybout, Lexington, MA: Lexington Books, 69-90. Clark, M., S. Milberg and J. Ross (1983), Arousal Cues Arousal-related Material in Memory: Implications for Understanding Effects of Mood on Memory, Journal of Verbal Learning and Verbal Behavior , 22, 633649. Clore, Gerald L., Norbert Schwarz and Michael Conway (1994), Affective Causes and Consequences of Social Information Processing, in Handbook of Social Cognition, Second Edition, (1), eds. R. Wyer, Jr. and T Srull, Hillsdale, NJ: Lawrence Erlbaum Associates, 323-417. Craik, Fergus I.M. and Robert S. Lockhart (1972), Levels of Processing: A Framework for Memory Research, Journal of Verbal Learning and Verbal Behavior , 11, 671-684.

Devine, Patricia G. (1989), Stereotypes and Prejudice: Their Automatic and Controlled Components, Journal of Personality and Social Psychology, 56 (1), 5-18. Eagly, Alice H. and S. Chaiken (1993), The Psychology of Attitudes, New York: Harcourt, Brace Jovanovich. Edell, Julie A. and Marian Chapman Burke (1987), The Power of Feelings in Understanding Advertising Effects, Journal of Consumer Research, Vol. 14 (December), 421-433. Eich, Eric (1984), Memory for Unattended Events: Remembering With and Without Awareness, Memory and Cognition, 12, 105-11. Ellis, H. C. (1985), On the Importance of Mood Intensity and Encoding Demands in Memory: Commentary on Hasher, Rose, Zacks, Sanft and Doren, Journal of Experimental Psychology: General, 114, 392-395. Fazio, Russell H., David M. Sanbonmatsu, Martha C. Powell and Frank R. Kardes (1986), On the Automatic Activation of Attitudes, Journal of Personality and Social Psychology, 50 (February), 229-238. Feldman, Jack M. and John G. Lynch, Jr. (1988), Self-Generated Validity and Other Effects of Measurement on Belief, Attitude, Intention and Behavior, Journal of Applied Psychology, Vol. 73 (August), 421-435. Friestad, Marian and Esther Thorson (1986), Emotion-Eliciting Advertising: Effects on Long-Term Memory and Judgment, in Advances in Consumer Research, ed. Richard J. Lutz, 13, Provo, UT: Association for Consumer Research, 111-116. ------------ and Esther Thorson (1993), Remembering Ads: The Effects of Encoding Strategies, Retrieval Cues and Emotional Response, Journal of Consumer Psychology, 2 (1), 1-23. Gardner, Meryl Paula (1985), Mood States and Consumer Research: A Critical Review, Journal of Consumer Research, Vol. 12 (December), 281-300. Graf, Peter, George Mandler and P. E. Haden (1982), Simulating Amnesic Syndrome in Normals, Science, 218, 1243-44. ------------ and Daniel L. Schacter (1987), Selective Effects of Interference on Implicit and Explicit Memory for New Associations, Journal of Experimental Psychology: Learning, Memory and Cognition, 13, 45-53. Hardin, Curtis and Mahzarin Banaji (1990), Affective Intensity and Valence in Memory, Paper presented at the annual meeting of the American Psychological Association, Boston. Holbrook, Morris B. and Elizabeth C. Hirschman (1982), The Experiential Aspects of Consumption: Consumer Fantasies, Feelings and Fun, Journal of Consumer Research, 9 (September) 132-140. Isen, Alice M. (1989), Some Ways in Which Affect Influences Cognitive Processes: Implications for Advertising and Consumer Behavior, in Cognitive and Affective Responses to Advertising , eds. Patricia Cafferata and Alice M. Tybout, Lexington, MA: Lexington Books, 91-118. Izard, Carroll E. (1977), Human Emotions, New York: Plenum Press. Jacoby, Larry L. and M. Dallas (1981), On the Relationship Between Autobiographical Memory and Perceptual Learning, Journal of Experimental Psychology: General, 3, 306-340.

Kitayama, Shinobu (1990), Interaction Between Affect and Cognition in Word Perception, Journal of Personality and Social Psychology, 58 (2), 209-217. ------------and Paula M. Niedenthal (1994), Introduction, in The Hearts Eye: Emotional Influences in Perception and Attention, ed. Paula M. Niedenthal and Shinobu Kitayama, San Diego, CA: Academic Press, 5-14. ------------ and Susan Howard (1994), Affective Regulation of Perception and Comprehension: Amplification and Semantic Priming, in The Hearts Eye: Emotional Influences in Perception and Attention , ed. Paula M. Niedenthal and Shinobu Kitayama, San Diego, CA: Academic Press, 41-67. Kihlstrom, John F. (1993), The Psychological Unconscious and the Self, in Experimental and Theoretical Studies of Consciousness, Ciba Foundation Symposium 174, England: John Wiley & Sons, 147-155. Krishnan, H. Shanker and Dipankar Chakravarti (1993), Varieties of Brand Memory Induced by Advertising: Determinants, Measures and Relationships, in Brand Equity and Advertising: Advertisings Role in Building Strong Brands, eds. David A. Aaker and Alexander L. Biel, Hillsdale, NJ: Lawrence Erlbaum Associates, 213-234. ------------ and Stewart Shapiro (1996), Comparing Implicit and Explicit Memory for Brand Names From Advertisements, Journal of Experimental Psychology: Applied, (June), Kunst-Wilson, William Raft and Robert B. Zajonc (1980), Affective Discrimination of Stimuli that Cannot be Recognized, Science, 307, 557-558. Lazarus, Richard S. (1982), Thoughts on the Relations Between Emotions and Cognition, American Psychologist , 37, 1019-1024. Lee, Angela Y. (1995), Effects of Stimulus Exposure on Information Processing: An Implicit Memory Perspective, Unpublished Doctoral Dissertation. Lynch, John G. Jr., Howard Marmorstein and Michael F. Weigold (1988), Choices from Sets Including Remembered Brands: use of Recalled Attributes and Prior Overall Evaluations, Journal of Consumer Research, 15 (September), 225-233. ------------, Srull, Thomas K. (1982), Memory and Attentional Factors in Consumer Choice: Concepts and Research Methods, Journal of Consumer Research, 9, 18-37. Mehrabian, Albert and James A. Russell (1974), An Approach to Environmental Psychology, Cambridge, MA: MIT Press. Nedungadi, Prakash (1990), Recall and Consumer Consideration Sets: Influencing Choice without Altering Brand Evaluations, Journal of Consumer Research, 17 (December), 263-276. Niedenthal, Paula M. (1990), Implicit Perception of Affective Information, Journal of Experimental Social Psychology, 26, 505-527. ------------and Carolin Showers (1991), The Perception and Processing of Affective Information and its Influences on Social Judgment, in Emotion and Social Judgment , ed. Joseph P. Forgas, Oxford: Pergamon Press, 125-143. Page, Thomas J., Jr., Esther Thorson and Maria Papas Heide (1990), The Memory Impact of Commercials Varying in Emotional Appeal and Product Involvement, in Emotion in Advertising: Theoretical and

Practical Explorations, eds. Stuart J. Agres, Julie A. Edell and Tony M. Dubitsky, New York: Quorum Books, 255-268. Park, C. Whan and S. Mark Young (1986), Consumer Response to Television Commercials: The Impact of Involvement and Background Music on Brand Attitude Formation, Journal of Marketing Research, XXIII (Feb.), 11-24. Plutchik, Robert (1980), Emotion: A Psychoevolutionary Synthesis, New York: Harper and Row. Ratchford, Brian T. (1987), New Insights about the FCB Grid, Journal of Advertising Research, August/September, 24-38. Roediger, Henry L. III (1990), Implicit Memory: Retention without Remembering, American Psychologist , 45 (9), 1043-1056. ------------, Kavitha Srinivas and Mary Susan Weldon (1989), Dissociations Between Implicit Measures of Retention, in Implicit Memory: Theoretical Issues, eds. Stephen Lewandosky, John C. Dunn and Kim Kirsner, Hillsdale, NJ: Lawrence Erlbaum Associates, 67-86. Scarborough, D.L., L. Gerard and C. Cortese (1979), Accessing Lexical Memory: The Transfer of Word Repetition Effects Across Task and Modality, Memory and Cognition, 7, 3-12. Schacter, Daniel L. (1987), Implicit Memory: History and Current Status, Journal of Experimental Psychology: Learning, Memory and Cognition, 13 (3), 501-518. ------------, Jeffrey Bowers and Jill Booker (1989), Intention Awareness and Implicit Memory: The Retrieval Intentionality Critierion, Implicit Memory: Theoretical Issues, eds. Stephen Lewandosky, John C. Dunn and Kim Kirsner, Hillsdale, NJ: Lawrence Erlbaum Associates, 47-65. Schwarz, Norbert and G. L. Clore (1983), Mood, Misattribution and Judgments of Well-Being: Informative and Directive Functions of Affective States, Journal of Personality and Social Psychology, 45, 513-523. Smith, Eliot R. and Nyla R. Branscombe (1988), Category Accessibility as Implicit Memory, Journal of Experimental Social Psychology, 24, 490-504. Smith, S. M. (1988), Environmental Context-dependent Memory, in Memory in Context: Context in Memory, ed. D. M. Thomson and G.M. Davies, New York: Wiley. Srull, Thomas K, (1983), Affect and Memory: The Impact of Affective Reactions in Advertising on the Representation of the Product Information in Memory, in Advances in Consumer Research, Vol. 10 ed. Richard Bagozzi and Alice M. Tybout, Provo, UT: Association for Consumer Research. Stayman, Douglas M., David A. Aaker and Donald E. Bruzzone (1989), The Incidence of Commercial Types Broadcast in Prime Time 1976-1986, Journal of Advertising Research, July, 26-33. Stout, Patricia A., Pamela M. Homer and Scott S. Liu (1990), Does What We See Influence How We Feel? Felt Emotions versus Depicted Emotions in Television Commercials, in Emotion in Advertising: Theoretical and Practical Explorations, eds. Stuart J. Agres, Julie A. Edell and Tony M. Dubitsky, New York: Quorum Books, 195-210.

Thorson, Esther and Marian Friestad (1989), The Effects of Emotion on Episodic Memory for Television Commercials, in Cognitive and Affective Responses to Advertising , eds. Patricia Cafferata and Alice M. Tybout, Lexington, MA: Lexington Books, 305-326. ------------ and Thomas J, Page, Jr. (1988), Effects of Product Involvement and Emotional Commercials on Consumers Recall and Attitudes, in Nonverbal Communication in Advertising , ed. Sid Hecker and David W. Stewart, New York: DC Heath and Company, 111-125. Tobias, Betsy A., John F. Kihlstrom and Daniel L. Schacter (1992), Emotion and Implicit Memory, in The Handbook of Emotion and Memory, ed. Sven-Ake Christianson, Hillsdale, NJ: Lawrence Erlbaum Associates, 67-92. Tulving, Endel and D. M. Thompson (1973), Encoding Specificity and Retrieval Processes in Episodic Memory, Psychological Review, 80, 352-372. Vanden Abeele, Piet and Douglas L. Maclachlan (1994), Process Tracing of Emotional Responses to TV Ads: Revisiting the Warmth Monitor, Journal of Consumer Research, 20 (March) 586-600. Wright, Peter (1973), The Cognitive Processes Mediating Acceptance of Advertising, Journal of Marketing Research, 19 (February) 53-62. Zajonc, Robert B. (1980), Feeling and Thinking: Preferences Need no Inferences, American Psychologist, 35 (2), 151-175. Zielske, Hubert A. (1982), Does Day-after Recall Penalize Feeling Ads? Journal of Advertising Research, 22 (1), 19-23.

ENDNOTES
In examining the impact of emotional advertising appeals, researchers have often drawn distinctions between the broader concept of affect and the more specific terms, mood and emotions or feelings. Mood states are typically thought of as relatively mild, more enduring affective states, not evoked in response to a single encounter with a particular stimulus (cf. Isen, 1989). In contrast, emotions or feelings (often used interchangeably, e.g., Burke and Edell 1989) are stimulus-specific responses which are typically shortlived (Gardner 1985). Drawing on prior research which focuses on emotional responses to persuasion appeals rather than mood effects (e.g., Batra and Ray 1986, Aaker, Stayman and Hagerty 1986), we focus only on emotions or feelings in this research. ii Note, however, that Bjork and Richardson-Klavehn (1988) distinguish between intentional explicit memory, which is characterized by a conscious, strategic attempt to remember, and involuntary explicit memory, defined as the spontaneous re-experiencing or reconstruction of an episode. It is the former type of explicit memory that is the focus of discussion throughout this research. iii While theoretically it is possible to cross the task and memory types, typically direct measures are used to assess explicit memory, while implicit memory is measured indirectly, and these terms appear to be emerging as the standard vocabulary for measuring these effects (cf. Merikle and Reingold 1991). iv As in research in cognitive psychology which exposes subjects to multiple repetitions of the target stimulus (Bjork and RichardsonKlavehn 1987; hence the term repetition priming), subjects saw the target advertisement twice in the mock-magazine. v To identify important and unimportant attributes of film, an independent set of participants rated a series of 12 film attributes on their importance (1 = not at all important; 7 = very important). The three most important attributes identified were: Color quality (m = 6.69); sharpness of image (m = 6.53); and price ( m = 5.08). The three least important attributes were: Package design (m = 2.00); manufacturer participation in event sponsorship (m = 2.23); and country of origin (m = 2.00). vi A real rather than fictitious brand was chosen for methodological reasons, specifically for the conceptual priming measure, which will consist of the generation of category membership. While a fictitious brand would be a cleaner stimulus for other measures, it was deemed unlikely that a fake brand would appear on a list of category members, unless participants relied upon their explicit memory for the fictitious brand name. Thus, consistent with the approach taken in the cognitive psychology literature, Agfa was chosen as it is a lesser-known member of the photographic film category. To insure this, a pretest was conducted in which participants (N = 13) were asked to list all brands of film they could recall. Two participants (15%) included Agfa on that list (Kodak and Fuji were listed by all participants). Subsequently, participants rated their familiarity with Agfa (m = 2 .53), Kodak ( m = 6.31) and Fuji ( m = 5.46), as well as their use of each brand (Agfa m = 1.61, Kodak m = 5.61, Fuji m = 3.92) on scales of 1 to 7 (where 1 = not at all familiar, 7 = very familiar; and 1= never use, 7 = always use). These results suggest that while Agfa is known by American consumers, it is much less well-known than Kodak and Fuji, and is thus suitable for the assessment of conceptual priming. vii Specifically, the SAS procedure CATMOD was used to analyze all the memory data in the paper. This is a procedure which allows for the analysis of categorical data that can be represented by a two dimensional contingency table. Here, the term ANOVA is used to denote the analysis of response functions and the partitioning of variation among those functions into various sources (SAS/STAT Users Guide, Volume 1, Version 6). viii Most subjects indicated that the task was an attempt to see which brands came to mind most quickly. ix This change was made for logistical reasons as it was quite time-consuming and expensive to create more than 200 full-color copies of the target magazine. x To select the brand in each category, an independent set of pretest participants (N = 19; 9 female) were asked to list all the brands of batteries and gift box chocolates they could recall. To mirror the stimuli used in Experiment 1, real, but lesser-known brands were selected. While the Energizer and Duracell brands of batteries were listed by virtually all participants (100% and 95% respectively), Rayovac Batteries was mentioned by just 25% of subjects. Similarly, Whitmans chocolates was mentioned again by 25%, compared with Godiva and Sees Candies, which were mentioned by nearly all participants (98% and 93% respectively). xi Pretest participants generated three highly important attributes for batteries: length of life ( m = 6.0), price ( m = 5.45) and an included battery tester ( m = 5.3) and three low important attributes: U.S. company (m = 2.35), colorful batteries (m = 2.8) and unique company logo (m = 2.3). The three most important attributes for gift box chocolates were taste ( m = 6.8); smell ( m = 6.0); and flavor variety ( m = 5.30); the three least important attributes were colorful chocolates (m = 3.3); company sponsors special events (m = 2.75); and old-fashioned company logo (m = 2.65). xii Additional analysis indicated that emotional diagnosticity did not significantly affect Aad ( F = 1.79, p < .19), though the means were directionally consistent with such an effect ( m high diagnosticity = 2.45; mlow diagnosticity = 1.85). The diagnostic emotional appeal generated significantly more warm feelings (m = .47) and less negative feelings (m = .32) than did the non-diagnostic appeal (m warm = .12; m negative = .76; ps < .05). There were no differences in the number of upbeat (F = 1.12, p < .30) or total ( F < 1) feelings about the appeals.
i

Você também pode gostar