Emotions and Motivationa in Advertising

John R. Rossiter, Australian Graduate School of Managment
Larry Percy, Lintas:USA
ABSTRACT - In this paper, we update Rossiter and Percy's (1987) theory of emotions and motivations in advertising, which states that: (1) a brand attitude must link the brand to a purchase motivation; (2) this linkage is effected via one or more benefit claims, which contain a cognitive component and an emotional component with the elicited emotion or emotions serving the purchase motivation; (3) informational (negative reinforcement) motives require at least a negative-to-neutral sequence of emotions whereas transformational (positive reinforcement) motives require at least a neutral-to-positive shift although the neutral part of this sequence is usually assumed as the consumer's initial exposure state; and (4) various informational and transformational motives are most effectively served by specific types of emotions. We present a pilot test of the theory. We also comment on the role of arousal and on other advertising theorists' attempts to explain the application of emotions in advertising.
[ to cite ]:
John R. Rossiter and Larry Percy (1991) ,"Emotions and Motivationa in Advertising", in NA - Advances in Consumer Research Volume 18, eds. Rebecca H. Holman and Michael R. Solomon, Provo, UT : Association for Consumer Research, Pages: 100-110.

Advances in Consumer Research Volume 18, 1991      Pages 100-110

EMOTIONS AND MOTIVATIONA IN ADVERTISING

John R. Rossiter, Australian Graduate School of Managment

Larry Percy, Lintas:USA

ABSTRACT -

In this paper, we update Rossiter and Percy's (1987) theory of emotions and motivations in advertising, which states that: (1) a brand attitude must link the brand to a purchase motivation; (2) this linkage is effected via one or more benefit claims, which contain a cognitive component and an emotional component with the elicited emotion or emotions serving the purchase motivation; (3) informational (negative reinforcement) motives require at least a negative-to-neutral sequence of emotions whereas transformational (positive reinforcement) motives require at least a neutral-to-positive shift although the neutral part of this sequence is usually assumed as the consumer's initial exposure state; and (4) various informational and transformational motives are most effectively served by specific types of emotions. We present a pilot test of the theory. We also comment on the role of arousal and on other advertising theorists' attempts to explain the application of emotions in advertising.

INTRODUCTION

One of the most important jobs for creative people in advertising is to select the proper types of advertising stimuli that are likely to elicit appropriate emotions in the target audience. Creative people are usually given the benefit or benefits that they are supposed to associate with the client's brand, and their task becomes one of generating a creative idea or overall theme for the advertising as well as converting the benefits into benefit claims that distinguish the brand from competing brands. Whereas benefits can be conceptualized as verbal or visually implied statements linking a brand to an attribute, benefit claims make this verbal or visually implied statement in a way that elicits an appropriate emotion or sequence of emotions (Rossiter and Percy 1987). In short, a benefit lacks emotion, but a benefit claim depends upon emotion. Closely competing brands usually target the same benefit or set of benefits. In the detergent category, both Tide and All claim to get all the dirt out. In the world of trucks, Ford and Chevy talk about basically the same rugged performance features. And how is one to tell the difference between really good perfumes such as Obsession and Opium? More and more, what differentiates the effectiveness of advertising in most categories is not benefit selection but rather the ability of the brand's advertising agency to make benefit claims in an emotionally more compelling (informational) or engaging (transformational) way.

Emotional stimuli, of course, are not just inserted ad lib in ads. Rather, it is our contention (Rossiter and Percy 1987) that emotional stimuli are inserted, or should be inserted, to serve an underlying purchase (or usage) motivation (a motivation or motive being defined as a behavioral energizing mechanism). This emotional energizing takes place in the brand attitude communication effect. In the first section of this paper, we describe more fully (with some updating) our 1987 theory of emotions and motives in advertising. In the second section, we present a pilot test of the theory. In the final section, we comment on attempts by other advertising theorists and researchers to conceptualize and measure emotions in advertising.

ROSSITER AND PERCY'S THEORY

Rossiter and Percy (1987, Chapters 6-10) propose what amounts to a theory of emotions and motivations in advertising. Their theory centers on the communication effect of brand attitude (or preference, as called by marketing scientists). They define brand attitudes consisting essentially of a propositional link between a brand and a purchase motivation; the proposition can be established by, among other causes, advertising. A consumer or buyer could thus hold several attitudes (but only several) toward a brand, depending on the purchase motivation operative at the time. In defining brand attitude in terms of motivation, Rossiter and Percy (see also Katz 1960; Fennell 1975, 1978, 1989; Lutz 1978) depart from the "overall evaluation" definition of attitude popularized by Fishbein and Ajzen (1975) and argue that brand attitude must have an energizing function. "Overall evaluation" or "global affect" becomes merely a particular type of attitude where the purchase motive is sensory gratification. In our theory, there is always evaluation for a purpose (see especially Fennell 1989). We maintain that advertising practitioners (creatives) are much more likely to implicitly use our definition of brand attitude than to advertise a brand on "overall affect." However, a belief-evaluation or expectancy-value conceptualization is still included in the theory in that this is the way that benefit claims (the "surface structure" in ads) are hypothesized to make the propositional link between the brand and the motivation (this link, the attitude, being the "deep structure"). Figure 1 depicts this conceptualization of brand attitude, showing: the brand, as represented by the prior communication effect of brand awareness; one or more belie& associated with the brand; an emotion associated with each belief; with the emotions tapping the purchase motivation. The standard multi-attribute formulation can be applied to this model with the addition of a subscript, m, representing the particular purchase motivation:

FIGURE 1

ROSSITER AND PERCY'S (1987) CONCEPTUALIZATION OF BRAND ATTITUDE

EQUATION

where

"bm = attitude toward brand b for purchase motivation m

Bbim = belief that brand b delivers on benefit claim i in relation to purchase motive m

Eim = emotional effectiveness of benefit claim i in energizing motive m

n = number of benefit claims.

Rossiter and Percy (1987) proposed eight motives (Table 1) as energizing all human behavior including purchase and consumption behavior. Their textbook was not the place to explain the theoretical basis of these motives but perhaps this paper does present an appropriate forum. Essentially we adhered to a homeostatic or equilibrium concept of motivation (Figure 2) in which there are two fundamental motivating mechanisms: the onset of a negative stimulus, which motivates the consumer to reduce or remove that stimulus; and the onset of a positive stimulus, which motivates the consumer to seek that stimulus until satiation sets in and the consumer returns to equilibrium. It is in the postulation of a positive (but temporary) departure from equilibrium that we differ from Fennell's ground-breaking conceptualization of motivation (most recently in her 1989 paper). Fennell postulates that all motives operate through deprivation and therefore involve negative reinforcement. Our negative motives (with her closest equivalents shown in parentheses) are as follows: problem removal (escape), problem avoidance (prevention), incomplete satisfaction (frustration), mixed approach-avoidance (conflict), and normal depletion (maintenance). Our positive motives (again with her motives' closest equivalents shown in parentheses) are as follows: sensory gratification (sensory pleasure opportunity), intellectual stimulation or mastery (exploratory interest), and social approval (no equivalent). Fennell's last two motives are seen as negative in that a sensory pleasure opportunity is one that the "actor is not yet enjoying," and exploratory interest is, according to her theory, due to the actor "experiencing difficulty in processing information" (1989, p. 45). Thus, Fennell sees these positive stimulus onsets as being externally-induced deprivations, a viewpoint with which we disagree (surely, the whole world is not motivated by deprivation!). It should be noted, however, that this theoretical point is somewhat moot. Arguing slightly differently from Fennell, our positive motives entail a transition from non-reward, but not necessarily punishment as Fennell implies, to reward. As Wagner (1969) points out, whereas the transition from non-reward to reward is reinforcing, it is not clear whether the reinforcement is due to the initiation of reward or to the termination of non-reward. But we have an additional argument for our two-mechanism model of motivation over Fennell's, and this is shown in our emotional components subsequently. For positive motives, there is no necessity, and indeed rarely any advisability, in showing deprivational states in advertising prior to portraying positive stimulus onset in the form of sensory gratification, intellectual stimulation, or social approval.

TABLE 1

ROSSITER AND PERCY'S (1987) EIGHT PURCHASE MOTIVES

FIGURE 2

THE TWO FUNDAMENTAL MOTIVATING MECHANISMS, S- ONSET AND S+ ONSET, AND THEIR RESULTING DRIVE AND AFFECTIVE DYNAMICS

TABLE 2

HAMMOND'S (1970) RECONCEPTUALIZATION OF MOWRER'S (1960A,B) THEORY OF EMOTIONS

Emotions enter our motivational theory by extending the innovative conceptualization proposed by Mowrer (1960a, b). To introduce Mowrer's conceptualization, we have to distinguish emotions from motivations. Motivations have as their antecedents the operations of deprivation and presentation as implied in Figure 2 earlier (see also Skinner 1938; Estes and Skinner 1941; Millenson 1967; Strongman 1987) and are usually controlled by internal stimulus changes. Emotions, on the other hand, have as their antecedents abrupt external stimulus change. Emotional responses relate to abrupt external stimulus changes, and thus to motivations, as shown in Table 2, which is a summary of Mowrer's conceptualization but with modifications proposed by Hammond (1970). For Mowrer, external stimulus changes which elicit hope and relief are postulated to produce approach behavior, whereas those that elicit fear and disappointment are postulated to elicit avoidance behavior. In Hammond's reconceptualization, hope and fear lead to excitatory behavior whereas relief and disappointment are inhibitory. Hammond's conceptualization fits precisely with our dual-mechanism homeostatic conceptualization.

Our theory of emotions and motivations in advertising gains two important advantages from its origins in Mowrer's theory. The first advantage is that, while there are numerous typologies of emotion in psychology (Davitz 1969), and now in advertising (see Batra and Holbrook 1990 for a recent typology), our motivational theory of emotions in advertising proposes that the effective emotions will be close variations of Mowrer's four, namely: fear, hope, relief and disappointment. Secondly, Mowrer's theory emphasizes stimulus change and thus a potential sequence of emotions as being motivational. Every other advertising theorist whose work we are aware of, with the possible exception of Fennell although she proposes a motivational theory without emotional specifics, has proposed a single emotion theory (e.g., Aaker, Stayman, and Hagerty 1986; Edell and Burke 1987; Batra and Holbrook 1990). By fitting emotions to motivations, it can be seen that, for motives that are subject to negative reinforcement, at least a two-step sequence of emotions is necessary: from some negative state (fear [S- ON] or disappointment [S+ OFI 1) to relief (S- OFF). On the other hand, the motives that are subject to positive reinforcement may begin with a negative state (as Fennell suggests) or may simply start from neutrality and move to hope (S+ ON). In fact, recalling Charles Revson's famous comment about perfume, that "in the store we sell hope," we could turn Fennell's conceptualization around and argue that all ads make a "promise" and thereby invoke hope--whether this hope be for termination of a negative state (negative reinforcement) or for onset of a positive state (positive reinforcement). In this sense, although this is probably unnecessarily complicating the theory, all motives could be seen as energized, in advertising, by S+ onset, which is precisely the opposite of Fennell's view that all motives are energized by S- onset. The important point, however, is that in many brand attitude situations in advertising, dynamic sequences of emotions are necessary and it would be entirely misleading to look for the presence of a single emotion except in the special case of transformational advertising which we propose as being most suited to positive motives. This point will be discussed further in the applications comments at the conclusion of the paper.

A further aspect of the motivationally relevant role of emotions in our theory is that the dual-mechanism conceptualization of "negative disequilibriation" and "temporary positive stimulus- seeking" led us to appreciate the importance not only of the two ubiquitous dimensions of emotions, unpleasantness-pleasantness (P) and intensity or arousal (A), but also the third dimension as identified by Osgood (1955; and later Mehrabian and Russell 1974; Mehrabian 1980; see also Donovan and Rossiter 1982) of dominance (D) although we prefer Osgood's term "control." Human beings are much concerned with control of their motivational states and it seems evident that any homeostatic theory of motivation that proposes a dynamic sequence of emotions would emerge with a dimension of control. The P-A framework that we favored was that of Russell and Pratt (1980; see also Rossiter and Percy 1987, pp. 211-212) but we also added back the dimension, D, of dominance or control.

TABLE 3

PRESENT VERSION OF ROSSITER AND PERCY'S HYPOTHESIZED RELATIONSHIPS LINKING EMOTIONS TO MOTIVATIONS IN ADVERTISING

The latest version of our hypothesized relationships linking emotions to motivations in advertising is presented in Table 3 (modified slightly from Rossiter and Percy 1987, p. 213, Table 84). In the theory as proposed in 1987, we stated (p. 213) that "it is possible to give a general indication of typical emotions that might be used to portray [the motives] (and thus stimulate the motivation) in advertisements": anger -> relief (problem removal), fear -> relaxation (problem avoidance), disappointment -> optimism (incomplete satisfaction), guilt -> peace of mind (mixed approach-avoidance), mild annoyance -> convenience (normal depletion), neutral or dull -> elated (sensory gratification), neutral or bored -> excited (intellectual stimulation), neutral or apprehensive -> flattered (social approval). A colleague, Robert J. Donovan, has been applying our 8-motive scheme in advertising research for a number of years now (Rossiter, Percy, and Donovan 1990) and has suggested two modifications. The first modification is that "intellectual stimulation" should be bracketed with the additional motive descriptor "mastery," so we have added the emotion sequence of neutral or naive -> competent to the original intellectual stimulation motive's emotion sequence of neutral or bored -> excited. The second modification is that "social approval" should be bracketed with an additional motive descriptor conformity," so we have added the emotion sequence of neutral or ashamed -> proud to the original social approval motive's emotion sequence of neutral or apprehensive -> flattered; however, just as we cautioned originally (1987, p. 213) that social approval can sometimes stem from social anxiety and thus more correctly be classified as problem avoidance, so too can conformity, as is shown by Bearden and Rose (1990). Moreover, in the first stage of the pilot test of our theory, we employed the original emotional adjective pair anger -> relief to represent problem removal but have decided, based on feedback from the respondents, to replace anger with the less emotionally intense term "annoyance," since very few ads attempt to deliberately elicit the intense emotion of anger. Similarly, the second emotion used originally for sensory gratification, "elated," seemed too intense, so we substituted the term "sensory anticipation." The final modification, also based on the pilot test, is that mixed approach-avoidance seems better characterized by the emotion pair conflict -> peace of mind than by guilt -> peace of mind, so the former pair is now employed. It may also be noted that we were not entirely happy with the second term in the emotion pair for normal depletion, mild annoyance -> convenience. The technically correct term for "convenience," since it is not really an emotion, would be "equilibrium" but this is not very descriptive. We are still searching for a better term for this emotion.

Rossiter and Percy's (1987) theory also makes explicit predictions about the relative contribution of the emotional component of benefit claims according to the type of brand attitude that is being targeted. In 1984, we proposed a quadrant theory which is an attitudinal classification that vaguely resembles the well-known FCB grid but is much more sophisticated and operational (Rossiter, Percy, and Donovan 1984; Rossiter and Percy 1985; Rossiter and Percy 1987; Rossiter, Percy, and Donovan 1990). In the low involvement-informational quadrant, we predicted that the problem-solution emotional sequence would be most effective (which fits the negative motives' emotion sequences in Table 3) and that this would operate directly on brand attitude (Ab in conventional terminology). In the low involvement-transformational quadrant, we predicted that the (here positive) emotional portrayal would work via Aad (and this is the only quadrant where we hypothesized a major mediating role for Aad to subsequently affect Ab) because "emotional authenticity" (Wells 1981) of the advertising is essential to influence this type of brand attitude. In the high involvement-informational quadrant, we predicted that correct emotional portrayal would in general be less important in affecting Ab except early in the product category life cycle where emotional portrayal is necessary to sell the category need, and that the cognitive component would make the dominant contribution. In the high involvement-transformational quadrant, we predicted that the high involvement aspect would have an arousal-like multiplicative tendency to amplify and spread "emotional authenticity" such that there would be significant individual differences in what was perceived to be an authentic emotional portrayal, leading us to postulate authenticity for "lifestyle" segments, with this authenticity operating directly on Ab, not on Aad if Aad is defined as "overall ad likability." As we will show in a forthcoming review of the empirical literature since 1987, these predictions have generally been supported.

PILOT TEST OF THE THEORY

To provide a pilot test of our theory for this conference, we decided to conduct a small-scale study with graduate students. As mentioned, Rossiter and Percy's theory has received many "real world" tests, via our applied research colleagues, but necessarily under less controlled conditions.

For this test, we omitted the normal depletion motive. The reason for this is that the normal depletion motive, in our theory, is not a motivation for purchasing a brand initially. Rather, it is a motive that applies only to a brand-loyal target audience and it motivates simple re-stocking behavior; the brand must have been purchased originally on the basis of one of the other seven motives. For the seven motives, we chose a pair of print ads that, in our expert judgment, utilized the respective motives. One of each pair was for a likely low purchase-risk (low involvement) brand and the other was for a likely high purchase-risk (high involvement) brand. It should be noted that we did not pay any deliberate attention to emotional stimuli in the ads; however, we recognize the possibility of some self-serving selection here, perhaps operating unconsciously, in testing our own theory.

The seven pairs of ads, 14 in total, were presented in random order in two sets of seven in two class sessions, the high involvement ads in the first session and the low involvement ads in the second session. The experimental subjects, a class of 35 MBA students, were given booklets containing a randomized list of the emotion adjective pairs (for informational motives) and single emotional adjectives (for transformational motives). They were asked to look at a color transparency version of each ad on the screen, the color being important to-reproduce the emotional portrayal in the pictorial content of the ads in their original form (although several ads actually were black and white). Subjects also had black and white copies of the ads in their questionnaire booklets to which they could refer for the detailed copy. They were told to look at the ad and read the copy, then pick the adjective pair or adjective which "best describes how tie advertiser is trying to make you feel."

The first finding was that informational and transformational ads, in general, are emotionally distinguishable. For informational ads, subjects selected informational emotions on .79 (79%) of the trials (low involvement-informational, .79; high involvement-informational, .79; chance, .44; both Z = 2.50, p = .006). For transformational ads, subjects selected transformational emotions on .94 (94%) of the trials (low involvement-transformational, .99; high involvement-transformational, .89; chance, .56; Z = 6.43, p = .0001 and Z = 2.92, p = .002, respectively). This finding is important because it implies that emotion sequences (used to represent the negative-to-neutral informational motives) are accurately detected as compared with single emotions (used to represent the positive transformational motives) as suggested by our theory.

The second finding was that "correct" emotional detection for individual informational and transformational motives ranged considerably. Given a chance level of .11 for subjects' selection of the correct (by our judgment) emotion pair or single emotion, the results, which are averaged here for convenience over low and high involvement pairs of ads, were: problem removal, .53; problem avoidance, .20; incomplete satisfaction, .10; mixed approach-avoidance, .07; sensory gratification, .69; intellectual stimulation/mastery, .29; and social approval-conformity, .50. A confusion matrix, to be reported elsewhere, revealed that the three informational motives of problem avoidance (fear -> relaxation), incomplete satisfaction (disappointment -> optimism) and mixed approach-avoidance (guilt -> peace of mind) were often interchanged and that the transformational motive of intellectual stimulation/mastery (excited; competent) was often perceived as sensory gratification (sensory anticipation). Either we as expert judges see different motives in ads than do lay people (consumers!) or else we still have not found adequate emotional descriptors for four of the seven motives. To the extent that the latter is true, an approach in terms of types of emotions (Batra and Holbrook 1990), which we would have to complexly amend, for informational motives, to "sequences of types" (see discussion of their paper below), may have to be tried.

Overall, the first finding, of an informational versus transformational distinction in terms of emotional sequence types, is encouraging; but the second finding perhaps underscores the view that emotions are extremely discrete.

OTHER THEORETICAL AND APPLIED ISSUES

As this is a special topics forum, we take this opportunity to comment on the ideas of other theorists who are investigating the role of emotions in advertising. Also, we comment on some applications by advertising researchers.

Batra and Holbrook (1990) Study

Holbrook and Batra have been the two most productive advertising theorists working in the area of emotions in advertising. In a very important theoretical and empirical study (Batra and Holbrook 1990), they argued that, whereas conventional R-type factor analysis can be used (cf. Osgood 1955) to identify the dimensionality of emotions, a cluster analysis procedure (Cattell 1978) should thereafter be used to isolate types of emotions that then will have projections on the original R-type dimensions. A perhaps self-critical comment, since we also advocate a types-of-emotions theory, is that these investigators deleted single emotions if they did not correlate with a cluster of other emotions. It is possible that these "outliers" might be precisely the unique emotions that a creative person is seeking for a particular campaign. But our main criticism is as follows. Batra and Holbrook's theory is, of course, "motiveless." Consequently, it fails to identify sequences of emotions which we deem essential for informational (negatively reinforcing) motives. Moreover, in their empirical test of the typology, where the 12 emotional clusters were used to predict Aad, Ab, and purchase intention (PI) with very impressive results, it is quite possible that they inadvertently picked up sequences of emotions since the ads in the study were rated on multiple emotion types or clusters. The use of multiple emotion ratings of ads also raises a practical issue. In their predictive regression analysis, any of the 12 emotion types could have contributed to prediction of the dependent variables. In the real world, creatives want to know how well their particular advertising execution has tapped a particular emotion or sequence of emotions. In this sense, Batra and Holbrook's theory is rather ad hoc in terms of its implications for advertisers. A final comment on their important work is that these investigators typically ask respondents to rate how the ads used for the study made them (the respondents) feel. In our study, in contrast, we asked respondents to rate how the advertiser is trying to make them feel, that is, we used a more "projective" measure that did not require the respondent to actually experience the emotion. We agree with Batra and Holbrook that ads will be more effective in terms of influencing brand attitudes to the extent that the portrayed emotion or emotion sequence is subjectively experienced by the consumer during ad processing. However, as our brand attitude quadrant predictions (outlined earlier) suggest, subjective experience is much more critical for transformational motives. Indeed, for informational motives, which are more utilitarian, "cold emotion" (to coin a term which may be the opposite of Abelson's concept of "hot cognition") may be sufficient as long as the consumer cognitively anticipates the emotional sequence that would result if he or she were to buy and use the product. This latter supposition implies the concept of emotion memory (Arnold 1970; Posner and Snyder 1975, Bower 1981; Bower and Cohen 1982). The concept of emotion memory, and its corollary of vicariously or not-really-experienced emotions but rather ''imaged'' emotions during advertising exposure (Rossiter and Percy 1980; 1983), has obvious implications for Pavelchak, Gardner, and Broach's (1990) theory presented in this session.

Pavelchak, Gardner, and Broach (1990) Study

Pavelchak, Gardner, and Broach (1990) advocate a generalized arousal theory that implies the concept of "unitary arousal." Evidence suggests that generalized arousal is a rather empty construct (cf. the debate between Anderson 1990 and Neiss 1990). Neiss argues that emotions are discrete psychobiological states and that generalized arousal cannot distinguish, for example, the emotions of exhilaration, fear, anger, or sexual excitement. Also, evidence for the notion of "optimal level of arousal" at the individual level is completely lacking (Strongman 1987; Neiss 1990). Arousal of the physiological kind is not in itself sufficient for an emotional state to be present, as can be exemplified by arousal during hard physical exercise (Strongman 1987). In our theory, we see arousal as representing no more than the subjective intensity dimension (though we label it "A") which largely distinguishes discrete emotions within the same type--for instance, happiness intensifying into joy or elation, or, as in our pilot study, annoyance being a less intense form of anger. Thus, we would direct Pavelchak et al. away from the generalized arousal view toward arousal as represented by the subjective intensity dimension of discrete emotions. We would also suggest that a more fruitful perspective on individual differences, rather than optimal level of arousal, might be Eysenck's (1981) theory of anxiety and extraversion, whereby anxiety is related to chronic levels of ANS arousal and extraversion is related to chronic levels of CNS arousal (see also Rossiter and Percy 1987; and see Cetola and Prinkey 1986 for a successful test of the introversion-extraversion dimension of Eysenck's theory in an advertising setting). Pavelchak et al.'s theory, while it examines the P and A dimensions of emotions, neglects the D (or control) dimension which we believe to be essential. These authors also promote the misleading generalization (in their special topic proposal) that "with pleasure reactions, if an ad makes you feel good, you like it and, via transfer, like the brand as well." The "affect transfer" notion has been overgeneralized and is far too simplistic (Allen and Shimp 1990). We remind that our theory predicts, with considerable empirical support (Rossiter and Percy 1987, p. 242), that pleasant stimuli in ads will influence brand attitude via Aad only with low involvement-transformational advertising.

Yi (1990) Study

In a recent study, Yi (1990) provides evidence -that supports our emotions and motivations theory even though his stated conclusions from his data imply otherwise. Yi's experiment utilized magazine ads for a new, fictitious brand of automobile which, in our framework, would be a high involvement-informational choice (informational given the attributes focused on, which were safety and gas consumption). Positive and negative affect, respectively, were induced by having subjects read positively or negatively affectively-toned magazine articles prior to ad exposure. Induced affect did significantly affect Aad. But it affected Aad only; there was no significant carryover to either attitude toward the brand (Ab) or brand purchase intention (PI). This is precisely the result that Rossiter and Percy's (1987) theory would predict, given a high involvement-informational brand attitude. As a practical implication of his study, we contended in our 1987 book that media context hardly ever affects advertising effectiveness except under very extreme conditions (such as might obtain in laboratory studies). To our knowledge, there is no field study evidence to support the widespread notion that media context makes a difference (Rossiter and Percy 1987; Rossiter 1988; Appel 1987). Our recommendation for advertisers is that "a good ad will work anywhere." Laboratory studies with often extreme induced affect and forced exposure do little to counter this recommendation.

Left Brain-Right Brain Theory

We are also on record (Rossiter 1980) as being entirely skeptical of the pursuit of cortical location of emotions or of ad processing in general in the right or left hemispheres of the brain. Not only are such cortical responses difficult to interpret (Olson and Ray 1983) but the whole pursuit presumes a physiological reductionism that is not adequate to explain emotions (Strongman 1987). In a practical sense, who cares where emotions may occur as long as they do occur? Cognitive theorists would also maintain that it is the subject's report of, and labeling of, an emotional occurrence that is determinant rather than whether there is indeed any physiological correlate (Strongman 1987). We will be very surprised if physiological measures of advertising effectiveness produce anything valid or practically worthwhile. For instance, Rothschild and Hyun's recent (1990) demonstration that EEG patterns may predict visual ad recognition is of little practical value. If you can measure visual ad recognition directly (e.g., BRC research service; Rossiter and Percy 1987, pp. 561, 587), then why bother measuring EEG?

Picture Sorts as Measures of Emotions

Based on the twin assumptions that emotions are accurately reflected through facial expression (Darwin 1878; Ekman 1965; Ekman and Oster 1979; Izard 1972) and that verbal measures are inadequate for measuring emotional experiences, several advertising agencies have developed "picture sorts" of faces representing various emotions to measure consumers' emotional responses to advertisements. For instance, FCB has developed VIP, Visual Image Profile, and ICON, Image Configurations, which show faces and situations respectively (Ratchford 1987) and BBDO has developed EPD, Emotional Photo Deck, which shows faces (Russell and Starkman 1989). In our opinion, photo sorts are another questionable pursuit. In the first place, the photo sort methodology asks consumers to pick the face that best represents how they feel after viewing the commercial (the photo sort method appears to have been mainly used with TV commercials rather than with ads in other media). This after-only measure would miss the emotional sequence that is incorporated in many ads and thus would have limited predictive validity as well as limited diagnostic value. It would, of course, be possible to ask consumers to select several photographs, to represent a sequence of emotions should such a sequence be intended in the commercial, but this has not been done. Secondly, there is a fundamental fallacy inherent in the picture sort methodology. The method of validating whether various pictures accurately represent various emotions has universally been to correlate picture sorts with verbal (adjectival) descriptions of emotions. This raises the question of: why bother switching to pictures at all? Picture sorts are more time-consuming as an applied advertising research task than merely checking off adjectives and no one has yet shown that verbal checklists of emotions are in any way inadequate. Indeed, in Rossiter and Percy (1987, pp. 209-212), we cited an extensive study by Sweeney, Tinling and Schmale (1970) demonstrating that the same emotion can be accurately portrayed and interpreted in virtually any medium. By "media" here we mean the types of stimuli available to creative people to elicit emotions, namely: heard words and sound effects, music, seen words, pictures, color, and movement--which are themselves obviously constrained by the actual advertising medium employed (Rossiter and Percy 1987, pp. 209-210). As we have seen in our pilot study, picking the right emotional adjective can be almost as difficult as it must be to pick accurate pictorial expressions of an emotion. However, it seems a lot easier and more practical to use verbal descriptors than to employ pictures, except perhaps when the subjects are pre-school children (Rossiter 1977). From a practitioner's standpoint, we suggest that picture sorts will turn out to be just another faddish sales gimmick used by advertising agencies under the guise of proprietary research techniques.

A-B-E Model of Benefit Claim Selection

The authors are working on a new model of benefit claim selection in advertising which was stimulated positively by an article by Moberg (1988) and negatively by our dissatisfaction with the "laddering" approach developed by Reynolds and Gutman (1984). Whereas the laddering approach probably taps emotions in relation to motives as advocated by our theory, the approach is silent as to which level on the "ladder" a particular benefit claim in an ad should be pitched at for maximum effect on brand attitude. We conceptualize three basic levels (assuming initially that each is tied to the same motive) which can be described as: the physical attribute (A),- the subjective benefit or consequence resulting from that attribute (B); and emotional responses (E) which can be either the emotional consequence of perceiving the attribute (A) or of experiencing the benefit ( 3) or a completely independently inserted emotion as suggested by classical conditioning or by the "peripheral cue" concept (Petty and Cacioppo 1986). The current Zeitgeist among advertising agency practitioners, which is also implied by laddering theory, is that effective ads must focus on the emotional consequence (E) of benefits. This is a unitary (and therefore wrong) model for at least two reasons. Firstly, emotional consequences, and indeed emotional reactions in general, are often specific and idiosyncratic--especially under high involvement conditions (recall Rossiter and Percy's 1987 contention that high involvement amplifies emotions and leads to individuals', or individual segments', "authenticities" being necessary). Only in low involvement-transformational advertising, such as for Coca-Cola, can one hope to focus on an emotional consequence that is almost universally the same for everyone in the audience. However, even Coca-Cola uses at least two emotional approaches as witnessed by the "high energy" creative idea used in the summer commercials versus the "heavy pathos" creative idea used in the Mean Joe Green commercial. An alternative benefit claim strategy would be to focus instead on the attribute (A) level and let the audience "read in" their own emotional consequences. In this way, for varied emotional "reasons," a broader audience may be targeted by the ad. Of course, there is probably a tradeoff here between audience size and persuasive effectiveness in moving from the A end of the continuum to the E end (Lautman and Percy 1984). However, we note that many low involvement-informational and high involvement-informational ads--such as retail store newspaper ads announcing sales and specials--appear to be successful with hardly any emotional content in the benefit claim other than that it tries to elicit the mildly affectively-laden belief of "That's good value." Which level, or even multiple levels, of A B or E to focus on when designing benefit claims for ads is a complicated decision that cannot be solved by any unitary theory as is popular in advertising at present and indeed may resolve to a purely empirical question that can only be answered by constructing conceptually alternative A-B-E versions of ads and subjecting them to a formal ad test with the target audience, or target audiences if a broad target with less persuasion (an A ad version) is being tested against a narrow target with more persuasion (a B or E ad version). Which "level" of benefit claim to employ in their A-B-E sense is one of the most frequent questions asked in advertising practice.

SUMMARY

We (Rossiter and Percy 1987 and-this paper) have proposed a theory in which emotional stimuli, to be effective in influencing brand attitude, must serve an underlying purchase motivation. The theory proposes eight types of purchase motivations capable of energizing the consumer to act by offering either negative reinforcement or positive reinforcement. Our theory further hypothesizes that particular types of emotional sequences have to be included in the ad, with the sequence for informational (negative) motives being from negative-to-neutral and the emotional sequence for transformational (positive) motives being from neutral-to-positive. Empirical support for the two emotional sequences looks promising but the task of selecting efficient emotional descriptors for the specific motives remains challenging.

REFERENCES

Aaker, D.A., D.M. Stayman, and M.R. Hagerty (1986), "Warmth in Advertising: Measurement, Impact and Sequence Effects," Journal of Consumer Research, 12 (December), 365-381.

Allen, C.T. and T.A. Shimp (1990), "On Using Classical Conditioning Methods for Researching the- Impact of Ad-Evoked Feelings," in Emotion in Advertising: Theoretical and Practical Explorations, S. Agres, J.A. Edell, and T.M. Dubitsky, eds., Westport, CT: Quorum Books.

Anderson, K.J. (1990), "Arousal and the Inverted-U Hypothesis: A Critique of Neiss's 'Reconceptualizing Arousal'," Psychological Bulletin, 107 (January), 96-100.

Appel, V. (1987), "Editorial Environment and Advertising Effectiveness," Journal of Advertising Research, 27 (August/September), 11-16.

Arnold, M.B. (1970), Feelings and Emotions: The Loyola Symposium, New York: Academic Press.

Batra, R. and M.B. Holbrook (1990), "Developing a Typology of Affective Responses to Advertising," Psychology & Marketing, 7 (Spring), 11-25.

Bearden, W.O. and R.L. Rose (1990), "Attention to Social Comparison Information: An Individual Difference Factor Affecting Consumer Conformity," Journal of Consumer Research, 16 (March), 461-471.

Bower, G.H. (1981), "Mood and Memory," American Psychologist, 36 (January), 129-148.

Bower, G.H. and P.R. Cohen (1982), "Emotional Influences in Memory and Thinking: Data and Theory," in Affect and Cognition, M.S. Clark and S.T. Fiske, eds., Hillsdale, NJ: Lawrence Erlbaum.

Cattell, R.B. (1978), The Scientific Use of Factor Analysis in Behavioral and Life Sciences, New York: Plenum.

Cetola, H. and K. Prinkey (1986), "Introversion-Extraversion and Loud Commercials," Psychology & Marketing, 3 (Summer), 123-132.

Darwin, C.R. (1872), The Expression of Emotions in Man and Animals, London: Murray.

Davitz, J.R. (1969), The Language of Emotion, New York: Academic Press.

Donovan, R.J. and J.R. Rossiter (1982), "Store Atmosphere: An Environmental Psychology Approach," Journal of Retailing, 58 (Spring), 34-<7.

Edell, J. and M.C. Burke (1987), 'The Power of Feelings in Understanding Advertising Effects," Journal of Consumer Research, 14 (March), 421433.

Ekman, P. (1965), "Differential Communication of Affect by Head and Body Cues," Journal of Personality and Social Psychology, 2 (October), 726-735.

Ekman, P. and H. Oster (1979), "Facial Expression of Emotion," Annual Review of Psychology, 30, 527-554.

Estes, W.K. and B.F. Skinner (1941), "Some Quantitative Properties of Anxiety," Journal of Experimental Psychology, 29 (December), 390400.

Eysenck, H.J. (1981), A Model For Personality, New York: Springer.

Fennell, G. (1975), "Motivation Research Revisited," Journal of Advertising Research, 15 (June), 23-28.

Fennell, G. (1978), "Consumers' Perceptions of the Product Use Situation," Journal of Marketing, 42 (April), 3847.

Fennell, G. (1989), "Action Vs. Attitude: Motivation Makes the Difference," in Proceedings: Society for Consumer Psychology, D.W. Schumann, ed., Knoxville, TN: University of Tennessee.

Fishbein, M. and I. Ajzen, (1975), Belief, Attitude, Intention, and Behavior: An Introduction to Theory and Research, Reading, MA: Addison-Wesley.

Hammond, L.J. (1970), "Conditioned Emotional States," in Physiological Correlates of Emotion, P. Black, ed., New York: Academic Press.

Izard, C.E. (1977), Human Emotions, New York: Plenum.

Katz, D. (1960), "The Functional Approach to the Study of Attitudes," Public Opinion Quarterly, 24 (April), 163-204.

Lautman, M.R. and L. Percy (1984), "Cognitive and Affective Responses in Attribute-Based Versus End-Benefit Oriented Advertising," in Advances in Consumer Research: Vol. 11, T. Kinnear, ed., Ann Arbor, MI: Association for Consumer Research.

Lutz, R.J. (1978), "A Functional Approach to Consumer Attitude Research," in Advances in Consumer Research: Vol. 5, H.K. Hunt, ed., Ann Arbor, MI: Association for Consumer Research.

Mehrabian, A. (1980), Basic Dimensions for a General Psychological Theory, Cambridge, MA: Oelgeschlager, Gunn, and Hain.

Mehrabian, A. and J.R. Russell (1974), An Approach to Environmental Psychology, Cambridge, MA: MIT Press.

Millenson, J.R. (1967), Principles of Behavioral Analysis, New York: Macmillan.

Moberg, G.D.11988), "Strategy Testing: To Execute or Not to Execute?" Marketing News, January 4, p. 30.

Mowrer, O.H. (1960a), Learning Theory and Behavior, New York, Wiley.

Mowrer, O.H. (1960b), Learning Theory and the Symbolic Processes, New York: Wiley.

Neiss, R. (1990), "Ending Arousal's Reign of Error: A Reply to Anderson," Psychological Bulletin, 107 (January), 101-105.

Olson, J.C. and W.J. Ray (1983), "Using Brain-Wave Measures to Assess Advertising Effects," Working Paper, Marketing Science Institute, Cambridge, MA.

Osgood, C.E. (1955), "Fidelity and Reliability," in Information Theory in Psychology, H. Quastler, ed., Glencoe, I.L.: Free Press.

Pavelchak, M.A., M.P. Gardner, and V.C. Broach (1990), "Ad-Induced Arousal: Effects on Ad and Brand Attitudes," Abstract of Paper for Special Topic Session, 1990 ACR Conference.

Ratchford, B.T.(1987), "New Insights About the FCB Grid," Journal of Advertising Research, 27 (August/September), 24-38.

Reynolds, T.J. and J. Guttman (1984), "Laddering: Extending the Repertory Grid Methodology to Construct Attribute-Consequence-Value Hierarchies," in Personal Values in Consumer Psychology, R. Pitts and A. Woodside, eds., Lexington, MA: Lexington Books.

Rossiter, J.R. (1977), "Reliability of a Short Test Measuring Children's Attitudes Toward TV Commercials,'' Journal of Consumer Research, 3 (March), 179-184.

Rossiter, J.R. (1980), "Point of View: Brain Hemisphere Activity," Journal of Advertising Research, 20 (October), 75-76.

Rossiter, J.R. (1988), "The Increase in Magazine Ad Readership," Journal of Advertising Research, 28 (October/November), 35-39.

Rossiter, J.R. and L. Percy (1980), "Attitude Change Through Visual Imagery in Advertising," Journal of Advertising, 9 (2), 10-16.

Rossiter, J.R. and L. Percy (1983), "Visual Communication in Advertising," in Information Processing Research in Advertising, R.J. Harris, ed., Hillsdale, NJ: Lawrence Erlbaum.

Rossiter, J.R. and L. Percy (1985), "Advertising Communication Models," in Advances in Consumer Research, E.C. Hirschman, and M.B. Holbrook, eds., Provo, UT: Association for Consumer Research.

Rossiter, J.R. and L. Percy (1987), Advertising and Promotion Manage ne, It, New York: McGraw-Hill.

Rossiter, J.R., L. Percy, and R.J. Donovan (1984), 'The Advertising Plan and Advertising Communication Models," Australian Marketing Researcher, 8 (December), 744.

Rossiter, J.R., L. Percy, and R.J. Donovan (1990), "A Better Advertising Planning Grid," Working Paper, Australian Graduate School of Management, University of New South Wales.

Rothschild, M.L. and Y.J. Hyun (1990), "Predicting Memory for Components of TV Commercials from EEG," Journal of Consumer Research, 16 (March), 472-478.

Russell, D.A. and D.L. Starkman (1989), "Measuring the Emotional Response to Advertising," Reclame En Onderzoek, 1 (April), 15-26.

Russell, J.A. and G. Pratt (1980), "A Description of the Affective Quality Attributed to Environments," Journal of Personality and Social Psychology, 38 (August), 311-322.

Skinner, B.F. (1938), The Behavior of Organisms, New York: Appleton-Century-Crofts.

Strongman, K.T. (1987), The Psychology of Emotion, 3rd edn., Chichester, UK: Wiley.

Sweeney, D.R., D.C. Tinling, and A.H. Schmale (1970), "Dimensions of Affective Expression in 4 Expressive Modes," Behavioral Science, 15 (5). 393407.

Wagner, A.R. (1969), "Frustrative Non-reward: A Variety of Punishment," in Punishment and Aversive Behavior, B.A. Campbell and R.M. Church, eds., New York: Appleton-Century-Crofts.

Wells, W.D. (1981), "How Advertising Works," Working Paper, Needham, Harper & Steers Advertising, Chicago.

Yi, Y. (1990), "Cognitive and Affective Priming Effects of the Context for Print Advertisements," Journal of Advertising, 19 (2), 4048.

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