1=2: When a Singular Experience Leads to Dissociated Evaluations

Size: px
Start display at page:

Download "1=2: When a Singular Experience Leads to Dissociated Evaluations"

Transcription

1 HONEA, DISSOCIATED MORALES, EVALUATIONS FITZSIMONS JOURNAL OF CONSUMER PSYCHOLOGY, 16(2), Copyright 2006, Lawrence Erlbaum Associates, Inc. 1=2: When a Singular Experience Leads to Dissociated Evaluations Heather Honea San Diego State University Andrea C. Morales Arizona State University Gavan J. Fitzsimons Duke University This research examines how a single experience with a salesperson can lead to the formation of dual representations of the salesperson resulting in opposing spontaneous and deliberative affective responses and dissociated evaluations. Consumers may either use their spontaneous affective reactions to form an evaluation or may respond more deliberately using information that contradicts this initial affect. As a result, they can hold 2 evaluations of the same salesperson one that is more spontaneously generated and one that is more deliberate. The 2 cognitive bases for these evaluations can coexist in memory even when they are opposite in valence, and consumers switch back and forth between the 2 evaluations, depending on the level of cognitive activity in which they engage. George Orwell s (1977) novel 1984 linked the concept of doublethink the power of holding two contradictory beliefs in one s mind simultaneously, and accepting both of them to life in a mind-controlling totalitarian society. Such dire circumstances, however, are not required for an individual to hold at least two alternate versions of something in his or her mind. Consumers might feel negatively toward a company that collects data about their spending habits because their privacy has been invaded. At the same time, they might appreciate the company s knowledge of their tastes and service they receive. Likewise, diners might have positive reactions to a delicious dessert when they focus on taste but negative reactions to the same dessert when they consider their health. Previously, researchers assumed that such conflicts pit heart against mind (Shiv & Fedorikhin, 1999). However, consumer preferences are not simply a function of whether emotion wins over cognition. Even when consumers rely on affect to perform evaluations (e.g., Pham, 1998), contradictions arise because lower-order (spontaneous) versus higher-order (controlled or specific) affective responses can Correspondence should be addressed to Heather Honea, College of Business Administration, San Diego State University, San Diego, CA hhonea@mail.sdsu.edu result in very different judgments (Giner-Sorolla, 1999; Shiv & Fedorikhin, 2002). When spontaneous versus deliberative processes generate conflicting reactions to the same object, consumers may end up with contradictory mental representations of that object (Brendl, Markman, & Messner, 2003). And if level of processing impacts which representation is interpreted, it may also determine whether consumers spontaneously use affect as a basis for judgment independent of subsequent affective information that calls the implications of this initial affect into question. Thus, consumers may not only make decisions on the basis of information they have previously discredited, but may also switch back and forth between contradictory positions. This means that impulsive purchase decisions could have entirely different outcomes than those made with deliberation, even when consumers have recently analyzed their preferences. Moreover, products, salespeople, or companies that generate two contrasting representations may create schizophrenic consumers whose evaluations are inconsistent from 1 min to the next. Although consumer research has documented divergent evaluations of a single object (Brunel, Tietje, & Greenwald, 2004), the possibility that a consumer can have contradictory evaluations and accept both of them has not been directly addressed in the marketing literature.

2 DISSOCIATED EVALUATIONS 125 A number of studies have verified that spontaneous and deliberative attitudes can be quite contradictory (e.g., Greenwald & Banaji, 1995). Some research has even bolstered the concept of dual attitudes, theorizing that spontaneous and deliberate evaluations of the same object can coexist in memory (Wilson, Lindsey, & Schooler, 2000). Work on the psychology of attitudes, however, has not considered the relation of mental representations and their associated affective reactions with regard to disassociation. If representations of an object are formed at lower versus higher levels of processing and consequently are applied independently at the time of judgment (Wyer & Srull, 1989), this could lead to dissociated evaluations. Previous research has indicated that two independent representations of an object can be formed and sampled independently for use in subsequent judgments (Carlston, 1980). Kardes (1986), for instance, found that consumers retrieve initial stimulus-based versus attribute-based representations of a product independently. This indicates that individuals earlier reactions to a product may or may not influence their subsequent evaluations. However, neither of these studies addressed how level of processing determines which representation influences judgment. Nor did they connect affective reactions associated with these representations to evaluations or consider the circumstances under which earlier affect may influence subsequent evaluations. This research explores the potential for a single object to generate dual representations that result in conflicting affective reactions and considers the circumstances under which initial affective reactions continue to impact evaluation. In conceptualizing the influence that different representations can have on judgments, we made the following assumptions. First, an encounter with an object may result in the formation of a mental representation. If this object generates a socially ingrained, spontaneous affective reaction, this affect will become associated with the representation. Later, the individual may need to use this information in a manner he or she did not anticipate, such as making a judgment regarding the object. If at that time his or her level of processing is constrained, he or she will simply rely on that representation and its associated affect to make a judgment. When processing is not constrained, however, the individual may have the opportunity to retrieve that representation and construe its implications in the context of additional situational information. This deliberate, cognitive activity may result in the formation of a second representation in which the evaluative implications differ from those of the first. These two representations, although formed at different times, may coexist in memory. Second, the level of processing will determine which representation influences judgment. If an individual has to make a quick judgment, the first representation is likely to be retrieved and the affect it elicits may provide the basis for the judgment. Although the second representation might also be retrieved, its implications for judgment take time to compute. If an individual has sufficient time to perform these computations, the effects of these implications could override the effects of the affective reaction elicited by the first representation alone and be used as a basis for judgment. If processing time is limited, however, the affective reactions elicited by the first representation may dominate. In other words, the second representation and its associated affective reactions are only applied to judgment when the opportunity to think about the judgment is fairly unrestricted. If restrictions exist, individuals are likely to apply the first representation to subsequent judgments, even if the second representation is accessible. To evaluate these possibilities in a consumer context, we considered the different interpretations that consumers might give to a salesperson s behavior (e.g., whether the salesperson s attention to them is manipulative or sincere) and the different evaluations that result from this interpretation. We identify an event that is likely to elicit a spontaneous positive reaction based on its social value; namely, a salesperson exerting extra effort in serving a customer. This reaction, in turn, becomes associated with a global feature of the salesperson. If individuals are later asked to make a quick judgment about the salesperson, they may base their evaluation on this representation and its associated affect. However, suppose customers have time to consider the motives that underlie the salesperson s behavior and have the objective of forming an impression. Then they might perceive the exertion of effort to be manipulative, and might form a second representation that contains the implications of this perception. This latter representation may elicit mistrust. We show that even though this second representation has negative reactions that drive judgment when processing is unconstrained, the representation that is formed on the basis of an individual s initial spontaneous reaction continues to drive judgment when time constraints exist. Thus, if dual impressions of an object are formed, not only do the situational conditions surrounding an evaluation determine which impression informs judgment, but individuals may actually switch back and forth between conflicting judgments, depending on the situation at hand. SPONTANEOUS AND DELIBERATIVE PROCESSES Evaluation Spontaneous evaluations classify stimuli as good or bad without requiring extensive thinking (Chen & Bargh, 1999). Under some circumstances, however, this initial evaluation can be overridden by more deliberative thought processes. Such deliberative processes require time and ability, and they can serve to correct spontaneous judgments by incorporating the implications of additional situational information (Gilbert, 2002). Johar and Simmons (2000), for example, found that even if consumers pay attention to a product disclosure, the disclosure will only influence their opinion if they have sufficient time to think about it. Similarly, Camp-

3 126 HONEA, MORALES, FITZSIMONS bell and Kirmani (2000) found that when consumers have limited time and mental resources available to process information, they perceive a sales agent s flattery as sincere. When they are able to deliberate on the motives underlying the behavior, however, they perceive it as insincere and update their judgments accordingly. In certain situations, however, deliberative inferences may not actually correct initial evaluations. For instance, studies have examined race- or gender-based stereotypes that were supposedly replaced by more positive or egalitarian views. However, these views continued to hold power when people were not consciously monitoring their responses (Banaji & Greenwald, 1995; Banaji, Hardin, & Rothman, 1993; Devine, 1989; Devine, Plant, Amodio, Harmon-Jones, & Vance, 2002). This suggests that circumstance determines whether implicit (spontaneous) or explicit (deliberative) evaluations are applied to an object (Devine, Monteith, Zuwerink, & Elliot, 1991). Dissociated Evaluations Studies of both implicit and explicit evaluation measures often find inconsistent correlations between the two (e.g., Dovidio, Kawakami, Johnson, Johnson, & Howard, 1997). This discrepancy is labeled dissociation, because when spontaneous and deliberative measures are uncorrelated, they tend to predict qualitatively different behaviors and preferences (Greenwald & Banaji, 1995). For example, Caucasians show a spontaneous preference for advertisements that feature a Caucasian over an African American spokesperson, although their deliberative attitudes toward each spokesperson are equivalent. African Americans, in contrast, display a more positive deliberative attitude for advertisements featuring a African American over Caucasian spokesperson although their spontaneous preferences for each spokesperson are similar (Brunel et al., 2004). The lack of correlation between these preferences suggests that dissociated attitudes toward a single object coexist (Wilson et al., 2000). Dissociation also appears to be a possibility in the formation of attitudes toward newly encountered individuals. Wilson, Lindsey, and Anderson (1998 as cited in Wilson, Lindsey, & Schooler, 2000) examined whether individuals can erase an initial attitude that was generated in an experimental setting once they determine that their attitude is inaccurate. Although individuals who are not under time pressure reported corrected attitudes, individuals under time pressure did not. In addition, the latter individuals responses to a set of implicit measures suggested that they were still relying on their noncorrected attitude. Affect and Dual Representations Dissociated evaluations require some form of duality during encoding or interpretation of information, or both. This indicates that dissociated evaluations may be driven by the presence of dual representations. Dual representation is implied by the Wyer and Srull (1986, 1989) bin model. The model suggests that information can be interpreted at different levels of abstractness and that processing will determine the nature of the representation that is formed. Lower-level or more general encoding will occur relatively spontaneously at the initial comprehension stage of processing and a representation of this information may be formed on the basis of this encoding. A second, higher-level encoding may take place at a later stage of processing, depending on the objectives that exist at the time, and an additional representation might be formed on the basis of this encoding. Thus a representation formed at one level of processing may be available independently of the representations formed at other levels (Wyer & Srull, 1986). As a result, spontaneous or preconditioned affective reactions may be associated with a representation that is separate from the representation that results from deliberative processing of the available information. All things being equal, the most recently formed representation should be most accessible (Wyer & Srull, 1986). Nevertheless, level of processing can also determine which representation and its associated affect informs judgment. Devine s (1989) two-stage model of evaluation suggested that individuals initial responses to certain objects are a function of their social experience. Thus, a representation of an object that is consistent with a stereotype or based on a common social interpretation (or both) can be processed immediately with minimal cognitive effort, thereby motivating spontaneous evaluations. Because such processing involves minimal cognitive activity, the affective reactions that result are likely to be relatively generalized positive or negative responses (Cohen & Areni, 1991; Leventhal, 1980). At a later point, in the second stage of evaluation, individuals can either rely on their initial response or, if they are motivated, focus on a more detailed representation that resulted from deliberation. An evaluation of the implications of this latter representation will generally take time and resources. However, these more controlled cognitive processes will result in more discriminated affective reactions or specific emotions (Frijda & Zeelenberg, 2001). This suggests that when thinking is constrained, individuals are likely to rely on a representation in which the implications can be processed quickly (e.g., a representation that spontaneously elicits affective reactions), and the implications of this representation will inform judgments (Gilbert & Hixon, 1991; Wilson et al., 2000). When individuals have an opportunity to deliberate, however, they are likely to assess the implications of a representation that differs from those of the first, and the affective responses that result from this more deliberative assessment may inform their judgments. As a result, when a deliberate representation has implications that differ from those of its spontaneous counterpart, different levels of processing can result in different judgments.

4 DISSOCIATED EVALUATIONS 127 In summary, we propose that certain objects often generate dual representations, each of which may have different implications for judgments. On one hand, a spontaneous affective response can become associated with a characteristic of an object through learning, and this reaction can also become associated with the representation that is formed. In addition, more deliberate responses to the same characteristic might generate a different representation with implications that contradict those of the spontaneous reaction. Furthermore, the implications of this second representation for judgment require time to evaluate. Thus, the use of a representation and its associated affective reactions that inform judgment depend on the level of processing at the time of judgment. As a result, as level of processing changes, individuals may switch between judgments. Alternative Conceptualizations Although a dual-representation conceptualization is consistent with previous work in impressions and judgment (e.g., Adaval & Wyer, 2003; Kardes, 1986), alternative possibilities should be considered. Theoretically, dissociated evaluations could result in the absence of any mental representations of an object. For example, every piece of encoded information about the object might have its own memory trace (Hintzman, 1986). Then, if certain encoded pieces of information have different implications, judgments might be determined by whatever comes to mind most easily. Dissociation could also result if information is organized into a single mental representation. Theory and research in person memory (Srull & Wyer, 1989), for example, assumes that the representation of a person consists of a central evaluative concept to which a number of specific behaviors are peripherally linked. Some of these behaviors, however, may have implications that are evaluatively inconsistent with the concept with which they are associated, and may often be easier to recall than the consistent ones (Srull & Wyer, 1989). When processing time is limited, individuals might use the central concept of this representation as a basis for judgment. When they have more time, however, they might consider features of the associated behaviors and thus make judgments that differ from those they might make on the basis of the central concept alone. Although these alternatives might account for the phenomena to be investigated, the dual-representation conception seems viable on a priori grounds. The dual-representation conceptualization can be tied specifically to degree of processing in which the implications of a spontaneously formed representation can be interpreted and applied immediately, whereas the implications of a deliberatively formed representation require processing and are only understood when there is sufficient time. This parallels the fact that level of processing impacts the type of information used in consumer judgments (e.g., Campbell & Kirmani, 2000; Johar & Simmons, 2000). For the purposes of this research we explore the implications of a dual-representation conceptualization without specifically precluding the possibility of the no-representation or single-representation alternatives. These alternative conceptualizations will be evaluated further after the results are reported. Dissociation in Salesperson Judgment Salesperson effort represents a referent characteristic that can elicit a highly positive generalized spontaneous reaction. People who grow up in a community that emphasizes the value of hard work are likely to have a spontaneous positive reaction to effort. Exerting effort is viewed as a valuable trait, and it is generally rewarded regardless of how that effort pays off (Cameron, 1989; Haas & Deseran, 1981; Struthers, Weiner, & Allred, 1998; Weiner, 1980; Weiner, Heckhausen, & Meyer, 1972). Consequently, consumers may typically have more positive evaluations of salespeople who put forth more effort to market their products, regardless of the actual quality of the products. However, this impression can be altered depending on how consumers feel about the motives or intentions surrounding the effort (Morales, 2005). That is, if people take time to think about the salesperson s intentions and conclude that they are probably self-serving, their impressions are likely to be negative. To examine the implications of dual representations for reactions to the effort exerted by a salesperson, we ran two experiments. Experiment 1 establishes that when individuals experience a single incident that has multiple implications, they can experience contradictory responses to that incident based on level of processing. Experiment 2 demonstrates that these contradictory responses are not necessarily integrated into an overall judgment because they are actually driven by different representations. More important, the impact of initial affect does not necessarily abate. In Experiment 2, we introduced constraints on the time to make a judgment to influence which representation would drive consumers responses and influence judgments. Regardless of the order in which judgments were measured, there was a difference between individuals judgments when they were under time constraints than when they were not. EXPERIMENT 1 Experiment 1 was designed to demonstrate that consumers can experience both spontaneous and deliberate responses to a salesperson based on the salesperson s effort, and that these responses inform subsequent judgments of the salesperson. This study established that salesperson effort can elicit contradictory spontaneous and deliberate responses. That is, it can either elicit a generalized positive reaction or the more specific negative reaction of mistrust. To evaluate this possibility before receiving information about the salesperson s effort, some participants were ex-

5 128 HONEA, MORALES, FITZSIMONS posed to information that activated the concept that business activities can be somewhat underhanded. Other participants were not exposed to this material. We expected that all participants would experience a spontaneous, positive response to the salesperson s effort. Therefore, we predicted that when participants were under time pressure, this initial, positive reaction to effort should inform their evaluation, and so their evaluations would be quite favorable regardless of whether suspicion was primed. However, we expected that when participants were not under time pressure, priming suspicion would lead them to attribute the salesperson s effort to self-serving motives, leading to lower evaluations of the salesperson than would otherwise be the case. Method Participants. A total of 77 undergraduate students participated in the study as partial fulfillment of a course requirement. Experiment 1 was a 2 (Effort: High vs. Low) 2 (Processing Conditions: Constrained vs. Unconstrained) between-subjects design, and it was run in a behavioral lab using a Visual Basic computer program. Stimuli and procedure. At the beginning of the session, participants were informed that they would be participating in two separate computer-based research activities and would be completing a series of ostensibly unrelated tasks. Furthermore, they were informed that to simulate reality, some aspects of these tasks might have time restrictions. (In fact, these restrictions were only imposed under constrained processing conditions.) With this preamble, participants were told they would be using a computer to read several business industry articles and would be asked to answer questions about each article. They then read two filler articles and one target article. The target article was designed to prime suspicion by discussing ethical issues in pricing. Specifically, it indicated that some businesses engage in questionable behaviors and practices when it comes to pricing. Consistent with the guise for the first part of the study, participants were asked at the end of each article to respond to questions regarding the industry that was discussed in the article. The questions were general inquiries regarding the content of the article unrelated to the suspicion prime. After completing the first set of questions, participants were introduced to the second activity. Participants were asked to imagine that they were going shopping for a Mother s Day present for their mom. They were told that they had decided to buy her a pair of earrings from a local jewelry store. On arriving at the store, they see a pair that they think their mom might like. They ask to look at the earrings and describe their mother s general style to the salesperson. In the high-effort condition, the salesperson was described as having spent 50 min with them discussing the artist who created the earrings, the meaning behind them, and how they would look on their mom. In the low-effort condition, the salesperson was said to have spent only 5 min describing the same things. All participants were then told to assume that they had been in the store for 1 hr in total and had to make a decision. Measures. To assess consumer judgment, participants after reading the scenario were asked whether they would recommend the salesperson to a friend. Furthermore, one half of the participants were told they had only 5 sec to respond to the question. To emphasize the time limit, the screen blinked as each second passed and showed the remaining seconds in bright red numerals in the center of the screen. This time-constrained measure has been used in previous research to impact level of processing (e.g., Wilson, Hodges, & LaFleur, 1995) and captures the impact of participants spontaneous reactions on judgment (Wilson et al., 2000). Participants in the unconstrained processing condition had 1 full min to respond to the same question and this measure served as the deliberative judgment measure. In addition to providing their judgments of the salesperson, participants were also asked to report the specific affective responses they experienced when thinking about the salesperson in the store. A series of positive and negative affect terms were compiled from affect-related articles and embedded in this list were two items intended to measure feelings of mistrust specifically distrust and suspicion. All items were measured on a sliding bar scale, with the cursor positioned at the midpoint. The endpoints not at all and to a great degree were subsequently translated to a1to100 scale. Results The percentage of participants who recommended the salesperson under each experimental condition is shown in the top half of Table 1. Results of a 2 (Effort: High vs. Low) 2 (Processing Conditions: Constrained vs. Unconstrained) logistic regression show that when participants were under time pressure, they were more likely to recommend the salesperson to a friend when the salesperson exerted a great deal of effort (89.5%) than when the salesperson had not (65.0%), χ 2 (1, N TABLE 1 Recommendation of Salesperson and Feelings of Mistrust as a Function of Level of Processing: Experiment 1 Salesperson Effort Measure and Processing Condition High Effort Low Effort Percentage recommending salesperson Constrained processing condition 89.5 a 65.0 b Unconstrained processing condition 20.0 b 61.1 c Feelings of mistrust Constrained processing Unconstrained processing a n = 19. b n = 20. c n = 18.

6 DISSOCIATED EVALUATIONS 129 = 77) = 6.2, p =.01. In the absence of time constraints, however, they were substantially less likely to recommend the salesperson in the former condition than the second (20% vs. 61.1%), χ 2 (1, N = 77) = 3.0, p =.08. The interaction implied by these data was significant, χ 2 (1, N = 77) = 8.52, p =.004. To explore the role of specific affective responses, a combined measure of mistrust was created by averaging the distrust and suspicion responses from the affect item list (r =.81). As shown in the bottom half of Table 1, participants reported greater mistrust when the time to respond was not constrained (M = 63.0) than when it was constrained (M = 50.2), F(1, 73) = 8.71, p =.004, and greater mistrust when the salesperson exerted effort (M = 61.9) than when the salesperson did not (M = 51.30), F(1, 73) = 5.95, p =.02. As expected, however, the effect of effort was only evident when judgment time was unconstrained (73.3 vs. 52.7, respectively); when participants were under time pressure, the effect of the salesperson s effort on judgments was negligible (50.5 vs. 49.9). The interaction of Time Pressure Salesperson Effort was significant, F(1, 73) = 5.33, p =.02. The final goal of this study was to establish that there is a relation between the specific affective responses generated and the deliberate evaluation formed of the salesperson. That is, feelings of mistrust were expected to mediate the effect of effort on salesperson judgments in the unconstrained processing conditions. Because there was no difference between the high- and low-effort mistrust responses in the constrained processing condition, a mediation analysis was only performed on the two cells in the unconstrained processing condition. When mistrust was used as a covariate in an analysis of salesperson evaluations in these conditions, the effect of this covariate was significant, χ 2 (1, n = 38) = 4.10, p =.04, and accounted for.21 of the variance. However, the previously significant effect of effort on salesperson judgments disappeared (p >.10), and the proportion of variance accounted for by the effort variable was reduced from.17 to.05. Thus, feelings of mistrust mediate the effect of effort on salesperson judgments. Discussion This study confirms that individuals indeed have different responses to effort depending on level of processing. When processing was constrained, individuals respond positively to high levels of effort. However, when they are primed with suspicion and deliberate on high levels of effort they appear motivated to generate more discrete negative emotions related to mistrust. The mediation analysis indicates that feelings of mistrust inform deliberative judgments. Thus, we can confirm that when processing is unconstrained, individuals rely on more specific negative affective reactions to make their judgments, whereas when processing is constrained, individuals appear to rely on their initial generalized positive reactions. The study demonstrates that people report different judgments of salesperson effort depending on processing conditions and as a result a different representation may be formed at lower levels of processing than at higher levels. However, the between-subjects design does not indicate the implications when dual representations are actually available. Experiment 2 investigates these matters. EXPERIMENT 2 Experiment 1 established that a salesperson exerting effort can elicit a spontaneous positive affective reaction, but when individuals are able to deliberate on that effort, they are likely to generate more specific affective responses. The spontaneous reactions that have become associated with the salesperson may be elicited merely by mentioning his or her name, independent of other information about him or her. If this happens, and if individuals have to make a quick judgment, they should base their judgment on their reactions to the salesperson s name alone. If individuals have time to deliberate, however, they will be able to perform the computations required to assess the implications of the representation that carries information related to salesperson motives, and they will base their evaluation on this representation instead. If deliberation results in positive feelings such as appreciation, the representation that is formed should have similar implications for judgments of the salesperson, even if it is independent of the representation that is formed on the basis of spontaneous reactions to his or her behavior. If deliberation results in specific negative feelings, however, evaluations will be higher when they are based on the representation that elicits spontaneous positive affect than when they are based on the representation that was formed through deliberate cognitive processing. To produce a difference between a judgment based on spontaneous responses to effort and a judgment based on a representation that requires more time to evaluate, we asked some participants to perform an ostensibly unrelated task that primed concepts associated with suspicion after receiving the effort information. Other participants were primed with neutral concepts unrelated to suspicion. Information that the salesperson has expended high effort should result in the spontaneous activation of a positive response regardless of which prime is applied. Furthermore, in the absence of priming concepts associated with suspicion, deliberation on the implications of this information is likely to generate specific, positive responses that are consistent with the generalized, spontaneous response. As a result, judgments are likely to be similar regardless of whether spontaneous or specific responses dominate. When participants are primed with suspicion, however, the opportunity to deliberate may lead them to interpret the salesperson s behavior unfavorably, leading to a representation that has negative implications. Consequently, judgments based on this representation should be more negative than corresponding judgments by participants who received the neutral prime.

7 130 HONEA, MORALES, FITZSIMONS We used a timed response task to measure spontaneous judgments and an untimed rating task to measure deliberate judgments. Furthermore, we varied the order in which these judgments were made. We assumed that if separate representations toward the same object differentially inform judgments, judgments made under different processing conditions should differ regardless of which evaluation is measured first. Method Participants. A total of 154 undergraduate students participated in the study to fulfill a course requirement. Experiment 2 was a 2 (Effort: Low vs. High) 2 (Prime: Neutral vs. Suspicion) 2 (Measure Order: Spontaneous Judgment First vs. Deliberative Judgment First) mixed design, and was run in a behavioral lab using a Visual Basic computer program. Stimuli and procedure. At the beginning of the session, participants were informed that they would be participating in a set of studies investigating visual and verbal comprehension in computer-mediated environments, and as a consequence, they would be completing a series of seemingly unrelated tasks. 1 In the first part of the computer program, participants were trained on a timed good bad response task used to measure spontaneous judgments. They were told that they would be required to participate in this same task several times during the session and that this first section was designed to allow them to practice the basic response. Single words were presented on the computer screen and participants were asked to indicate whether they thought the word was good or bad. They were asked to be as accurate and as fast as possible by pressing the appropriate key. Participants were then told to place their left index finger on the 1 key and their right index finger on the 0 key. They were instructed to press 1 if the word they saw was a good or positive word and 0 if it was a bad or negative word to them. After completing the training session, participants were asked to imagine the same Mother s Day scenario used in Experiment 1. The only difference was that in this scenario, the specific name Morgan was used to refer to the salesperson so that we could use this name in the good bad spontaneous judgment measure in a manner to be described. 2 Once participants had finished reading the scenario, they were given the priming task. We expected that a spontaneous reaction toward the salesperson would be generated simply on reading the scenario and would be associated with the global features of the salesperson (i.e., the salesperson s name). However, motivation is required to make the attributions that elicit specific affective responses. As a result, we assumed that participants would not form this second representation until they were asked to make a judgment. In response to this request, we predicted that participants would retrieve the initial episode and, during their deliberation, would apply any concepts that appeared to be relevant at the time. This cognitive activity would lead them to form a separate representation. However, this representation should influence their later judgments only if they have time to assess its implications. 3 If participants level of processing is once again constrained, they should rely on their first representation and its associated spontaneous affective reactions when making a judgment. The supraliminal prime was accomplished using a scrambled-sentence completion task (Srull & Wyer, 1979). In line with the cover story that the set of studies investigated visual and verbal comprehension in computer-mediated environments, participants were told that this particular study would be used to examine their linguistic and cognitive abilities. The instructions told them that they would see five lists each containing a total of six words on the screens that followed, and that at least one sentence could be formed using five of the words. They were asked to rearrange five of the words into a sentence, ignoring the sixth word, and to write the sentence underneath the list of words in the box provided. Participants in the suspicion-prime condition saw lists containing these target words: suspicious, shady, dubious, distrust, and wary. Participants in the neutral-prime condition saw lists containing neutral target words such as these: healthy, bright, vegetable, and scary. An example of a sentence used in the priming task is stones suspicious moss no rolling gather, where the constructed sentence is rolling stones gather no moss and the excluded target word, suspicious, is the prime. In some lists, the target word was the sixth word and was ignored, whereas in other lists the target word was actually included in the five words used to create the sentence. After completing the scrambled sentence-priming task, participants completed the spontaneous and deliberative judgment measures. Across subjects, the order of the spontaneous and deliberative judgment measures was rotated, with each measure appearing first one half of the time. For the spontaneous measure, we used the good bad timed response task on which participants trained at the beginning of the session. Although they were given a long list of words, including several other proper names, the actual spontaneous mea- 1 A pretest of the experimental design confirmed that participants found the explanation for differential activities in the study very consistent with the guise that the research purpose was to examine visual and verbal comprehension in computer-mediated environments. 2 Morgan was selected because it did not appear in the 1990 Census most-common name listing and because it did not indicate a specific gender for the salesperson. 3 Some research on impression formation has indicated that once information is encoded, individuals do not reinterpret or recode the originally encoded information in terms of concepts that are accessible at the time of making a judgment (Srull & Wyer, 1980). However, in cases when the purpose of information is introduced after the information is received and encoded, rendering it inadequate for making a judgment, individuals may engage in recoding based on concepts that are salient at the time. This was the case in this study.

8 DISSOCIATED EVALUATIONS 131 sure was how they responded to the name Morgan. Participants with a favorable judgment of the salesperson presumably responded with a 1 when they saw the name Morgan, whereas those with a less favorable judgment responded with a 0. For the deliberate judgment measure, participants were asked to respond to an untimed rating question about Morgan. Specifically, they were asked to give the salesperson, Morgan, a rating on a sliding bar scale defaulted at the midpoint, with endpoints very low to very high (subsequently translated to a rating from 1 100). In addition to these measures, participants were asked to respond to questions regarding the specific affective responses they experienced in thinking about Morgan while they were shopping. The series of valenced terms was the same as the series used in Experiment 1, with items embedded in the list that referred to feelings of mistrust. All these items were measured on a sliding bar scale and were subsequently translated to a 1 to 100 scale. Results The proportions of positive responses to Morgan in the timed judgment task were analyzed using logistic regression as a function of priming conditions, salesperson effort, and measurement order. Data relevant to this analysis are summarized in the top half of Table 2. The effects of salesperson effort did not depend on measurement order (χ 2 < 1). That is, participants showed the same pattern of responses to the name Morgan regardless of whether they generated these responses before or after the rating task. Pooled over order conditions, participants were more likely to respond positively to Morgan in the high effort condition (76.7%) than in the low effort condition (61.0%), χ 2 (1, N = 154) = 4.29; p <.05. Furthermore, this difference did not depend on whether suspicion was primed (73.7% vs. 60.5%) or not (79.5% vs. 61.5%), χ 2 < 1, and did not depend on whether responses were assessed before the spontaneous measure (81.6% vs. 63.4%) or afterward (71.8% vs. 58.3%), χ 2 < 1. The main effects for order and prime are also insignificant. The effects of priming and salesperson effort on deliberative responses showed a different pattern, as indicated in the bottom half of Table 2. In neutral prime conditions, participants rated the salesperson more favorably when the effort expended was high (M = 67.2) than when it was low (M = 56.4), F(1, 150) = 4.95, p <.05. When participants were primed with suspicion, however, the reverse was true; that is, they rated the salesperson less favorably when the effort expended was high than when it was low (50.3 vs. 60.2), F(1, 150) = 4.03, p <.05. The interaction of Priming Salesperson Effort was significant, F(1, 150) = 8.95, p <.01, and did not depend on measurement order (F < 1). To explore the role of more specific affective responses, distrust and suspicion were averaged to form an indicator for feeling mistrust (r =.75). Mistrust was low in the neutral prime condition regardless of whether the salesperson s effort was high (M = 38.6) or low (M = 44.0), F < 1. When suspicion was primed, however, mistrust was greater when the salesperson expended high effort than when the salesperson expended low effort (54.7 vs. 39.5, respectively), F(1, 150) = 8.70, p <.01. The interaction of Priming Salesperson Effort was significant, F(1, 150) = 8.10, p <.01. We expected the deliberative judgment (salesperson rating) to be mediated by the mistrust evoked by high effort. In fact, when mistrust was used as a covariate in the initial analysis, its effect was significant, F(1, 149) = 37.2, p <.001, and accounted for.18 of the variance. Furthermore, the previously significant interaction between the Prime Effort was reduced to nonsignificance, and the proportion of variance it accounted for was reduced from.05 to.01. Thus, feelings of mistrust mediate the effect of priming suspicion and salesperson effort on salesperson ratings. Discussion Consistent with predictions, spontaneous responses to the salesperson, as inferred from the timed response task, increased with the salesperson s effort level. This was true regardless of whether participants were primed with suspicion. In contrast, untimed judgments of the salesperson only increased with the effort the salesperson expended as long as participants were not stimulated to feel concerned about the salesperson s motivation. When participants were primed TABLE 2 Spontaneous and Deliberate Evaluations of the Salesperson as a Function of Priming, Salesperson Effort, and Measurement Order: Experiment 2 Deliberate Measure First Spontaneous Measure First Measure and Priming Condition High Effort Low Effort High Effort Low Effort Percentage spontaneous positive reaction Suspicion prime 68.2 a 56.3 b 81.3 b 63.6 a Neutral prime 76.5 c 60.0 d 81.8 a 63.2 e Deliberate evaluations Suspicion prime Neutral prime a n = 22. b n= 16. c n = 17. d n = 20. e n = 19.

9 132 HONEA, MORALES, FITZSIMONS with suspicion, they made lower ratings of the salesperson when he or she had exerted extra effort than when he or she had not. The difference in evaluations for the neutral- versus suspicion-prime conditions can be attributed to the difference in feelings of mistrust. No differences in mistrust were evident in the neutral prime condition. When suspicion was primed, however, feelings of mistrust were much higher when the salesperson had exerted extra effort than when he or she had not. In addition to demonstrating a different pattern of results for spontaneous versus deliberative judgments in the suspicion-prime condition, the lack of order effects in our analyses indicates that initial affect can continue to influence evaluations independent of more deliberate information that calls this initial affect into question. Regardless of whether individuals report their spontaneous or deliberate judgment first, subsequent judgments still appear to be a function of level of processing. Specifically, the reporting of a spontaneous judgment displays the same pattern whether it precedes or follows the reporting of a deliberate judgment. Note that if only one representation had been constructed and used as a basis for judgment, such order effects are unlikely to be observed. Specifically, if participants based their initial judgments on responses to information that is consistent with social learning but updated their subsequent judgments based on responses to information inconsistent with social learning, under time pressure individuals would not revert back to their initial judgments. This was not the case. Specific affective responses only impacted judgments when level of processing was high. This suggests that level of processing not only drove differential responses to a single incident but this single incident was most likely stored as separate representations. GENERAL DISCUSSION Dual representation appears to drive contradictory responses to salesperson effort. Individuals experience a spontaneous, positive response to salesperson effort that can inform their judgment of a salesperson. When sufficient processing resources are available, however, individuals are able to undertake the cognitive computations necessary to elicit more specific valenced responses. Therefore, judgments informed by these responses are not necessarily consistent with spontaneous responses. However, even when individuals have reported a deliberate judgment, they still revert back to their spontaneous judgment when the level of processing is low. Such responses suggest that separate representations of the salesperson exist and can generate contradictory reactions. It is important to note that when spontaneous and deliberate judgments are inconsistent, individuals do not reconcile this apparent conflict. The dual representations that inform evaluations coexist, and depending on circumstances, individuals rely on one representation or the other, and its associated reactions, when forming judgments. This study also suggests that it is not only possible, but appropriate, to draw a specific connection between primary and secondary affective responses and dissociated evaluations. Even when specific affective responses are elicited that contradict spontaneous affect, spontaneous affect may continue to impact judgments. Theoretically, this indicates a particular uniqueness to judgment. Certain contextual factors may be better able to impact spontaneous versus deliberate judgments because their influence operates at a primary versus secondary level. If this is so, the key to influencing judgment may be quite complex. Changing specific affective responses may do little to change judgments informed by spontaneous processes (e.g., Shiv & Fedorikhin, 2002). And even if a situation dictates that a more deliberative judgment drives preference, consumers may switch back to their more spontaneous judgment as soon as processing conditions change and a different representation is driving their response. In fact, this switching back and forth between deliberative and spontaneous judgments may be unavoidable so long as information with a common social interpretation or learned cultural reaction is available. Alternative explanations of our findings should be considered. First, individuals might not form separate representations, but might encode and store each piece of information separately; thus leading every piece of information to have its own memory trace (Hintzman, 1986). In this case, participants would presumably base their judgments on whatever piece of information comes to mind most easily. It is unclear, however, how this conceptualization would account a priori for the effects of levels of processing that we observed in Experiment 2. Second, information might be organized into a single mental representation consisting of a central favorable concept of the salesperson to which peripheral concepts of the person s behavior (in some cases, perceived as manipulation) are attached (Srull & Wyer, 1989). Judgments might then depend on whether the central concept or its peripheral associates was identified and used. However, the fact that the spontaneous and deliberative judgments observed in Experiment 2 were independent of the order in which these judgments were made is difficult to explain on the basis of this conceptualization. That is, once a judgment was formed that incorporated evaluatively inconsistent information, it is not obvious why an individual would switch back to the use of the central concept instead of integrating the inconsistent information and the central concept into a new evaluation. Thus, although this conceptualization cannot be completely discredited, the dual-representation conceptualization we propose provides the most parsimonious a priori account of our results. Future research would benefit from expanding the domain of inquiry. Although our work focused on responses to effort, a number of other firm salesperson actions, marketing stimuli, and product features might also have the potential to generate dissociated consumer re-

10 DISSOCIATED EVALUATIONS 133 sponses. For instance, the fact that consumers may have contradictory evaluations and accept both of them could potentially account for situations in which consumers buy products without thinking, despite knowledge that the products are inferior or pose dangers to their health. If consumers have spontaneous positive reactions to products that have been discredited, such behavior is likely to occur. This research focuses on a subset of specific negative responses contrasted to a general positive initial response. It will also be important to show similar effects when a spontaneous response is negative and more intentional responses are positive. Work in the area of stigmas, for instance, suggests that two psychological systems may be involved in reactions to a stigma one of which generates an immediate negative reaction and another of which generates a more positive reaction in individuals who delay their response to the stigmatized stimulus (Pryor, Reeder, Yeadon, & Hesson- McInnis, 2004). Additionally, our research indicates that dual representations lead individuals to switch back and forth between judgments. However, we were not able to unequivocally verify distinct systems. Implementation of technology such as Functional Magnet Resonance Imaging (fmri) is increasingly able to examine the existence of separate systems (Lieberman, Jarcho, & Satpute, 2004), and this may be an important tool in verifying distinct psychological systems such as those we have presented. The connection between spontaneous and deliberative responses and judgment may provide insights regarding dual coding in theories of memory and the role of affective information in judgment. Carlston s (1980) dual-coding theory suggested that individuals may store two independent representations of a stimulus, and that each judgment can be sampled independently. He found that an inference based on an assessment computed from initial information becomes more accessible than any spontaneous evaluation. However, this finding only holds in the case of a single-implication episode (e.g., an episode implies an individual is either unkind or honest), but not in the case of a multiple-indication episode (a single episode implies both unkindness and honesty). A multiple-indication episode represents a circumstance that more closely mirrors the effort episode in this article. One could speculate that an inference based assessment is not necessarily more interpretable than a spontaneous evaluation when an episode has multiple implications. This may be particularly true if the representation that informs the spontaneous evaluation represents a common social interpretation and the inference assessment is inconsistent with this common social interpretation. Perhaps dissociated representations are formed and level of processing determines the representation applied to judgment. Such issues would benefit from further exploration. In other research, Levine, Wyer, and Schwarz (1994) identified conditions in which descriptive judgment-relevant information took precedence over affect as a basis for judgment when it was highly accessible in memory, but affect took precedence in global judgments of self-esteem. Perhaps certain assessments, such as those related to self-esteem, display similar effects to certain representations formed in response to effort (representation containing information related to a common social interpretation) in that both are quite ingrained and habitual. Clearly, additional examination of the relation between higher- versus lower-order responses, consumer memory, and assessments could be a fruitful extension to this research. The application of dual representation in the consumer context also has the potential to provide important insights regarding the types of responses that might impact attitude change and drive dual attitudes in the consumer context. Consider a case in which consumers change their attitude in response to a persuasive message, but eventually revert back to an originally held attitude. Is it possible that the message generated a new representation, but that the representation containing the old attitude and responsible for generating a spontaneous reaction was unaffected? Alternatively, if aspects of a message or a product elicit spontaneous affective responses (because they are conditioned based on social learning), an associated attitude may be persistent, although it came about through a peripheral route. Even when more deliberate processes generate explicit attitudes, these attitudes may fail to impact preference and behavior if consumers do not have sufficient processing capacity at the time of judgment or choice. Finally, future research should continue to investigate the conflict underlying dissociated evaluations. The examination of duality in consumer judgment will benefit from better knowledge of the contextual factors that form representations and elicit spontaneous and deliberative responses, and the manner in which these responses interact and direct evaluations. This will promote our understanding of the roles that common social interpretations and social learning play in consumer doublethink. More important, it will provide additional insight regarding whether consumers are really wired to maintain some level of mental contradiction through dual representations. ACKNOWLEDGMENTS We would like to thank Patti Williams, Cristel Antonia Russell, and Jaideep Sengupta for their helpful comments on earlier drafts of this article. REFERENCES Adaval, R., & Wyer, R. S. (2003). Communicating about a social interaction: Effects on memory for protagonist s statements and nonverbal behaviors. Journal of Experimental Psychology, 40, Banaji, M. R., & Greenwald, A. G. (1995). Implicit gender stereotyping in judgments of fame. Journal of Personality and Social Psychology, 68,

11 134 HONEA, MORALES, FITZSIMONS Banaji, M. R., Hardin, C., & Rothman, A. J. (1993). Implicit stereotyping in person judgment. Journal of Personality and Social Psychology, 65, Brendl, C. M., Markman, A. B., & Messner, C. (2003). The devaluation effect: Activating a need devalues unrelated objects. Journal of Consumer Research, 29, Brunel, F. F., Tietje, B. C., & Greenwald, A. G. (2004). Is the implicit association test a valid and valuable measure of implicit consumer social cognition? Journal of Consumer Psychology, 14, Cameron, S. (1989). The unacceptability of money as a gift and its status as a medium of exchange. Journal of Economic Psychology, 10, Campbell, M. C., & Kirmani, A. (2000). Consumers use of persuasion knowledge: The effects of accessibility and cognitive capacity on perceptions of an influence agent. Journal of Consumer Research, 27, Carlston, D. E. (1980). The recall and the use of traits and events in social inference processes. Journal of Experimental Social Psychology, 16, Chen, M., & Bargh, J. A. (1999). Consequences of automatic evaluation: Immediate behavioral predispositions to approach or avoid the stimulus. Personality and Social Psychology Bulletin, 25, Cohen, J. B., & Areni, C. S. (1991). Affect and consumer behavior. In T.S. Robertson & H. H. Kassarjian (Eds.), The handbook of consumer behavior (pp ). Englewood Cliffs, NJ: Prentice Hall. Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and controlled components. Journal of Personality and Social Psychology, 56, Devine, P. G., Monteith, M. J., Zuwerink, J. R., & Elliot, A. J. (1991). Prejudice with and without compunction. Journal of Personality and Social Psychology, 60, Devine, P. G., Plant, A. E., Amodio, D. M., Harmon-Jones, E., & Vance, S. L. (2002). The regulation of explicit and implicit race bias: The role of motivations to respond without prejudice. Journal of Personality and Social Psychology, 82, Dovidio, J. F., Kawakami, K., Johnson, C., Johnson, B., & Howard, A. (1997). On the nature of prejudice: Automatic and controlled processes. Journal of Experimental Social Psychology: Special Issue on Unconcious Processes in Stereotyping and Prejudice, 33, Frijda, N. H., & Zeelenberg, M. (2001). Appraisal: What is dependent? In K. R. Scherer, A. Schorr, & T. Johnstone (Eds.), Appraisal processes in emotion (pp ). New York: Oxford University Press. Gilbert, D. T., (2002). Inferential correction. In T. Gilovich, D. Griffin, & D. Kahneman (Eds.), Heuristics and biases: The psychology of intuitive judgment (pp ). Cambridge, England: Cambridge University Press. Gilbert, D. T., & Hixon, J. G. (1991). The trouble of thinking: Activation and application of stereotypic beliefs. Journal of Personality and Social Psychology, 60, Giner-Sorolla, R. (1999). Affect in attitude: Immediate and deliberative perspectives. In S. Chaiken & Y. Trope (Eds.), Dual-process theories in social psychology (pp ). New York: Guilford. Greenwald, A. G., & Banaji, M. R. (1995). Implicit social cognition: Attitudes, self-esteem, and stereotypes. Psychological Review, 102, Haas, D. F., & Deseran, F. A. (1981). Trust and symbolic exchange. Social Psychology Quarterly, 44, Hintzman, D. L. (1986). Schema abstraction in a multiple-trace memory model. Psychological Review, 93, Johar, G. V., & Simmons, C. J. (2000). The use of concurrent disclosures to correct invalid inferences. Journal of Consumer Research, 26, Kardes, F. R. (1986). Effects of initial product judgments on subsequent memory-based judgments. Journal of Consumer Research, 13, Leventhal, H. (1980). Toward a comprehensive theory of emotion. Advances in Experimental Social Psychology, 13, Levine, S. R., Wyer, R. S., & Schwarz, N. (1994). Are you what you feel? The affective and cognitive determinants of self-judgments. European Journal of Social Psychology, 24, Lieberman, M. D., Jarcho, J. M., & Satpute, A. B. (2004). A dual-process model of reactions to perceived stigma. Journal of Personality and Social Psychology, 87, Morales, A. C. (2005). Giving firms an E for effort: Consumer responses to high-effort firms. Journal of Consumer Research, 31, Orwell, G. (1977) San Diego, CA: Harcourt Brace Jovanovich. Pham, M. T. (1998). Representativeness, relevance, and the use of feelings in decision making. Journal of Consumer Research, 25, Pryor, J. B., Reeder, G. D., Yeadon, C. Y., & Hesson-McInnis, M. (2004). A dual-process model of reactions to perceived stigma. Journal of Personality and Social Psychology, 87, Shiv, B., & Fedorikhin, A. (1999). Heart and mind in conflict: The interplay of affect and cognition in consumer decision making. Journal of Consumer Research, 26, Shiv, B., & Fedorikhin, A. (2002). Spontaneous versus controlled influences of stimulus-based affect on choice behavior. Organizational Behavior and Human Decision Processes, 87, Srull, T. K., & Wyer, R. S. (1979). The role of category accessibility in the interpretation of information about persons: Some determinants and implications. Journal of Personality and Social Psychology, 37, Srull, T. K., & Wyer, R. S. (1980). Category accessibility and social perception: Some implications for the study of person memory and interpersonal judgment. Journal of Personality and Social Psychology, 38, Srull, T. K., & Wyer, R. S. (1989). Person memory and judgment. Psychological Review, 96, Struthers, C. W., Weiner, B., & Allred, K. (1998). Effects of causal attributions on personnel decisions: A social motivation perspective. Basic and Applied Social Psychology, 20, Weiner, B. (1980). A cognitive (attribution)-emotion-action model of motivated behavior: An analysis of judgments of help-giving. Journal of Personality and Social Psychology, 39, Weiner, B., Heckhausen, H., & Meyer, W. -U. (1972). Causal ascriptions and achievement behavior: A conceptual analysis of effort and reanalysis of locus of control. Journal of Personality and Social Psychology, 21, Wilson, T. D., Hodges, S. D., & LaFleur, S. J. (1995). Effects of introspecting about reasons: Inferring attitudes from accessible thoughts. Journal of Personality and Social Psychology, 69, Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A model of dual attitudes. Psychological Review, 107, Wilson, T. D., Lindsey, S., & Schooler, T. Y. (2000). A model of dual attitudes. Psychological Review, 107, Wyer, R. S., & Srull, T. K. (1986). Human cognition in its social context. Psychological Review, 93, Wyer, R. S., & Srull, T. K. (1989). Memory and cognition in its social context. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Received: July 26, 2003 First revision received: February 21, 2004 Second revision received: April 28, 2005 Accepted: July 26, 2005

12

13