The Dual Role of Emotions in Consumer Satisfaction/Dissatisfaction

H. Shanker Krishnan, Indiana University
Richard W. Olshavsky, Indiana University
ABSTRACT - Prior studies on consumption emotions have not made a distinction between emotions experienced directly during consumption and emotions evoked during the evaluation of these directly experienced emotions. This distinction is particularly relevant since a negative emotion (e.g., fear) directly experienced during consumption can evoke a positive emotion (e.g., joy) if a negative emotion was desired (e.g., for horror movies, rollercoaster rides). We present a model that attempts to capture this dual role of emotions on satisfaction. Results from two exploratory studies (a focus group and a lab experiment) provide initial support for the model.
[ to cite ]:
H. Shanker Krishnan and Richard W. Olshavsky (1995) ,"The Dual Role of Emotions in Consumer Satisfaction/Dissatisfaction", in NA - Advances in Consumer Research Volume 22, eds. Frank R. Kardes and Mita Sujan, Provo, UT : Association for Consumer Research, Pages: 454-460.

Advances in Consumer Research Volume 22, 1995      Pages 454-460


H. Shanker Krishnan, Indiana University

Richard W. Olshavsky, Indiana University


Prior studies on consumption emotions have not made a distinction between emotions experienced directly during consumption and emotions evoked during the evaluation of these directly experienced emotions. This distinction is particularly relevant since a negative emotion (e.g., fear) directly experienced during consumption can evoke a positive emotion (e.g., joy) if a negative emotion was desired (e.g., for horror movies, rollercoaster rides). We present a model that attempts to capture this dual role of emotions on satisfaction. Results from two exploratory studies (a focus group and a lab experiment) provide initial support for the model.


Current models of satisfaction have explicitly considered the role of "consumption emotion" patterns as predictors of satisfaction (e.g., Mano and Oliver 1993; Oliver 1993; Oliver and Westbrook 1993; Westbrook 1987). "Consumption emotion refers to the set of emotional responses elicited specifically during product usage or consumption experiences, as described either by the distinctive categories of emotional experience and expression (e.g., joy, anger, and fear) or by the structural dimensions underlying emotional categories, such as pleasantness/unpleasantness, relaxation/action, or calmness/excitement" (Westbrook and Oliver, 1991, p. 85). The general finding is that distinct patterns of affective experiences are systematically related to different levels of a unidimensional satisfaction response. Also, Oliver (1993) found that attribute satisfaction affects both emotions and overall satisfaction.

These studies make an important contribution to our understanding of the role of emotions in satisfaction. However, a distinction has not been made between emotions experienced directly during the consumption episode, and emotions evoked during the evaluation of this directly experienced emotion. Making a distinction between these two types of emotions is particularly relevant for hedonic attributes (Holbrook and Hirschman 1982). Whereas utilitarian attributes seem to be based on relatively objective product features such as fluoride content and miles per gallon, hedonic attributes, by definition, imply a direct affective experience. For example, consumers may experience excitement and interest during a music concert, and evaluate the experience as "joyful." Moreover, the valence of these two types of affect may be different. For instance, consumers may be extremely frightened (a negative emotion) during a rollercoaster ride but their evaluation of this experience may be joyful (a positive emotion), since they wanted to be frightened. These direct affective experiences may provide particularly strong influences on overall satisfaction for products/services that are primarily hedonic in nature.

The purpose of our paper is to further explore these two distinct types of emotions that occur at consumption and to examine their specific roles in understanding overall satisfaction. First we present a model that explicates the dual role of emotions in consumption and their separate effects on satisfaction responses. Then we present exploratory data from two studies that provide some initial support for these two separate roles of emotions for hedonic products. Finally, we discuss implications for future research in this area.


Building on the constructs expressed in Westbrook and Oliver (1991), overall satisfaction is viewed as the postchoice evaluation judgment concerning the purchase of a specific product or service. Overall satisfaction is modeled as a unidimensional affective concept. As shown in the Figure, following Oliver (1993), overall satisfaction is determined by the cumulative impact of satisfaction experienced on each of the salient attributes (possibly in proportion to their importance weights). Although not depicted in the Figure, overall satisfaction may also be determined, in part, by preexisting or concurrent affective states such as mood.

Consistent with prior research, we suggest that consumers approach consumption experiences with desires (Di) and expectations (Ei) pertaining to each attribute. For hedonic attributes, consumers may describe both desires and expectations in terms of specific emotions ("I want to see an exciting movie") and the corresponding level ("I think this concert will be extremely boring"). Expectation is assumed to influence perceived performance (only). Perceived performance (PPi) refers to the individual's perceptions of how the product performed on that attribute. Whenever the performance of a hedonic attribute is "perceived," one or more emotions (EM1) are automatically evoked (Mano and Oliver 1993). Individual attributions (A1) on the origin of these emotions moderate the specific types of emotions experienced, and also moderate the effects of these emotions on satisfaction. For example, if a rollercoaster ride operator fails to secure the safety restraint then PP may be negative and other-oriented, with emotions of anger, disgust, or contempt experienced; if fastening the safety restraint was your responsibility, then PP may be negative and self-oriented with emotions of shame and guilt evoked (following Oliver 1993). These moderated emotions represent the first of the two hypothesized roles of emotions in determining satisfaction.

After experiencing a product, a comparison process takes place between desire and perceived performance (Olshavsky and Spreng 1989; Spreng and Olshavsky 1993; see Woodruff et al, 1983 for other comparison standards). This comparison process is cognitive in nature and is labeled evaluated performance (EPi); [Spreng and Olshavsky (1993) refer to this as "desires congruency."] it essentially tells the individual whether they "got what they wanted." For hedonic attributes, the comparison process leads to a second emotion (EM2) based on the individual's evaluation of the (earlier) directly experienced emotion; evoked emotions may be positive, negative or of mixed valence. Again, individual attributions (A2) moderate the type of emotions experienced, and also moderate the effects of these emotions on satisfaction. Note that these second emotions and the corresponding attributions are distinct from those that occur directly during consumption. This second role of emotions in determining satisfaction appears consistent with prior treatments of this concept. [Cognitive responses to utilitarian and hedonic attributes are also expected to occur and to impact satisfaction, but these influences are not addressed in this paper.]



Satisfaction at the level of each attribute (Si) is thus determined by the moderated emotions that are generated directly from consumption (i.e., EM1) and by the moderated emotions that are evoked following the evaluation of the directly generated emotions (i.e., EM2). As noted earlier, the emotion directly experienced and the emotion evoked during the evaluation of performance may be different in type and even opposite in valence. For example, some consumers are extremely frightened while flying. While most of these consumers will have a negative emotional response to this frightening experience, some consumers may have a positive emotional response to this frightening experience (because they desire to be frightened). Also, many consumers seem to greatly enjoy fearful experiences such as horror movies, some types of amusement park rides, and bungee jumping.

As one source of complexity, some consumption experiences may lead to such strong direct emotions (EM1) that these emotions overwhelm the emotions that arise from the comparison process (EM2) on one or more of the other attributes (hedonic or utilitarian). For example, the tremendous fear that some consumers experience (EM1) while flying could overwhelm any other emotions elicited by the comparison process (EM2) on attributes such as staff friendliness, food quality, and meeting flight schedule times. Results of a study reported in the Marketing News (Vol. 28, No. 17, August 15, 1994) illustrate the above idea; results show that among shoppers in an upscale, enclosed mall in Miami the group that felt "unsafe...appeared so concerned about its safety that only 27.9% liked the atmosphere/decor of the mall. In contrast, 43.4% of the safe group liked the atmosphere."

An important aspect of our model is that we explicitly suggest that perceptions of product performance and comparisons of product performance to desires are based on salient product attributes (Oliver 1993), and that these attributes can be categorized as either utilitarian or hedonic, or a mix of the two types (Holbrook and Hirschman 1982; Mano and Oliver 1993). Recognition of the dual emotional responses to hedonic attributes is crucial to our model. But, note that if all of the attributes are utilitarian in nature, the only emotions that will impact satisfaction are those evoked in response to evaluations of the utilitarian attributes (EM2).


Study 1 had two goals: first, to determine whether consumers are able to identify their emotions at various stages of the consumption experience, and second, whether experienced emotions are different from evaluated emotions in terms of valence. In order to explore how consumers evaluate emotions from hedonic products, two focus groups were conducted (with seven and ten students). Undergraduate students recruited from marketing classes were offered course credit for participation. The moderator was a graduate student with prior experience in conducting depth interviews. Two relevant experiential services (rollercoaster rides and horror movies) and a set of questions were deemed satisfactory based on a pre-test.

After a general introduction in which the session was described as a discussion of consumption experiences with products/services, each session was broken up into two parts so that the discussion would focus on each of the two experiential categories in turn. In order to make a specific experience highly salient, the moderator asked each participant to name the "scariest" rollercoaster ride (horror movie) they had experienced. Participants were instructed to answer subsequent questions using this particular experience episode. Initial discussion focused on when they participated in these activities. Subsequently the focus turned to the participants' thoughts/feelings during the consumption experience. The discussion was structured around the following areas: expectations, desires, perceived performance, evaluated performance, and overall satisfaction.

Study 1 Results

The focus group results indicated that participants were able to relate their thoughts and feelings about all of the areas of inquiry. Although the experiences were sometimes a few months old (or more in some cases), participants had vivid memories, since the experiences had prompted intense feelings. Participants described their expectations based on their own earlier experiences with the same or similar activities and based on observations of other consumers engaging in these activities (e.g., "I think I was expecting to feel sick and just terrified."). Participants described their desires relevant to engaging in these activities in terms of emotions such as fear, thrills, arousal, and excitement (e.g., "I wanted to feel scared and out of control because the roller coaster, you really can't C it's not like you can tell them to stop and you get off...").

Importantly, participants were also able to describe the specific emotions they felt during the consumption experience (i.e., perceived performance). Illustrative comments are presented in Table 1. In general, participants' experienced emotions corresponded most closely with the fear, joy, and interest emotions from Izard's affect taxonomy (1977). Unexpected events led to the emergence of surprise as another emotional element. The intensity of the emotion varied during the experience episode, and seemed to depend partly on believability. Compared to horror movies, the rides seemed to be more "real" in terms of their potential to induce intense emotions.

Participants were also able to describe their emotions as they evaluated the consumption experience (i.e., evaluated performance). They evaluated the movie in terms of their desires (e.g., "it wasn't scary enough"). Illustrative comments are shown in Table 1. An outcome relevant to our theoretical perspective, was that participants evaluated felt negative emotions in positive terms. The rollercoaster ride may have been intensely scary, but participants walked out of it with the evaluation that it was fun (joyful). It is important to note that these are not akin to mixed emotions which generally refer to distinct emotions that are experienced concurrently. Rather, participants seemed to be referring to one set of emotions during consumption that gave rise to other emotions during evaluation.

Finally, participants were able to describe their satisfaction levels (see Table 1). Satisfaction seemed to depend upon evaluated performance (i.e., whether or not the ride or movie was as scary as was desired). In most cases, participants indicated considerable satisfaction; in some cases they were willing to stand in line (again) for two hours or more to repeat the experience or to rent other movies that were sequels to the original horror film. But, if the movie (or ride) was not scary enough, the dissatisfaction with these attributes would lead to overall negative evaluations of the movie (ride). Finally, of considerable interest, the direct emotions experienced by some of our participants were so negative and intense (i.e., the effects were strong and enduring) that they never wanted to see another horror movie or go on another amusement park ride.

The first study generally supported the notion that consumers can, when specifically asked to do so, describe emotions at various stages of a consumption experience. Specifically, participants were able to articulate desires and expectations with respect to emotions (antecedents of consumption emotions), the actual experienced emotion itself, and their evaluation of this emotion in terms of other emotions and satisfaction levels. Although these were experiences that they were asked to relate from memory, participants were able to describe their feelings before, during, and after the experience. Given that our procedure focused on a memory based "re-experiencing" of the original consumption, other studies that focus on concurrent evaluations are needed to corroborate this finding. The second goal of our study was also achieved since consumers' evaluations of the negative emotions directly experienced during consumption were positive, and thus opposite in valence. Both of these findings however provide only initial support for our model; the demand effects inherent in using focus groups lead us to a second exploratory study.


The second exploratory study used a lab experiment to further explore the notion of the dual role of emotions in experiential aspects of consumption. Specifically we wished to explore whether experienced negative emotions, if desired, could lead to favorable satisfaction responses. Whereas in Study 1 we explored actual emotions experienced during a prior consumption episode, in the second study we asked for consumer reactions to scenarios. Manipulations of desired positive and negative emotions were embedded in these scenarios and participants were asked for their reactions to these scenarios.

Study Design

Ninety eight undergraduate students at the School of Business, Indiana University participated in this study for course credit. A 2 (desires: joy vs. fear) x 2 (expectations: low vs. high) x 2 (perceived performance: low vs. high) factorial design was used to explore the impact of these factors on satisfaction and other post-consumption responses. A basketball game and a horror movie were chosen as consumption experiences for positive (joy) and negative (fear) emotions respectively. Desires were manipulated with a statement that focused on the corresponding emotion. For example, for movies, the statement "you have just managed to get first show tickets for a new horror movie that you hope will be as scary as the last one from this producer" was used to make the desire for a negative emotion (i.e., fear) particularly salient. Expectations were manipulated with a statement attributed to an independent newspaper review that the movie "is guaranteed to chill your spine" (high expectations) or that it "has been hyped up and is not very scary" (low expectations). Finally, perceived performance was manipulated by focusing on how predictable the movie was and hence not scary (low perceived performance) or that it kept you "on the edge of your seat the whole time and was very scary" (high perceived performance).





To reduce the demand effects inherent in a complete within-subjects design, each subject was asked to react to only two scenarios, one for the basketball game and one for the horror movie (desires manipulation). Moreover, the two scenarios to which each subject was exposed were in diagonally opposite cells for expectations and perceived performance. For example, if a particular subject was in the low expectations, low perceived performance cell for the horror movie, they would be placed in the high expectations, high perceived performance cell for the basketball game.

After reading each of the two scenarios, subjects were asked to evaluate how satisfied they would be in this situation, and to indicate the likelihood of other post-consumption behaviors such as positive word-of-mouth, complaining, and loyalty. To investigate the efficacy of the scenarios in manipulating expectations and desires, subjects were exposed to two additional scenarios (one for each consumption experience) that represented a different cell in the design. This time, their reactions to these scenarios focused on measuring their expectations and desires for enjoyable or scary experiences. The perceived performance manipulation was fairly direct and hence not conducive to a specific check.

Study 2 Results

Table 2 reports the mean values on the manipulation checks for desires and expectations used in the study. Separate 2 (expectations) x 2 (perceived performance) ANOVA models (one for each of the two desire conditions) were used to test the efficacy of the expectations manipulations. The expectations manipulation was successful as evident from a comparison of mean scores of expectations across the cells in the study. Compared to the low expectations condition, expectations for the basketball game (F(1,94)=278.75, p<.001) and the horror movie (F(1,94)=87.29, p<.001) were significantly higher in the high expectations condition. Similarly, anticipations were also higher in the high expectations condition for the basketball game (F(1,94)=147.18, p<.001) and the horror movie (F(1,94)=78.59, p<.001). Since the pattern of these expectations did not differ between the two consumption experiences (desires conditions), the expectations manipulation was successful.

Since desire was manipulated using scenarios involving experiences that differed in many, unknown ways (basketball game vs. horror movie), their differences cannot be directly tested. Rather, across the two desires conditions, we compare patterns of levels of desired emotions to assess the efficacy of the desires manipulation. Most importantly, the basketball game scenarios scored substantially lower (mean=2.20) than the horror movie (mean=6.09) on the scariness measure. On the other hand, subjects indicated that they desired more enjoyment from the basketball game (mean=6.12) compared to the horror movie (mean=5.84). Finally, desired excitement scores were equal for the basketball game and horror movie scenarios, which reflects arousal and/or the interest emotion. In summary, although we cannot test for statistical significance, the desires manipulation appears to be successful as indicated by these mean scores.

To test the effects of expectation and perceived performance on satisfaction, separate 2 x 2 ANOVA models were used for each of the two desire conditions. Table 3 reports mean values on satisfaction, positive word-of-mouth, complaining, and loyalty measures across the eight study cells. As expected, for desires relating to positive experiences (joy from the basketball game), high levels of perceived performance led to high levels of satisfaction (F(1,94)=109.96, p<.001). Moreover, expectations about the game did not have any effect on satisfaction (F<1). Hence, perceived performance appears to be more diagnostic in understanding satisfaction than expectations; this result is consistent with our model. Perceived performance also has a positive association with positive word-of-mouth and loyalty and a negative relationship with complaining behavior.

A similar pattern is evident for the horror movie in spite of the fact that the desires for this experience were characterized by a negative emotion, namely fear. The means in Table 3 show that when the horror movie was perceived as extremely scary, subjects were more satisfied (than when the horror movie was less scary, F(1,94)=398.58, p<.001)). Moreover, lower expectations led to higher levels of satisfaction (F(1,94)=7.82, p<.01). Finally, similar to the basketball game patterns, higher perceived performance led to higher loyalty and positive word-of-mouth, and lower levels of complaining behavior. The only interaction that was observed was for complaining behavior, with results indicating that subjects are more likely to complain after a low level of perceived performance when expectations are high than when expectations are low.

In summary, in this second study subjects reacted to consumption context descriptions regarding performance on hedonic attributes, and evaluated how they would feel in these situations. [Only the direct effect of perceived performance on satisfaction is examined in this study. As depicted in the Figure, the effect of perceived performance on satisfaction can also be partly mediated by evaluated performance.] The results show that subjects may experience a negative emotion (fear) during consumption, but if this is the desired emotion (as for horror movies), this leads to higher levels of satisfaction. This result contrasts with Westbrook's (1987) finding that consumption-based negative affect leads to dissatisfaction. Our Study Two results generally parallel the comments in our focus groups and hence, the dual role of emotions is substantiated in both studies. We should note here that the scenarios made only one attribute/emotion salient, and hence, attribute and overall satisfaction may not differ very much. When multiple attributes/emotions are involved, the overall satisfaction may be modeled as a complex function of satisfaction with individual attributes (Oliver 1993).




Our model and preliminary empirical results suggest that the relationship between emotions and satisfaction is considerably complex for products that are primarily hedonic in nature. Specifically, we have suggested that expectations and desires for hedonic attributes center on the specific emotions that form an integral part of that consumption experience. Moreover these pre-consumption standards may be formulated not only in terms of the type of emotion (fear, joy), but also the degree of emotion (extremely scared or very joyful). This idea is an extension of Oliver's (1993) notion of an attribute basis for satisfaction, except that the attributes are hedonic in nature.

Second, our model and exploratory research alerts researchers to the possibility that the emotion experienced during consumption is conceptually different from evaluations of this emotion. This distinction is critical for hedonic bases of consumption where the emotion is the consumption experience itself. Moreover, the separation of these two as distinct emotions raises the possibility that they may be quite different in valence. As demonstrated in our preliminary studies, positive emotions (joy) may be evoked in response to evaluations of negative emotions (fear). It is also possible that negative emotions could be experienced in response to positive emotions; for example, even pleasant stimuli (such as music) may, in certain conditions (e.g., when one is trying to concentrate on one's work), evoke negative emotions. Further explorations of these dual emotions is necessary for a complete understanding of these phenomena.

Finally, the relationships of these emotions to satisfaction is an important topic. Given that these consumption emotions may have strong direct effects (Westbrook 1987) or moderate effects of other variables on satisfaction, the separate role of these two types of emotions needs to be subjected to greater scrutiny. This implies that to correctly understand the nature of the relationship between emotions and satisfaction, future studies should have separate measures of those emotions that result directly from the hedonic experience and those that result indirectly from a comparison process.


Two separate studies offer initial support for the notion of dual role of emotions; however, these findings are exploratory. First, the focus group results are fairly tentative given the small sample and directive nature of the discussion. Hence, since no deception was involved, the possibility of demand effects and "artificial results" should not be ruled out. The second study used a different approach with manipulation of key concepts using scenarios, and hence does not suffer the same problems. However, even here, some limitations are present. First, the scenarios may be so directive that there was no possibility for other responses. Hence, once again demand effects may color the subjects' responses. Second, the desires manipulation is confounded with the nature of the service (the basketball game vs. the horror movie). Hence, future research needs to disentangle product effects. Third, these scenarios attempt to place students in various consumption contexts and assess their emotions. Since we want to investigate experiential aspects of consumption, it may be more instructive to assess emotions from actual experiences as they occur. Even though the first study focused on actual experiences, these consumption contexts were months and sometimes years in the past. Finally, although both studies offer support for parts of the dual emotions model, they do not test the entire model or compare this model with other models.

In conclusion, we have proposed and partially tested a model of satisfaction that parses emotions into consumption-based and evaluation-based responses. Our exploratory research offers evidence in support of these distinctions. Given the limitations discussed above, we call for a more systematic exploration of this topic.


Holbrook, Morris B. and Elizabeth C. Hirschman (1982), "The Experiential Aspects of Consumption: Consumer Fantasies, Feelings, and Fun," Journal of Consumer Research, 9 (September), 132-140.

Izard, Carroll E. (1977), Human Emotion. New York: Plenum Press.

Mano, Haim and Richard L. Oliver (1993), "Assessing the Dimensionality and Structure of the Emotion Experience: Evaluation, Feeling, and Satisfaction," Journal of Consumer Research, 20(December),451-66.

Oliver, Richard L. (1993), "Cognitive, Affective, and Attribute Bases of the Satisfaction Response," Journal of Consumer Research, 20 (December), 418-430.

Oliver, Richard L. and Robert A. Westbrook (1993), "Profiles of Consumer Emotions and Satisfaction in Ownership and Usage," Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior, Vol. 6, 12-27.

Olshavsky, Richard W. and Richard A. Spreng (1989), "A Desires as Standard Model of Consumer Satisfaction," Journal of Consumer Satisfaction, Dissatisfaction and Complaining Behavior," 2,49-54.

Spreng, Richard A. and Richard W. Olshavsky (1993), "A Desires Congruency Model of Consumer Satisfaction," Journal of the Academy of Marketing Sciences, 21, No. 3, 169-177.

Westbrook, Robert A. (1987), "Product/Consumption-Based Affective Responses and Postpurchase Processes," Journal of Marketing Research, 24 (August), 258-70.

Westbrook, Robert A. and Richard L. Oliver (1991), "The Dimensionality of Consumption Emotion Patterns and Consumer Satisfaction," Journal of Consumer Research, 18 (June), 84-91.