Journal of Marketing Rearch
V ol. XLV (December 2008), 654–664
654
©2008, American Marketing Association
ISSN: 0022-2437 (print), 1547-7193 (electronic)
*Leif D. Nelson is Assistant Professor of Marketing, Rady School of Management, University of California, San D iego (e-mail:ldnelson@ucsd.edu). Tom Meyvis is Associate Professor of Marketing,Stern School of Business, New York University (e-mail: u.edu). The authors acknowledge the helpful comments of Chris Janiszewski, Justin Kruger, and Anne-Laure Sellier on a previous draft of the manuscript. Both authors contributed equally to the manuscript; author order was determined randomly. Dilip Soman rved as associate editor for this article.
LEIF D.NELSON and TOM MEYVIS*
Six studies demonstrate that interrupting a consumption experience can make pleasant experiences m
ore enjoyable and unpleasant experiences more irritating, even though consumers avoid breaks in pleasant experiences and choo breaks in unpleasant experiences.Across a variety of hedonic experiences (e.g., listening to nois or songs, sitting in a massage chair), the authors obrve that breaks disrupt hedonic adaptation and, as a result, intensify the subquent experience.
Keywords :adaptation, affective forecasting, well-being, consumption,
hedonic experiences
Interrupted Consumption:Disrupting Adaptation to Hedonic Experiences管理新思维
Imagine a patient who is about to undergo a painful physical therapy ssion and is given the option to take a short break in the middle of the ssion. Would the patient accept the offer and break up the ssion, or would he or she prefer to endure the entire ssion without interruption?Regardless of the patient’s preference, would the break make the ssion more painful or less painful? Now, imag-ine a customer who is about to enjoy a relaxing massage and can choo to take a short break in the middle. Would the customer choo to break up the massage or to maintain the continuous experience, and regardless of his or her pref-erence, would the disruption make the massage more enjoy-able or less enjoyable?
As the two scenarios illustrate, consumers often have the opportunity to choo between interrupted or continu-ous experiences. This article investigates the conquences of taking breaks in affective experiences (i.e., when should consumers take a break?), as well as consumers’expecta-tions for the conquences (i.e., when do consumers take a break?). For example, would consumers enjoy the latest three-hour Bollywood musical more if they watched it in a theater that offers a brief intermission or in a theater that shows it without interruption—and which theater would they choo? Alternatively, if a person’s sweetheart coerces
him or her into attending a Goth-metal performance, would the experience be less painful if the band takes a break in the middle of its t or performs without interruption—and which band would he or she end up choosing?
THE PRESCRIPTIVE PERSPECTIVE: HOW DO BREAKS
INFLUENCE HEDONIC EXPERIENCES?
This study address two independent theoretical ques-tions: How does hedonic disruption actually influence experience, and how do consumers think hedonic disrup-tion will influence experience? Most important, we exam-ine how breaking up an experience actually affects con-sumers’enjoyment
of this experience. To asss this effect,we rely on two critical assumptions: (1) that consumers adapt to many experiences and (2) that breaks disrupt this adaptation process. Together, the assumptions imply that breaks will intensify affective experiences. Therefore, we propo that within certain boundaries, consumers should inrt breaks in positive experiences but not in negative experiences.
To understand how breaks influence consumers’affective experiences, we first need to consider how subjective expe-riences change over time. Over the cour of an experience,affective intensity can either increa (i.e., nsitization) or decrea (i.e., adaptation). Whereas nsitization often occurs for complex stimuli (e.g., high-quality wines), in many domains, adaptation ems to be the norm (for a review, e Frederick and Loewenstein 1999). People adapt surprisingly quickly and completely to a variety of positive and negative experiences, ranging from the buzz of a com-puter hard drive to extreme windfalls or calamities. Previ-ous rearch has shown that people adapt to repeated con-sumption of their preferred ice cream (Kahneman and Snell 1990), increas in income (Easterlin 1995), failure to
Interrupted Consumption655 achieve tenure (Gilbert et al. 1998), solitary confinement
(Suedfeld et al. 1982), and, to some extent, even extreme
and life-altering events, such as winning the lottery or
becoming a paraplegic after a vere car accident (Brick-
man, Coates, and Janoff-Bulman 1978).
In each of the situations, the subjective intensity of the
experience decreas over time as a result of either basic
nsory process (e.g., nsory-specific satiety) or com-
plex psychological mechanisms (e.g., coping). In addition
to the overall intensity reduction, adaptation creates a
downward trend in positive experiences and an upward
四叶草的英文trend in negative experiences, patterns that strongly influ-
ence overall subjective intensity (Ariely and Zauberman
腌泡菜
2000, 2003). We argue that inrting a break in the experi-
ence will disrupt the adaptive process. Becau adapta-
tion results from prolonged exposure, disrupting exposure
should reduce adaptation. Furthermore, this reduction in
adaptation will intensify the experience following the break
(relative to an uninterrupted experience). In short, con-
sumption experiences that are characterized by adaptation
will become more inten when breaks are inrted.
Although this reasoning suggests that people should
break up pleasant experiences but not unpleasant ones, this
recommendation necessarily requires some caveats. For
example, if an unpleasant experience is sufficiently aversive
(i.e., very inten or very prolonged), the experience may
need to be interrupted to provide people with the opportu-
nity to recover. Similarly, some positive events are primar-
ily enjoyed in the gestalt (e.g., films, sports events) and
may be best experienced continuously. We further discuss
the potential boundaries of our predictions in the “Gen-
eral Discussion” ction.
THE DESCRIPTIVE PERSPECTIVE: CONSUMERS’
FORECASTS AND PREFERENCES
Another objective is to examine whether and when
people prefer to break up hedonic experiences. Whether
consumers choo to inrt a break in an affective experi-
ence depends on their beliefs about how the break will
affect their enjoyment. As prior rearch has shown, people
are quite poor at predicting how their enjoyment of an
experience will evolve over time (Loewenstein and Schkade
1999; Wilson and Gilbert 2003). In particular, people have
trouble predicting hedonic adaptation. Although people
show substantial variation in intuitions about adaptation or
nsitization to different stimuli (Kahneman and Snell
1990), in general, they tend to underestimate adaptation
(Loewenstein and Frederick 1997). Indeed, people are more
likely to predict nsitization rather than adaptation for
ongoing, continuous experiences (e.g., loud nois, a last-
ing headache), exactly the type of stimuli that we study
herein (Kahneman and Snell 1990; Snell, Gibbs, and Varey
1995). Similarly, people often overestimate the lasting
influence of an event on their overall happiness (i.e., the
impact bias; Gilbert et al. 1998). For example, assistant
professors tend to overestimate the effect of the tenure deci-
sion on their future happiness (Gilbert et al. 1998), whereas
prisoners tend to underestimate adaptation to solitary con-
finement (Suedfeld et al. 1982).
In summary, people do not hold a uniform belief in adap-tation, and some evidence indicates that people may even intuit nsitization rather than adaptation. Therefore, we
1Note that this simple heuristic assumes some degree of myopia on the part of the decision maker. We assume that the decision maker will put greater emphasis on the start of the break (stopping the experience) than on what happens after the break (restarting the experience).
predict that consumers will fail to anticipate their adapta-tion to many consumption experiences and thus will not expect breaks to intensify the experiences by disrupting adaptation. Instead, we propo that consumers will rely on the simple mantic intuition that a broken-up experience is a weakened experience. That is, consumers will rely on a simple decision rule: Stopping a good experience is bad, and stopping a bad experience is good.1As a result, they will prefer to break up negative experiences but not positive experiences. In other words, we argue that consumers’pref-erences will neither maximize their enjoyment nor mini-mize their suffering. As such, the traditional adage of the rvice industry, “the customer is always right,” may not apply to decisions about structuring pleasant or unpleasant experiences, for which the customer may turn out to be wrong.
If the predictions hold, managers may need to structure rvice experiences differently depending on whether they aim to maximize actual customer enjoyment or initial con-sumer appeal. If the therapists in the aforementioned sce-narios want to enhance their customers’enjoyment, the physical therapist should not offer the opportunity to take a break (becau the customers would choo it but feel wor), whereas the massage therapist should build the break into the massage schedule (becau the customers would not choo it but feel better if they did). However, if a movie theater wants to increa its appeal to moviegoers, it should screen the movie without intermission (even though the audience would enjoy the movie less).
The studies we report empirically test our conjecture that breaks tend to intensify hedonic experiences, even when consumers hold the opposite intuition. However, before examining the actual impact of breaks on hedonic experi-ences, we examined people’s intuitions about this impact by prenting them with some hypothetical choices and vignettes. A preference for breaking up negative experi-ences but not positive experiences would support our assumption that people tend to regard broken-up experi-ences as weakened experiences.
In a first study, undergraduate students (n= 28) imagined that they would be parated from their romantic partner for a long period but could choo any two concutive weeks during this period to
be together with their partner. Partici-pants strongly preferred weeks in the middle of the period rather than weeks in the beginning or at the end, indicating a preference for breaking up this negative experience (86%;χ2= 14.29, p< .001). Furthermore, when participants were subquently asked to forecast the experience over time, fewer than one-third showed the pattern of decreasing intensity consistent with adaptation. Thus, most participants failed to anticipate adaptation to this negative experience, and most wanted to disrupt it (for methodological details for all studies in the article, e the Web Appendix at /jmrdec08).
In a cond study, the opposite pattern emerged for a positive experience. Undergraduate students (n= 39) imag-ined a summer working in the south of France, which afforded four days of vacation in Saint-Tropez. A signifi-
656JOURNAL OF MARKETING RESEARCH, DECEMBER 2008
Table 1
PREFERENCE FOR BREAKING UP EACH OF 15 DIFFERENT
EXPERIENCES Type of Experience
Percentage Preferring a Break
Positive Experiences First-class flight
5% (6 of 119)***Listening to great music 8% (10 of 119)***Going on vacation
14% (17 of 119)***Receiving a foot massage 15% (18 of 119)***Watching a movie 16% (19 of 119)***Eating ice cream
29% (35 of 119)***Generally pleasant experience
14% (17 of 119)***Negative Experiences Waiting in line 36% (43 of 119)**Dentist visit 40% (47 of 119)*Irritating noi
69% (82 of 119)***Holding hand in cold water 74% (88 of 119)***Introductory marketing class 75% (89 of 119)***Smelling a nasty odor 76% (90 of 119)***Painful headache
79% (94 of 119)***Generally unpleasant experience
72% (86 of 119)***
*p < .05.**p < .01.***p < .001.
Notes: For each experience, we compare the obrved proportions to 50%.
cant majority (90%; χ2= 24.64, p < .001) preferred to take the four days of vacation concutively, indicating a reluc-tance to break up this positive experience.
班级之星A third investigation compared similar positive and nega-tive experiences. Two groups of students (n = 138) consid-ered the prospect of either a pleasant massage or a ssion of painful physical therapy and reported whether they would prefer a break in this experience. As we expected,whereas most participants wanted a break in the physical therapy (63%, or 42 of 67; χ2= 4.31, p = .038), most avoided a break in the massage (28%, or 20 of 71; χ2=13.54, p < .001).
A final investigation asked a group of undergraduate stu-dents (n = 119) whether they would want a break in a ries of positive (e.g., listening to pleasant music) and negative (e.g., smelling a nasty odor) experiences. As Table 1shows,for all but two of the experiences, participants preferred continuous positive experiences and disrupted negative experiences. Furthermore, participants who did not expect to adapt to positive experiences showed a stronger reluc-tance to break up the experiences than tho who did expect adaptation (F(1, 116)= 5.15, p = .025; bad on an index of t
he ven positive events). Converly, participants who did not expect to adapt to negative experiences showed a stronger preference for breaking up the experiences than tho who did expect adaptation (F(1, 116)= 22.21,p < .001; bad on an index of the eight negative events).The scenario studies support our prediction that people want to break up negative experiences but do not want to break up positive experiences. There may be many reasons for this pattern of preferences, including the anticipated dis-ruption of nsitization (rather than adaptation), an over-estimation of the need for coping resources, or myopic preference for stopping unpleasant experiences and contin-uing pleasant ones. Our objective is not to distinguish between the different reasons (and we expect that there
are many) but rather to document the intuitions so that they can be compared with the actual effect of disrupting hedonic experiences. This actual effect of hedonic disrup-tion is the focus of the remainder of this article.
We propo that people adapt to a wide variety of experi-ences and that breaks disrupt this adaptation process and thus keep the experience at a high level of intensity. We begin by directly testing our assumption that inrting a break intensifies the subquent experience (Studies 1 and 2). This should increa both the irritation with a negative stimulus and the enjoyment of a positive st
imulus. We then attempt to rule out hedonic contrast effects as an alternative to the adaptation-disruption hypothesis (Studies 3 and 4).Finally, we track participants’ongoing experience to better understand the mechanism by which breaks affect overall enjoyment and irritation (Studies 5 and 6).
STUDY 1: DISRUPTION OF ADAPTATION TO A
NEGATIVE EXPERIENCE
The scenario studies we described previously indicate that people often prefer to break up negative but not posi-tive experiences. However, we argue that breaks disrupt people’s adaptation to the experience and therefore make positive experiences more enjoyable and negative experi-ences more unpleasant. We first test this for a negative experience:
浅圆仓H 1: Inrting a short break in an unpleasant experience disrupts
adaptation and makes the experience following the break more aversive, though people expect that it will become less aversive.
In three between-subjects conditions, we measured partici-pants’irritation with a five-cond fragment of a noi that was prented in isolation, at the end of a longer experi-ence, or immediatel
y following a break.
Method
暗香浮动恰好
One hundred forty undergraduate students participated in a study examining the evaluation of auditory stimuli. Par-ticipants were ated at a computer workstation, asked to put on headphones, and told that they would be listening to a brief sound clip of a vacuum cleaner. There were three groups of participants. The first group listened to only 5conds of the vacuum cleaner; the cond group listened to 40 conds; and the final group listened to 40 conds, fol-lowed by a 5-cond break, and then another 5 conds of the vacuum cleaner. Immediately after this experience, par-ticipants answered a question about the last 5 conds of the stimulus. To avoid possible scaling effects, we asked participants to compare the vacuum cleaner noi to another irritating noi. Therefore, we prented partici-pants with a 5-cond sample of a drilling noi, after which they reported their preference between the two sounds on a 201-point sliding scale from –100 (“definitely prefer the vacuum”) to +100 (“definitely prefer the drill”).Higher numbers reflect a more aversive vacuum noi experience.
Becau it was important that participants understood and complied with all elements of the procedu
re, we took a priori measures to eliminate tho who did not by using a previously validated procedure that was modified for this task (Oppenheimer, Meyvis, and Davidenko 2006; e also Simmons and Nelson 2006, Study 12). At the end of the
Interrupted Consumption
657
Figure 1
BREAKING UP AN IRRITATING NOISE DISRUPTS ADAPTATION,
MAKING THE EXPERIENCE MORE AVERSIVE
reported by experiencers (with standard error). Both measures have been rescaled to a 50-point scale; higher numbers reflect more irritation. Given the difference in the initial measures, absolute differences in ratings for forecasters versus experiencers cannot be interpreted.
ssion, participants briefly read about the importance of following instructions and were asked to ignore a scale on the screen and instead click on a red square at the top of the screen. Although most participants clicked on the red square, 14 people instead clicked on the unlabeled scale and were removed from subquent analys.平安夜是哪一天
Results and Discussion
Forecast . In addition to this procedure, we asked a pa-rate group of participants (n = 40) to make predictions about the effects of the manipulation. All the participants listened to a 5-cond sample of the vacuum cleaner, rated it on a 76-point scale of irritation (0= “a little irritating,”and 75= “incredibly irritating”), and then estimated how irritating they would find the same sample in each of the other two conditions. Consistent with our initial studies,participants expected nsitization rather than adaptation:They expected the 40-cond experience to increa the irritation relative to the 5-cond sample (M = 42.07 versus 26.07; t(39)= 4.86, p < .001). Furthermore, participants had some
belief that the 5-cond break would reduce their irritation with the stimulus relative to the continuous experience (M = 42.07 versus 35.92; t(39)= 1.68, p = .10).In summary, participants forecasted that the stimulus would become more irritating with prolonged exposure and that this increa would be partially mitigated by adding a break. The forecasts are the direct opposite of our hypothesis, which we test next.
Experience . We expected that people would adapt to the noi and therefore find it less aversive after 40 conds than after only 5 conds. Furthermore, we hypothesized that breaking up the experience would disrupt this adapta-tion process and reestablish the aversiveness of the stimu-lus. We tested this prediction with a planned contrast, com-paring evaluations by participants in the 40-cond experience with evaluations in the two remaining condi-tions. This proved to be a reliable contrast becau people experiencing the noi continuously for 40 conds judged the last 5 conds to be less aversive (M = –39.49) than people experiencing just the first 5 conds (M = –14.98) or people experiencing the 5 conds after a break (M =–16.44; t(123)= 2.10, p = .039; e Figure 1). The results indicate that though people want to break up negative expe-riences, this is not always a wi decision. Whereas listen-ing to the noi for an extended period made the noi less aversive, inrting a break made the noi just as aversive as it had been initially, suggesting that the break disrupted the adaptation process.
STUDY 2: DISRUPTION OF ADAPTATION TO A
POSITIVE EXPERIENCE
Study 1 demonstrated that breaks can disrupt adaptation to a negative stimulus. In Study 2, we shift to the positive domain and examine whether disruption of adaptation can make a positive experience more pleasant:
H 2: Inrting a short break in a positive experience makes the
experience more pleasant, though people expect that it will become less pleasant.
In this study, participants either experienced an uninter-rupted massage or had the middle part of the massage replaced with a 20-cond break. In addition, whereas par-ticipants in Study 1 were only asked to rate the last 5 c-
onds of the experience, participants in this study provided evaluations of the entire experience, thus enabling us to draw more general inferences about the normative implica-tions of inrting the break. We predicted that though people would prefer not to interrupt the massage, inrting a break in the middle of the experience would actually enhance their enjoyment of the massage.
Method
Forty-nine undergraduate students were ated at a com-puter in a chair outfitted with a massage cushion and were told that they would be testing the cushion. They first rated a 5-cond sample of the massage and then answered five questions about their general massage chair preferences.Most were irrelevant filler questions (“Do you prefer heat-ing in a massage cushion?”), but one question asked partici-pants whether they would prefer a continuous massage or a massage with a brief break in the middle. Participants were then randomly assigned to either the break or the continu-ous condition. In the continuous condition, participants learned that they would experience the massage continu-ously for 3 minutes, whereas tho in the break condition learned that they would be experiencing 80 conds of mas-sage, 20 conds of nothing, and then 80 more conds of massage. The experimenter then started the massage cush-ion and turned the cushion off and on according to the condition.
Note that in the break condition, we replace 20 conds of massage with a 20-cond break. Alternatively, we could have inrted the 20-cond break in the experience, thus lengthening the entire experience. Each of the strategies reprents a conceptual trade-off. The replacement strategy keeps the total length of the experience constant across con-ditions but results in a shorter
massage in the break condi-tion, whereas an inrtion strategy would have kept the
658JOURNAL OF MARKETING RESEARCH, DECEMBER 2008
2We detected unusually high variance on the measures. Therefore, we
removed outliers on each measure that were more than 2.5 standard devia-tions from the mean. As a result, there are slight variations in the degrees of freedom in each analysis. In addition, to control for preexperience dif-ferences in the enjoyment of the massage cushion, we included the 5-cond sample rating as a covariate in our analysis.
length of the massage constant but would have resulted in a longer total experience in the break condition. To ensure that our results are not limited by the specific strategy ud for inrting the break, we u different procedures across experiments, relying on the replacement strategy in two studies (Studies 2 and 5) and the inrtion strategy in the remaining studies (Studies 3, 4, and 6). Furthermore, in Studies 3 and 4, we also add a break before the stimulus in the continuous condition, thus ensuring that the total experience is of equal length in both conditions. The results are consistent across the different studies and across all instantiations of the manipulation.
After the massage, participants first reported how much they had enjoyed the experience using a 9-point scale (1=“not pleasant,” and 9= “extremely pleasant”). Next, they ud a 201-point sliding scale to compare their experience to listening to 3 minutes of their favorite song (–100=“definitely prefer the massage,” and +100= “definitely pre-fer the song”). This preference measure was added to avoid scaling effects. Participants then reported how much they would be willing to pay to repeat the experience and how much they would be willing to pay to buy the massage cushion. Finally, participants completed the same screening measure that was ud in Study 1.
Results and Discussion
Forecast . Before conducting any analys, we removed 8participants who incorrectly responded to the screening measure, leaving us with a sample of 41. We then examined participants’choices between continuous and interrupted massage experiences, which we measured after the 5-cond sample but before the 3-minute experience. As we expected, most participants preferred the continuous experi-ence (73%; χ2= 8.80, p = .003). However, although people predicted that a break would lesn enjoyment, postexperi-ence evaluations showed the exact opposite result.2
Experience . Compared with participants who experi-enced the continuous massage, tho who had
a break in their experience rated their experience as more pleasant (M = 7.05 versus 6.05; F(1, 37)= 4.59, p = .039), were less likely to prefer listening to their favorite song instead (M =4.65 versus 32.33; F(1, 38)= 4.20, p = .047), were willing to pay more than twice as much to repeat the experience (M = $3.71 versus $1.27; F(1, 33)= 6.69, p = .014), and were willing to pay almost twice as much to purcha the massage cushion (M = $26.59 versus $14.41; F(1, 36)=5.76, p = .022). Although most participants expected that inrting a break would detract from the massage, the results confirmed our hypothesis that the break does enhance the experience.
Why do breaks improve positive experiences and worn negative experiences? Although we argue that breaks inten-sify experiences by disrupting adaptation, the effects may be due to hedonic contrast: The vacuum cleaner is unques-tionably irritating, but it may em even more irritating when contrasted with the welcome silence of the break.
3We did not ask about the negative-break condition, becau this condi-
tion would not offer any additional insight into beliefs about disruption of adaptation (which both other break conditions test) or hedonic contrast (which the positive-break condition tests). The same logic applies to the forecasting study we report in Study 4.
Consistent with this possibility, contrast effects play a large role in the reporting of subjective well-being (Tversky and Griffin 1991). Thus, although recent rearch has failed to find contrast effects in hedonic consumption (Novemsky and Ratner 2003), hedonic contrast effects are both plausi-ble and intuitive. The next two studies clarify the role of contrast effects by systematically varying break valence in either a negative (Study 3) or a positive (Study 4) experi-ence. If the intensifying effect of the break is driven by dis-ruption of adaptation (rather than hedonic contrast), the effect should persist regardless of the valence of the break:
H 3: D isrupting a negative experience makes the experience
more aversive, regardless of the valence of the disruption.H 4: D isrupting a positive experience makes the experience
史铁生的母亲more pleasant, regardless of the valence of the disruption.
STUDY 3: DIFFERENT BREAKS IN A NEGATIVE
EXPERIENCE
Method
One hundred venty-eight undergraduate students were randomly assigned to one of four conditions. Participants in the continuous condition listened to 20 conds of silence,followed by 180 conds of vacuum noi. In the remaining three conditions, participants also listened to 180 conds of vacuum noi, but they were told that after 160 conds,this experience would be interrupted for 20 conds. This interruption consisted of 20 conds of silence in the neutral-break condition, classical piano music (Glenn Gould performing Bach’s Goldberg Variations) in the positive-break condition, and 20 conds of a child practic-ing scales on a violin in the negative-break condition.
After the experience, all participants evaluated the over-all experience on two measures; they rated their enjoyment of the overall experience (on a nine-point scale anchored by “not unpleasant” and “extremely unpleasant”) and their relative preference between listening to the vacuum cleaner noi and listening to a drilling noi (on the 201-point scale ud in the previous studies). Finally, participants completed the same screening measure ud in the previous studies (which led to the elimination of 16 participants).Results and Discussion
Forecast . A parate group of undergraduate students (n = 42) forecasted respons to the critical conditions of this experiment. Participants listened to a 5-cond sample of the noi and then read
descriptions of the continuous,positive-break, and neutral-break conditions.3Participants reported that adding a neutral break (M = 50.0) would not be any wor than the continuous experience (M = 50.9)but that the positive break would slightly improve the experience (M = 45.1; t(41)= 1.98, p = .054). With the forecasts in mind, we considered the actual impact of the different breaks on the experience.
Experience . Becau the two evaluation measures were reliably correlated, we standardized and combined them into a single index (higher numbers indicated more irrita-