Does exposure enhance liking for the chilli burn

Appetite, 1995, 24, 107-120

Does Exposure Enhance Liking for the Chilli Burn?

RICHARD J. STEVENSON and MARTIN R. YEOMANS Experimental Psychology, University of Sussex

It has been suggested that merely being exposed to a novel stimulus can enhance liking for it. Two experiments explored this hypothesis in relation to the development of liking for the chilli burn in a two-course meal. In the first experiment there was a linear increase in rated liking for the burning sensation produced by 2'5 and 5"0ppm capsaicin between the first and the fifth exposure to an experimental meal, whereas no significant change was seen for the vehicle food or the preceding non-chilli flavoured food. In the second experiment a similar linear trend was observed with the 5-0ppm stimulus. However, liking and sensory ratings for an ascending capsaicin concentration series in a tomato juice vehicle tested before and after the exposure period were unaffected by any change in liking for the chilli burn in food. This experiment also suggested that increases in liking were not dependent on initial unfamiliarity with the chilli burn or on changes in rated burn intensity or arousal. Potential explanations for the observed effects are considered.

INTRODUCTION

The mechanics underlying the development of food preferences are still poorly unders tood. This paper examines the role o f one part icular mechanism, mere exposure (Zajonc, 1968), in the development o f liking for the oral burning sensation experienced when eating chilli (the chilli burn). According to this hypothesis, an increase in liking and preference will result f rom repeated unreinforced exposure to an initially novel stimulus (Bornstein, 1987). Given that mere exposure has been invoked as a potential mechanism for the development o f food preferences (e.g. Rozin & Vollmecke, 1986) and more specifically for chilli liking (Rozin, 1990), it is surprising to find that mere exposure effects with respect to food have received little empirical analysis, with only four published investigations using human subjects (Stang, 1975; Pliner, 1982; Birch & Marlin, 1982; Crandall , 1984). The experiments reported here were designed to see whether regular exposure to foods containing the active principle in chilli, capsaicin, altered liking for the chilli burn, and to examine the mechanisms underlying such changes.

Rozin (1990) suggests that for chilli, liking develops initially by enhancement o f preference for the flavour o f food in which chilli is eaten. Gradually, preferred levels

This research was supported by a studentship for R. Stevenson, from the UK Science and Engineering Research Council.

The authors would like to thank Dr P. G. Clifton, Dr Z. Dienes and Mr P. Russell for their advice and assistance.

RJS now at CSIRO Sensory Research Centre, PO Box 52, North Ryde, NSW 2113, Australia. All correspondence and requests for reprints to Dr M. R. Yeomans, Experimental Psychology,

University of Sussex, Brighton, BN1 9QG, UK.

0195-6663/95/020107+ 14 $08.00/0 © 1995 Academic Press Limited

108 R.J . STEVENSON AND M. R. YEOMANS

of piquancy increase, as do the contexts in which chilli is consumed, until finally there is liking even for the isolated burn. Five types of mechanism have been suggested by Rozin (1990) to account for these changes in liking: these are associative learning, opponent process, mere exposure, benign masochism (in which the eater enjoys the constrained risk of the burn) and, finally, receptor desensitisation for oral chemogenic pain. Although this last mechanism appears to be unlikely (Rozin, Mark & Schiller, 1981), the other mechanisms all remain possible explanations of development of liking for chilli burn.

Mere exposure has been invoked by Rozin (Rozin and Schiller, 1980; Rozin, 1990) as one account of the development of burn liking. Although no previous study has specifically examined this claim, similar effects have been investigated with other food items. For example, Pliner (1982), studied the effect of exposure on liking for novel fruit juices, by giving subjects numerous cliscrete exposures to the stimuli mainly within one experimental session. Pliner found that liking increased between initial and final exposure. Crandall (1984) gave the employees of a remote Alaskan fish cannery a number of exposures over days to a food stimulus (doughnuts) not normally available in their diet. The doughnuts were presented among a variety of other foods during the workers' coffee breaks and Crandall found that overall consumption of the target food rose across exposures (in this experiment liking was presumed to be indicated by increased consumption). In the only previous study using chilli, Stang (1975) presented a mixture of spices, en m a s s e , in two experimental sessions. Stang found no overall increase in pleasantness ratings for the spices. However, he did not attempt to look at the changes for individual spices nor did he present the spices in a naturalistic context as part of a food stimulus.

The experiments reported here attempted to test the mere exposure hypothesis under relatively naturalistic conditions. Exposures occurred at weekly intervals to a two-part meal consisting of plain ratatouille followed by ratatouille with chilli. This design was adopted so that if any change in liking for the taste of the second, chilli- containing portion was observed, the plain meal would act as an exposure control to test for a general increase in liking for the ratatouille. No counterbalancing of portion order was possible because of the slow rate of decay of the chilli burn, which could have significantly altered liking for any subsequent food.

Previous studies (e.g. Pliner, 1982; Crandall, 1984) have stressed the importance of novelty as a prerequisite for a successful exposure effect. Foods were used for which there was known to be no previous exposure (Crandall, 1984) or that were unlikely to have been familiar to most individuals (Pliner, 1982). Novelty with respect to chilli-burn is difficult to establish as it occurs in a wide variety of foodstuffs and non-food materials (e.g. toothpaste). Conscious knowledge of previous exposure is confounded by intensity, in that a subject may be familiar with a very weak burn, but not with a more intense one. In an attempt to choose subjects who were likely to be unfamiliar with the burn intensities used in this study, who had a low rate of current exposure to chilli and who were willing to repeatedly experience it over the course of the experiment, a variety of strategies was employed. Initial selection was based on subjects showing neither liking nor dislike for chilli, to eliminate those who might be unwilling to consume chilli-containing foods repeatedly and those who might already have reached their maximum liking. All subjects within the selected range reported uniformly low rates of current exposure. Finally, subjects were assigned randomly to one of two groups in whom different concentrations of capsaicin (the principal irritant in the chilli pepper: Maga, 1975) were used. It was presumed

EXPOSURE AND BURN LIKING 109

that at least some subjects would be unfamiliar with the doses presented to them. Subjective novelty was then established in a questionnaire at the end of the experiment.

EXPERIMENT 1

Method

Subjects Students and employees of the University of Sussex who had completed an earlier food-preference questionnaire (R. J. Stevenson, unpublished data) formed the pool from which subjects were drawn. Selection was based on the subject's liking or disliking for chilli in cooked food, indicated by a mark on a 100-ram line anchored at the ends "Strongly like" and "Strongly dislike". Subjects whose responses fell in the mid-range (40-60 mm region) were eligible to participate. Frequency of consumption of chilli was also assessed by response on the categories Never, Less than once a year, More than once a year, More than once a month, More than once a week and Once a day. Twelve subjects agreed to take part and were randomly assigned to two groups, Group 2-5 (n=6 , two males and four females, mean age= 22.0, meal like/dislike score=49.5 mm, median frequency of chilli consumption= less than once a year) and Group 5'0 (n=6, three males and three females, mean age = 30"1, mean score = 50-0 ram, median chilli frequency = less than once a year). All subjects were paid for participating.

Food stimuli Subjects were presented with two portions of 195 g each of canned ratatouille

preheated in a microwave (serving temperature approx. 65°C). The second portion of ratatouille contained either 2.5 or 5-0 ppm capsaicin (Bush, Boake & Allan Ltd., 10% Capsicum Oleoresin), depending on condition. Capsaicin was dissolved in butter so that the addition of 1 g of butter to the ratatouille resulted in the appropriate capsaicin concentration in the food.

Procedure On experimental days, subjects were instructed to eat a normal breakfast and

then to refrain from eating until lunchtime, during which they were to receive the experimental meal. On arrival, subjects recorded when and what they had last eaten that day and when they had last eaten chilli or strong, spicy foods. Following this, subjects were familiarized with the computerized ratings used to collect all responses in the experiments reported here. This procedure involved stopping and then adjusting a bar which moved back and forth along a line between the anchors "Not at all" and "Very" (after Yeomans et al., 1990), except where other anchors are stated. Below the rating line was displayed the question to which subjects were responding. Responses were assigned numerical values of between 0 and 100 by the computer, where 0 represented a response at the "Not at all" anchor and 100 a response at the "Very" anchor.

All subjects initially rated their hunger ("How hungry are you now?") and were then prompted to call for the first phase of their meal (195 g of plain ratatouille). Once that portion had been eaten, subjects rerated their hunger, and then rated how pleasant they had found the food ("How pleasant was the taste of the food?"). At

110 R.J. STEVENSON AND M. R. YEOMANS

this point subjects were allowed a drink of mineral water and were reminded to abstain from I drinking except when prompted. The second phase of the meal then began, which consisted of a 195 g portion of ratatouille containing capsaicin at either 2.5 or 5"0 ppm, depending on group. Following cessation of eating, subjects rated how hungry they still were, how pleasant they had found the food and how intense ("How strong is the burning sensation now?") and how pleasant ("How pleasant is the burning sensation now?") they found the chilli burn. The time between the start and the end of food consumption in each phase was recorded by computer. This is important especially for the second, capsaicin meal, since subjects could develop a strategy of either allowing the food to cool or eating more rapidly, both of which might reduce the intensity of the chilli burn and thus alter affective responding (Sizer & Harris, 1985; Green, 1990).

All subjects were tested weekly for five weeks. At the end of the final session, subjects were asked what they thought the purpose of the experiment was, whether they had eaten chilli with ratatouille before the study and whether they thought they would change their chilli or ratatouille consumption frequency as a consequence of the experiment. Familiarity with the level of chilli burn used in this study was established by asking subjects "Would you normally eat chilli as strong as in this study?". Subjects responding positively were classed as familiar, whilst subjects responding negatively were classified as unfamiliar.

RESULTS

Mean responses to "How pleasant is the burning sensation now?" increased between the first and last session for both the 2.5 and 5-0ppm groups (Fig. 1). Trends analysis found a significant linear coefficient for burn liking across sessions, F(1,10)--5-56, p<0"05, but no higher trend was significant. The apparent difference in burn liking between groups (Fig. 1) failed to reach overall significance, F(1,10)= 1-53, NS, and there was no interaction between group and session. Thus, liking for the burning sensation was found to increase linearly across sessions.

Responses to "How strong is the burning sensation now?" appeared to differ between groups, but in trends analysis only the interaction term for cubic trend was significant, F(4,40)=2-55, p<0"05. This complex result appears to arise from a non- linear decrease in intensity for the 5.0ppm group but a contrasting non-linear increase in the 2"5 ppm group (Fig. 1).

Liking for the taste of ratatouille, both plain and with capsaicin, varied among sessions. Overall liking at the end did not differ from that at the start for any food, and no clear pattern of liking change was discernible. The only statistically significant effects were quadratic and cubic contrast interactions with meal type (Table 1). The plain meal was rated as tasting significantly more pleasant than the chilli meal (Table 1), F(1,10)= 13-61, p<0.005. However, subjects in the 2-5-ppm group liked both the plain and the chilli meal, whereas subjects in the 5-ppm group liked the plain, but not the chilli meal: interaction of meal type and group, F(1,10)= 8-08, p<0.025.

Hunger data were examined for a difference in initial hunger ratings confounding between groups. As expected, hunger ratings decreased significantly across the course of the meal, F(2,20)=72.84, p<0.0001. These ratings did not differ significantly between groups, or between the first and last session (Table 1).


2o i o

o

ioo

" 9o

80

70 (A)

6o

5o

4o

30

70

60

%

(B)

I I

1 2 3 5

I I I I l

1 2 3 4 5

Session

FIGURE 1. Changes in ratings (mm) of (A) liking and (B) burn intensity across the five exposure sessions for subjects eating foods containing 2-5 (O) and 5.0 (ll)ppm capsaicin.

No significant changes in meal length across the five exposure sessions were found for either group or food type (Table 1). There was no clear evidence that subjects were altering liking by manipulating meal length to reduce burn intensity.

Debriefing No subject reported eating chilli or strong spicy food in the 24 h preceding any

weekly exposure. Three subjects ate chilli outside of the experiment during the course of the weekly exposures, one on two occasions, the other two on one occasion. However, the effect of these extra exposures may have been small, since the difference in burn pleasantness ratings between sessions one and five (one-sample t-test with p = 0), with data collapsed across groups, was still significant excluding those three subjects: mean + SEM change 13-8 _ 5"8 mm; t(8) = 2.38, p<0"05.

All subjects of group 5"0 stated that they were unfamiliar with their level of burn and only two subjects in this group reported having eaten ratatouille with chilli prior to the experiment. Three subjects in group 2-5 were unfamiliar with their level of burn and three familiar, while four subjects had eaten ratatouille with chilli before. Burn pleasantness ratings increased in these subjects between sessions one and five: mean + SEM rating change 14.4 +4"8; one-sample t(8)=2"98, p<0"05. There was a non-significant increase in likelihood of consumption of chilli and ratatouille products (Table 2). No subject referred specifically to liking or preference in describing the purpose of the experiment. However, some subjects thought that the experiment concerned some form of sensory adaptation to the chilli burn. This type of response


TABLE 1 Liking for the flavour of the chilli and non-chilli Jbods, hunger ratings and meal duration (mean +_ SEM) in the five exposure sessions for groups receiving either 2.5 or 5.0ppm

chilli in experiment 1

Session

1 2 3 4 5

Pleasantness of the non-chilli food (mm) Group2-5 71.2+11.0 79"2+ 6.2 74.2+ 5.4 68.2+ 7.2 70.8+ 6.3 Group 5"0 85.5___ 4.3 88"8+ 3.7 82.3+ 6.3 85.0+ 5"6 91-3+ 3-4

Pleasantness ratings for the chilli food (mm) Group2"5 68"3+ 5"1 67.5+ 4.9 59.7+ 9-1 67.0+ 7.5 64.7+ 8-4 Group 5-0 29-0+12'2 26.5+ 9.5 31.2+14.4 34.2+12.36 31.8+14.2

Hunger ratings for group 2-5 (mm) Start of meal 70.5+ 6.5 72-7+ 6.2 60.8+ 4.6 71.0+ 8-0 71.5+ 7.2 Start of chilli portion 54.7+ 7.9 53-8+ 8.7 43.2+ 7.1 53.5+ 7-2 59"7+ 6.7 End ofmeal 27.8+ 6"9 29.0+ 7-6 29.7+ 6.4 38"5+ 7.7 32.8+ 7-1

Hunger ratings for group 5-0 (mm) Start of meal 65.3+ 6.5 66-8+ 6.2 67-8+ 4.6 71.3+ 8.0 66"3+ 7.2 Start of chilli portion 40"2+ 7-9 56.4+ 8"7 42.3+ 7.1 493"0+ 7.2 36.0+ 6.7 End of meal 21.0+ 6-9 29"0+ 7.5 21.8+ 6"3 26.0+ 7.7 21-2+ 7.1

Meal duration (s) Group 2'5, non- chilli 155.2+12.7 169.7+15.2 170-2___17.7 150.2+27.9 132.0+10"4 Group 2.5, chilli 165.2+ 16-8 192.2+29-8 196.2+33.8 190.8+30.1 163.3+20"6 Group 5-0. non- chilli 169.7+ 8.2 159.3+12-8 176.5+15-5 166.7+12.9 174.3+12.8 Group 5'0, chilli 209.3+12'8 185.0+30.8 205-7+16.9 158.7+13.2 194-7+28'5

TABLE 2

Likelihood of eating chilli or ratatouille again as a consequence of exposure, divided according to exposure

group from experiment 1

Category Group 2.5 Group 5"0

More likely to eat ratatouille 4 More likely to eat chilli 2

No change, for ratatouille 2 No change, for chilli 4

Less likely to eat ratatouille 0 Less likely to eat chilli 0

is exemplified in one subject 's reply, "To see if a person becomes used to the s t rength o f chilli in food over a per iod o f t ime". Consequent ly , the observed changes in perceived burn s t rength need to be viewed with some cau t ion because o f the potent ia l for d e m a n d characteris t ics .

EXPOSURE AND BURN LIKING 1 13

DISCUSSION

This experiment provides the first evidence that repeated exposure to capsaicin enhances liking for the chilli burn. This effect cannot be explained readily by sensory adaptation, since rated burn intensity did not change in the group exposed to 2-5 ppm capsaicin and, although intensity did decrease in the 5'0 ppm group, this change was apparent in only the last session, in contrast to the linear increase in liking for the burn. There was also no evidence that subjects had developed a strategy to minimize burn intensity through altering meal duration. However, the absence of any change in liking for the taste of the ratatouille with chilli is puzzling in the light of Rozin's (1990) contention that initially preference develops for the flavour that chilli adds, rather than for the burn per se. Stimulus novelty is an important prerequisite for generating exposure effects (Harrison 1977) and this might explain the findings in this experiment. Neophobia to foods has been demonstrated in humans (e.g. Lee, 1989; Pliner, 1991) and would be expected to occur towards the chilli burn. If subjects were neophobic when initially confronted with the burning sensation, then the increase in liking observed in this experiment may represent the attenuation of a neophobic response to a particular intensity of the chilli burn.

EXPERIMENT 2

As experiment 1 was the first demonstration of an exposure effect with chilli, an extended replication was conducted. Apart from testing the reliability of the observations in experiment 1, a number of further issues were examined. The first aim was to establish whether initial unfamiliarity with the chilli burn (as judged by the subject) predicted the degree to which liking changed between the first and last experimental meal. This would provide some indication whether the subjective novelty of a particular intensity of the chilli burn is a sufficient condition for the effects of mere exposure to occur.

The second aim was to explore the potential role of arousal in the exposure effect observed in experiment 1. If an attenuation of neophobia is responsible, then liking might increase through a decrease in arousal caused by increasing familiarity with the burn, since Berlyne (1971, 1973) claimed that arousal may be a principal determinant of liking responses. This has been observed in an analogous situation in rats, where exposure to a novel flavour was described as highly arousing (Domjan, 1977).

Rozin (1990) suggested that, as liking for the burn develops, liking for the contexts in which it is enjoyed also increases. Thus the third aim was to test whether changes in liking for the chilli burn produced by exposure to capsaicin in the experimental meals would generalise to another context where capsaicin was present, namely a separate test situation, in which capsaicin solutions were evaluated by liking and sensory intensity ratings. This is a sensitive method for examining the relationship between sensation and liking, and for distinguishing between the responses of chilli likers and non-likers (Stevenson & Yeomans, 1993). This test was made 1 week before subjects underwent their first experimental meal and repeated 1 week after their last meal. It was predicted that any change in liking during the meals would transfer to the solution context. Only the 5"0 ppm stimulus was used in experiment 2 since this was perceived as the more novel stimulus in experiment


1. Since perceived burn intensity decreased in subjects experiencing 5"0 ppm capsaicin in the previous experiment, a second function of the taste test was to see whether changes in burn intensity also transferred to different contexts.

Method

Subjects Students or employees of the University of Sussex were selected using the criteria

described in the preceding experiment. Fourteen subjects (ten female and four male) agreed to take part (mean age = 20"0, mean chilli liking = 53.3 mm, median frequency of chilli consumption=more than once a year), none of whom had completed previous studies involving chilli. Subjects were paid for participating.

Materials Stock solutions for the preference test were made by dissolving 10% capsicum

oleoresin (Bush, Boake and Allan Ltd) in 2% polysorbate Tween-80 (ICI Speciality Chemicals). On experimental days, stock solutions were diluted 50:50 (by volume) with tomato juice (to enhance palatability) to yield capsaicin solutions of l, 2, 4 and 8 ppm. Mineral water was used for mouth rinses and a sucrose rinse 00% sucrose w/w in tap water) was used to reduce carry-over of burns between trials (see Nasrawi & Pangborn, 1990). All solutions and rinses were presented at 24°C.

The food and its preparation were identical to those described for experiment l, except that all subjects received ratatouille containing capsaicin at 5 ppm.

Preference test All subjects completed the pre-experimental preference test and its week 7

replication at lunchtime, before having eaten. On arrival, subjects completed the pre-experimental questionnaire (as in experiment 1) and a practice trial, after which they were left alone to complete the test. Subjects were instructed to take the first solution (1 ppm), pour it into their mouth, swill vigorously and then expectorate. Following this, four sets of ratings were made, each set within a 30-s period. Each set comprised of a rating of burn pleasantness ("How pleasant is the burning sensation now?"), intensity ("How strong is the burning sensation now?") and arousal ("How passive or active do you feel now?" with anchors "Passive" and "Active" after Weinberger, 1984). However, in the first set of ratings of each solution, subjects also rated their degree of familiarity with that level of burn ("Have you experienced this strength of burn before?" with anchors "Definitely no" scored 100 and "Definitely yes" scored zero). Subjects were instructed when using the familiarity rating that responses favouring the "Definitely yes" anchor indicated familiarity with that intensity of burn, whilst responses favouring the "Definitely no" anchor indicated unfamiliarity with that intensity of burn.

On completion of all the ratings for a solution, subjects were instructed to rinse and expectorate with sucrose solution, and then mineral water. This exact procedure was repeated for the next solution, until all four capsaicin solutions had been sampled in ascending order of concentration. Although the sucrose solutions should reduce carry-over of burn between solutions, some was inevitable. However, this was not crucial since the key comparisons were within rather than between subjects.

EXPOSURE AND BURN LIKING 1 15

TABLE 3 Changes in ratings of liking and burn intensity across the five exposure sessions from

experiment 2

Session

1 2 3 4 5

Burn liking 31.8+6-2 32.8+6.6 31.0+7-2 39.4+7.2 40.6+7.1 Burn intensity 84.8+2-4 83.6+2.6 83-4+3-2 78.2+3-3 79-5+3.8 Liking for "plain" food 76-5+3"0 78'8+2'7 80"3+3-5 83'9+2-9 79-0+4-4 Liking for "chilli" food 38.9+7.0 40"5__+6'4 44-5+8-1 47"6__+8'7 49"5__+9"4

Experimental meals During weeks 2-6 the procedure observed in experiment 1 was employed with

two modifications. First, subjects rated arousal after hunger and meal pleasantness ratings for the plain ratatouille and for the chilli ratatouille. Secondly, subjects rated their degree of familiarity with that intensity of the chilli burn after completing all the other ratings following the chilli ratatouille.

Debriefing Following the preference test, subjects were asked whether their liking for

ratatouille or chilli had changed since the start of the experiment, whether they would now choose hotter, the same strength or weaker doses of chilli as a result of the experiment and, thirdly, whether their frequency of ratatouille consumption would alter as a result of the study. Finally, subjects stated what they thought the purpose of the experiment was.

Results

Experimental meal Ratings of pleasantness of the burning sensation now tended to increase linearly

across the five experimental sessions (Table 3), although the linear contrast just failed to reach significance, F(1,13)=4-02, p=0"06. However, in contrast to experiment 1, there was a significant linear decrease, F(1,13)= 5"63, p<0.05, in rated intensity of the burning sensation (Table 3). Quadratic and cubic contrasts were not significant for either rating. To test whether changes in liking were related to changes in sensation, changes in liking and burn intensity between sessions one and five were correlated. Spearman's rho was 0.41, which was not significant. Thus, increased liking may not have been due simply to a decrease in burn intensity.

As in experiment 1, there was no significant change in ratings of pleasantness of the taste of the food across the five exposures (Table 3); the increased mean liking for the chilli meal was principally the result of a large positive hedonic shift in two subjects. As before, subjects rated the plain ratatouille as pleasant and the chilli ratatouille as mildly unpleasant, F(1,13)= 20.76, p<0-001. Ratings of arousal, which were predicted to decrease for the chilli ratatouille as the subjects became more familiar with the burn, did not alter significantly across the five sessions in either


TABLE 4 Hunger ratings and meal duration (mean + SEM) across the five exposure sessions fi'om

experiment 2

Session

1 2 3 4 5

Hunger ratings (mm) Start of meal 72"0__ 4-1 65.7+ 4'2 53-1___ 7'1 67"4+ 3-6 55"9__ 6.1 Start of chilli meal 49.8+ 4-2 46.0+ 3.8 35-2+ 4.4 46.6+ 5.3 38.9+ 6-8 Endofmeal 25.1+ 5.0 19.9+ 2-9 19.1+ 4.2 28.0+ 5.9 19.8+ 5-8

Meal duration (s) No-chilli 198"4+ 6-0 189.3+ 6-4 195-8+ 7'3 180'0+ 7'9 194"0+ 7"7 Chilli 266.9+24.1 263.4+27-0 271-9+28.1 247-5+27-1 252.4+24-5

the plain or the chilli ratatouille (Table 3). Subjects did however, find the chilli ratatouille more arousing than the plain ratatouille, F(I, 13) = 28-17, p<0-001. Finally, ratings of burn novelty fell significantly on the second exposure, but remained stable thereafter, F(4,52) = 3"05, p<0.025.

As before, hunger ratings fell significantly throughout the meal (Table 4). There was some variation in ratings across sessions, however, with a marginal interaction of session by time of rating, F(8,104)= 1-74, p=0-09, probably as a result of a tendency for subjects to report lower initial hunger ratings as the study progressed. This effect could result from lowered expectancy, as on later meals subjects knew both the size, quality and type of food they were to receive. As in the previous experiment, there was no significant differences in meal duration between the first and last meal for either the plain or the chilli meal (Table 4).

Finally, to determine whether initial novelty of the chilli burn was related to the magnitude of change in burn liking, burn liking change scores (difference between first and last burn pleasantness ratings) were regressed against familiarity ratings for the capsaicin burn on the first experimental meal. There was no significant linear, quadratic or cubic relationship between initial familiarity and change in pleasantness of the burn.

Preference test There was no evidence of change between the first and last preference test in

arousal, burn intensity or burn pleasantness ratings from ANOVA. Thus changes in liking for the chilli burn in the ratatouille did not appear to result in any change in liking for the capsaicin in tomato juice. Finally, to test whether subjects preferred dose of capsaicin in juice had changed as a result of exposure, the individual's dose rated most pleasant at the point in time where burn intensity was greatest was calculated: preferred doses did not differ significantly between test and retest by Wilcoxon signed rank test; median preferred dose (ppm) on test = 1.0 and on retest = 1.5.

The stimuli were given higher familiarity ratings during retest than during the initial test exposure, as would be expected: test/retest, F(1,13)= 8-17, p<0-025. The largest alterations in familiarity occurred for the strongest concentrations, which


TABLE 5 Familiarity with the intensity of the capsaicin burn for each solution in the original preference test and the one following exposure to the five experimental meals from

experiment 2. All values are means+ SEM

Test/trial 1 ppm 2 ppm 4 ppm 8 ppm

Original test 8.9_+2'8 21-2+6-5 27"6+_8'1 44.1 _+9-5 Retest 8'5_+2"5 8-6_+2.4 10"0_+2"5 15-8_+4.6

TABLE 6 Changes in attitudes towards the experiment stimuli, reported during debriefing in

experiment 2

Positive Negative Question response No change response

Increase in liking for chilli Increase in liking for ratatouille Prepared to try "hotter" food More likely to eat ratatouille again

8 5 1 6 8 0 4 8 2 4 10 0

were judged less familiar in the original test than in the retest (Table 5): retest x trial interaction, F(3,36)=8-14, p<0-001. This confirms the ability of subjects to assess burn familiarity in a laboratory context.

Debriefing No subject reported that they had eaten chilli or strong spicy food in the 24 h

preceding any experimental session. Five subjects ate chilli outside of the experiment, two on two occasions, the other three each on one occasion. However, the effect of these extra exposures may have been small, since the difference in burn pleasantness ratings between meals one and five excluding those five subjects (mean_+ SEM rating change 12.7 +_ 5.2) was still significant, t(8)= 2.40, p<0-025, One-tailed. The incidences of responses in the "Positive" and the combined "No change" and "Negative" categories to questions about changes in likelihood of eating food containing chilli did not differ significantly by chi-squared (Siegel & Castellan, 1988). Again when describing the purpose of the experiment, no subject showed an awareness that changes in liking were of principal interest. Responses again suggested that subjects thought the experiment concerned some form of sensory adaptation to the chilli burn. This type of response was again exemplified in one subject's reply, "To see if after exposure the same amount of chilli burns less".

DISCUSSION

This experiment lends some support to the conclusions from experiment 1, that exposure to the capsaicin burn can enhance subjects' liking for it. There was no


evidence to support the contention that either stimulus novelty or changes in arousal were necessary for this effect to occur. However, the linear decrease in burn intensity suggests a relationship between changes in sensation and liking which was not evident in experiment 1. However, the lack of correlation between overall changes in burn liking and intensity implies that any such relationship is weak at best. The lack of awareness of a change in liking for the burn, as indicated in debriefing, may be attributed to the lack of salience in small changes. There was no evidence that the increase in burn-liking observed in the ratatouille meal phase of the experiment had transferred to the context of tomato juice tests.

GENERAL DISCUSSION

The results of these experiments provide preliminary evidence that positive changes in liking for the chilli burn can be induced using a mere exposure paradigm. They also give some indication of the mechanisms which might underlie such changes. Previously it has been suggested that stimulus novelty is a crucial component of the mere exposure effect (Harrison, 1977; Zajonc, 1980; Bornstein, 1989). However, the results of experiment 2 suggest that novelty to the capsaicin burn alone is not a sufficient condition for enhancement of liking to occur. This is surprising because in the three successful studies of food exposure effects, initial stimulus novelty was suggested to be the most important factor in the generation of enhanced liking or consumption (Birch & Marlin, 1982; Pliner, 1982; Crandall, 1984). Our data show that liking can change in a positive direction whether or not a person believes themselves to be unfamiliar with a particular intensity of the chilli burn. Thus, whereas stimulus novelty might be important in responses to capsaicin in people who are completely naive to chilli, these data indicate that novelty is unlikely to account for increases in preferred chilli concentration.

Three explanations could be extended to account for the pattern of results observed here.

Firstly, it is possible that an association was formed between the chilli burn and either the positive caloric consequences of ingestion or the pleasant taste of the ratatouille. Such mechanisms have been widely invoked to explain food preference development generally (see Rozin & Zellner, 1985). Though either type of association could lead to an increase in liking for the burn, that would not explain the absence of any transfer of liking to the capsaicin in tomato juice context. Calorie-based learning would have to be context-dependent, which has not been widely ac- knowledged to be the case (but see Booth, 1985).

Second, the observed effects could result from some demand characteristic of the experiments. If, for instance, subject's responses were being influenced by a belief that the experiment was intended to demonstrate an increase in liking for the chilli burn, then this could explain the exposure effect obtained in the two studies. However, the problem with this explanation is that the same factors were not operative in the tomato juice preference test, where the same rating was used ("How pleasant is the burning sensation now?"). Moreover, neither "liking" nor "preference" was men- tioned in any subject's statement of their belief about the purpose of the experiment. It would be premature to reject the potential for demand characteristics on this basis, but they do not fully explain the results of the two studies. Third, Schull's (1979) conditioned opponent theory is that an opponent B process (that is an affective


response of opposite sign to its triggering stimulus) can become associated with cues predicting the arrival of its triggering stimulus, in this case the chilli burn. It is possible that such a process occurred here, with some cue from the weekly experimental meals becoming associated with an affectively positive process which developed over the course of exposures, enabling the chilli burn to become liked more. This would also explain the absence of any transfer of liking to the capsaicin in tomato juice, because there is no cue predicting the arrival of the chilli burn (e.g. ratatouille) and so the opponent B process is not triggered. This possible explanation of our results has also been suggested by Rozin (1990) as a candidate mechanism of chilli liking development.

In conclusion, the experiments reported here provide preliminary evidence that liking for the chilli burn can increase as a result of exposure. However, the mechanism through which this change occurs is not clear from these data alone, although explanations based on sensory adaptations, stimulus novelty and arousal seem unlikely. Given the potential for demand characteristics, and the marginal nature of liking change in the second experiment, future research will be needed to clarify the nature of the mechanisms for the development of liking for spicy foods during exposure experiments.

REFERENCES

Berlyne, D. E. (1971) Aesthetics and psychobiology. New York: Appleton-Century-Crofts. Berlyne, D. E. (1973) The vicissitudes of aplopathematic and thelematoscopic pneumatology

(or the hydrography of hedonism). In D. E. Berlyne & R. A. Madsen (Eds.), Pleasure, reward, preference; their nature, determinants, and role in behavior. Pp. 1-28. New York: Academic Press.

Birch, L. L. & Marlin, D. W. (1982) "I don't like it; I never tried it": Effects of exposure on two-year-old children's food preferences. Appetite 3, 353-360.

Bornstein, R. F. (1989) Exposure and affect: overview and meta-analysis of research, 1968-1987. Psycholog&al Bulletin, 106, 265-289.

Domjan, M. (1977) Attenuation and enhancement of neophobia for edible substances. In L. M. Barker, M. R. Best & M. Domjan (Eds.), Learning mechanisms in food selection. Pp. 151-181. Baylor University Press.

Grandall, C. S. (1984) The liking of foods as a result of exposure: Eating doughnuts in Alaska. Journal of Social Psychology, 125, 187-194.

Green, B. G. (1990) Effects of thermal, mechanical and chemical stimulation on the perception of oral irritation. In B. G. Green, J. R. Mason & M. R. Kare (Eds.), Chemical Senses, Volume 2, Irritation. Pp. 171-195. New York: Marcel Dekker.

Harrison, A. A. (1977) Mere exposure. In L. Berkowitz (Ed.), Advances in experimental social psychology, Volume 10. Pp. 39-83. New York: Academic Press.

Lee, K. (1989) Food neophobia; major causes and treatments. Food Technology, 43, 62-72. Maga, J. A. (1975) Capsicum. Critical Reviews #7 Food Science and Nutrition, 7, 177-199. Nasrawi, C. W. & Pangborn, R. M. (1990) Temporal effectiveness of mouth-rinsing on

capsaicin mouth-burn. Physiology and Behavior, 47, 617-623. Pliner, P. (1982) The effects of mere exposure on liking for edible substances. Appetite, 3,

283-290. Pliner, P. (1991) Neophobia in humans and the special status of foods of animal origin.

Appetite, 16, 205-218. Rozin, P. (1990) Getting to like the burn of chilli pepper, biological, psychological and cultural

perspectives. In B. G. Green, J. R. Mason & M. R. Kare (Eds.), Chemical Senses, Volume 2, Irritation. Pp. 231-269. New York: Marcel Dekker.

Rozin, P., Mark, M. & Schiller, D. (1981) The role of desensitization to capsaicin in chilli pepper ingestion and preference. Chemical Senses, 6, 23-31.


Rozin, E & Schiller, D. (1980) The nature and acquisition of a preference for chilli pepper by humans. Motivation and Emotion, 4, 77-101.

Rozin, E & Vollmecke, T. A. (1986) Food likes and dislikes. Annual Reviews of Nutrition, 6, 433-456.

Rozin, P. & Zellner, D. A. (1985) The role of pavlovian conditioning in the acquisition of food likes and dislikes. In N. S. Braveman & P. Bronstein (Eds.), Experimental assessments and clinical applications of conditioned food aversions. Pp. 189-202. New York: New York Academy of Sciences.

Schull, J. (1979) A conditioned opponent theory of pavlovian conditioning and habituation. In G. H. Bower (Ed.), Psychology of learning and Motivation, Volume 13. Pp. 57-90.

Siegel, S. & Castellan, N. J. (1988) Nonparametric statistics in the behavioral sciences. New York: McGraw-Hill Book Company.

Sizer, F. & Harris, N. (1985) The influences of common food additives and temperature on threshold perception of capsaicin. Chemical Senses, 10, 279-286.

Stang, D. J. (1975) When familiarity breeds contempt, absence makes the heart grow fonder: effects of exposure and delay on taste pleasantness ratings. Bulletin of the Psychonomic Society, 6, 273-275.

Stevenson, R. J. & Yeomans, M. R. (1993). Differences in ratings of intensity and pleasantness for the capsaicin burn between chilli likers and non-likers; implications for liking development. Chemical Senses, 18, 471-482.

Weinberger, J. (1984) Reactions to uncertainty: a comparison of three motivational theories. Motivation and Emotion, 8, 1984.

Yeomans, M. R., Wright, P., Macleod, H. A. & Critchley, J. A. (1990) Effects of nalmefene on feeding in humans; dissociation of hunger and palatability. Psychopharmacology, 100, 426-432.

Zajonc, R. B. (1968) Attitudinal effects of mere exposure. Journal of Personality and Social Psychology, Monograph Supplement 9, 1-27.

Received 13 October 1992, revision 24 February 1993

Does exposure enhance liking for the chilli burn

Documents