- Email: [email protected]
The specificity of restrained versus unrestrained eaters' responses to food cues: general desire to eat, or craving for the cued food?
Appetite 41 (2003) 7–13 www.elsevier.com/locate/appet
Research Report
The specificity of restrained versus unrestrained eaters’ responses to food cues: general desire to eat, or craving for the cued food? Ingrid Fedoroffa,*, Janet Polivyb, C. Peter Hermanb a
Department of Psychiatry, University of British Columbia, St Paul’s Hospital, 1081 Burrard Street, Vancouver, BC, Canada V6Z 1Y6 b University of Toronto, Toronto, Canada Received 23 November 2001; revised 14 November 2002; accepted 30 November 2002
Abstract Previous research has shown that exposure to food cues increases eating, especially in restrained eaters. The present study attempted to determine whether this elevated consumption reflects a general desire to eat in response to food cues, or specific desire/craving for the cued food. Restrained and unrestrained eaters were exposed to the smell of either pizza, cookies, or no smell for 10 minutes, were asked to write their thoughts (corresponding to the smell cue) about pizza, cookies or in general, and were then presented with either pizza or cookies to ‘taste and rate’. Results indicated that restrained eaters are more responsive than are unrestrained eaters to pre-eating exposure to smell and thought cues, eating significantly more after such cues. An interaction with food type, however, indicated that restrained eaters ate more only when the food presented to eat was the same as the prior food cues. Self-reported desire to eat, liking, and craving for a particular food increased somewhat for restrained eaters after exposure to the smell and thought of that food. Restrained eaters’ intake of a food that differed from the pre-eating food cues did not differ from their intake following no prior exposure to food cues. Restrained eaters thus showed a highly specific response to exposure to food cues. q 2003 Elsevier Ltd. All rights reserved. Keywords: Cued food; Restrained eaters; Craving
Who can resist the lure of freshly baked chocolate chip cookies, or savory food, just out of the oven? Attractive food cues have been shown to influence consumption even in the laboratory. Fedoroff, Polivy, and Herman (1997) found increased eating following exposure to the odor of pizza and to thinking about pizza. This elevated consumption, however, was confined to restrained eaters (chronic dieters). Unrestrained eaters in that study did not significantly increase their intake following the food cues. In fact, it seems that restrained eaters are particularly responsive to exposure to food cues. For example, restrained eaters salivate more in response to attractive food cues (Klajner, Herman, Polivy, & Chhabra, 1981; Legoff & Spigelman, 1987), and both a 5-minute and a 10-minute exposure to food cues before eating stimulated more food ingestion in dieters than in nondieters (Jansen & van den Hout, 1991; Rogers & Hill, 1989). * Corresponding author. E-mail address: [email protected] (I. Fedoroff). 0195-6663/03/$ - see front matter q 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0195-6663(03)00026-6
The present authors previously investigated the effects of single and multimodal food cues (smell or thoughts of attractive food) on food intake, hunger, craving, and desire to eat (Fedoroff et al., 1997) in restrained and unrestrained eaters. After either smelling pizza cooking for 10 min, or thinking about pizza for 10 min (or both), restrained eaters ate more pizza than after no cues, and more than unrestrained eaters did in each of these conditions (except the no cues condition). Moreover, after exposure to the cue manipulations, restrained eaters reported more desire to eat, hunger, liking and craving for pizza, and desire to eat pizza than did unrestrained eaters (Fedoroff et al., 1997). The pattern of results found by Fedoroff et al. (1997) raises questions about how food cues exert their effect on consumption. Do food cues generate a specific desire for the cued food, or a more general desire to consume something/ anything? Weingarten (1985) speculated that exposure to a food cue evokes a craving for that particular food, in rats at least. Similarly, when previously conditioned food cues are presented to an animal, the animal’s eating is directed
8
I. Fedoroff et al. / Appetite 41 (2003) 7–13
specifically at the previously conditioned source of that food (e.g. the food chute in an operant chamber), even if that food is currently available elsewhere in the chamber (e.g. food dish; Osborne, 1977; Weingarten, 1985). Moreover, animals conditioned with food as a reinforcer do not substitute an alternative consummatory behavior (such as drinking) in a later test situation (Grant & Milgram, 1973). Food cue specificity in humans has also been reported (Cornell, Rodin, & Weingarten, 1989). Males who had already eaten to satiety were given a priming taste of either ice cream or pizza (or no prime), and were then presented with large portions of both foods to eat ad lib. The participants ate more of the food that they had previously tasted than of the food that they had not tasted. In line with these findings, Weingarten has proposed that external food stimuli produce a motivational state (incentive-induced hunger) that may be manifested as a food craving, which may, in turn, be specific to the food expected (Cornell et al., 1989; Weingarten, 1985; Weingarten & Elston, 1990). On the other hand, if cravings reflect a physiological response, it may not be as important that precisely the craved item be supplied; Weingarten and Elston (1990, p. 236) state ‘items which affect a common brain neurochemical system should be capable of satisfying cravings evoked by any other item whose biological activity is mediated by that same system’. Thus, in order to satisfy a food cue-induced craving it may be sufficient for a different but equally attractive food to be offered. The present study was designed to test whether food cues elicit specific cravings or general desires to eat, and whether such desires are especially elevated in restrained eaters. The Cornell et al. (1989) study did find specificity of response to food cues, but although craving was discussed as a possible mediator of this response, it was not actually measured. Moreover, the role of dieting (or other individual difference) status in response to food cues was not assessed. In the present study, we presented two different food cues (smell of pizza or chocolate chip cookies) and test foods (the pizza and cookies) to restrained and unrestrained eaters. We measured both eating and subjective craving, hunger, and desire to eat each food, thus combining ideas presented in the papers by Fedoroff et al., 1997; Cornell et al., 1989. Participants were presented with a pizza smell, or cookie smell, or no pre-eating food cue, and then were given either pizza or cookies to eat ad lib. We predicted that, as in our earlier study, exposure to food cues would result in more eating by restrained eaters than would no exposure. On the premise that a food cue may act as both a signal to eat and a message indicating what to eat (Weingarten, 1985) and that restrained eaters are more sensitive to such messages (see e.g. Polivy, 1996, for a review), we hypothesized that restrained eaters would eat significantly more of the food only when it was the same as the cued food. We expected unrestrained eaters to be less influenced by exposure to food cues and not to eat significantly more food of either type after
pre-eating exposure to food cues, responding primarily to internal signals such as increased insulin secretion or salivation (nonspecific appetitive responses to food cues).
Method Participants. Participants were 132 female first year psychology undergraduate students, aged 17 – 47 years (M ¼ 21:2; SD ¼ 2.2). Participants received course credit for introductory psychology, and gave their written consent to participate in a study of food preferences. The 10-item Revised Restraint Scale (Polivy, Herman, & Howard, 1988) was used to divide participants into restrained ðn ¼ 60Þand unrestrained ðn ¼ 72Þ eaters. The restraint scores ranged from 5 to 28. All subjects were classified as either restrained or unrestrained eaters as determined by scores on the Revised Restraint Scale (Polivy et al., 1988). Those with scores of 14 and below were classified as unrestrained and those who scored 15 and above were classified as restrained. Dieting status was not evaluated. Ethical guidelines as recommended by the American Psychological Association were adhered to. Experimental design. Participants were randomly assigned to one of three cue conditions: no cue, cookie cue, or pizza cue, and were presented with two foods, pizza and chocolate chip cookies, in counterbalanced order. All conditions were crossed with restraint. This yielded a three (cue condition) by two (order of food presented—pizza first versus cookies first) by two (restraint) design. Procedure. Participants were scheduled for a 60 min session between the hours of 11:00 am and 6:00 pm. They were instructed not to eat for 2 h prior to their appointment. All participants were tested individually. The experiment was presented as an investigation into food preferences and taste perception. Students were told that they would be asked to give their opinion on various foods, and that there would be some food-tasting, but they were not informed as to what food would be sampled. Upon arrival at the lab, the participant was seated at a table in a private room, where she signed a consent form and completed a series of visual analogue scales (VAS) rating ‘hunger’, ‘fullness’, and ‘desire to eat’. She also rated her ‘liking’, ‘desire to eat’, and ‘craving’1 for four different foods: roast beef sandwiches, chocolate-chip cookies, pizza, and ice cream. The scales were anchored by ‘not at all’ and ‘extremely’ on a 100 mm line. The ratings for nontargeted foods were included in order to divert the participant’s attention from the experimenter’s interest in the target foods, pizza and chocolate-chip cookies. After completing the VAS, the participant was exposed to one of the three cue conditions (pizza, cookie, no cue) for 10 min. During this 1
Craving was not defined for participants. There are many definitions of craving and it was decided to allow participants to define craving for themselves.
I. Fedoroff et al. / Appetite 41 (2003) 7–13
time she was given written and verbal instructions to write her thoughts about chocolate chip cookies (cookie cue condition) or pizza (the pizza cue condition) or her thoughts in general (no cue condition). The smell of pizza or chocolate-chip cookies came from an oven, positioned in the room but behind and out of sight of the participant. Odor stimulation was controlled so participants were exposed to the odor for a constant amount of time (10 min) with the same amount of food being cooked (one mini pizza or six chocolate chip cookies). The door to the testing room was edged with weatherstripping so as to contain the odor as much as possible. Following cue exposure, the participant completed a second set of VAS as above. She was then given a large plate of freshly baked pizza or chocolate-chip cookies to taste and rate. The pizza plate consisted of three freshly cooked individual frozen pizzas cut into quarters. The cookie portion consisted of a plate of 12 warm chocolate-chip cookies, each about three inches in diameter. The participant was instructed to help herself to the food, and asked to rate it on several sensory variables, including pleasantness of taste and smell, intensity of spicy, sweet and salty tastes, as well as her desire to eat and her craving for this food. She was told that she had 10 min to complete the ratings, that she should take as many pieces as she needed to complete her ratings accurately, and that if she finished her ratings before the time was up, she could help herself to more of the food if she wished, as there was plenty in the lab. After the taste test, she was given more VASs (assessing hunger, fullness, etc., as above) to complete. The participant was given a glass of water (approximately 450 g) to drink from ad libitum. The participant was then given a second plate of food, either pizza or chocolate-chip cookies, whichever she had not received first. The same instructions were given for the second taste test as for the first. Thus, a student in the pizza cue/pizza condition would be exposed to the pizza cue and would then receive pizza first to taste and rate, getting chocolate-chip cookies second. Participants had no reason to expect that a second plate of food would be presented, and would presumably have eaten their fill of the first food. Eating of the second food was thus, in some sense, completely gratuitous. Finally, a last set of VASs, the Revised Restraint Scale, and a post-experimental questionnaire were administered. Participants were then debriefed and asked not to discuss the experiment with other students.
Results The data were analyzed by 3 £ 2 £ 2 ANOVAs. When significant main or interaction effects were found, post-hoc Newman – Keuls tests were performed to determine which means differed significantly. All differences between means
9
Fig. 1. Mean consumption of pizza or chocolate chip cookies as a function of cue conditions and restraint.
presented below are supported by ðp , 0:05Þ significant Newman – Keuls post-hoc tests, unless otherwise stated. Main analysis: first-food intake A three-way ANOVA involving restraint, cue (pizza, cookie, no cue) and food type (pizza, cookie) was conducted for the intake of whichever food the participant received first (immediately after the food cue). Food intake was measured in grams.2 Because analyses of VAS ratings of initial hunger indicated a difference for restraint, an analysis of covariance was computed with pre-cue hunger as the covariate (see the section VAS Ratings Over Time). There was a main effect of cue, Fð2; 119Þ ¼ 3:30; p , 0:05; indicating greater intake in the cued conditions than the nocues condition. There was also a significant main effect for food type, Fð1; 119Þ ¼ 20:71; p , 0:001; pizza intake was greater than was cookie intake. A food type by cue interaction, Fð2; 119Þ ¼ 5:16; p , 0:01; shows elevated intake of pizza following the pizza cue relative to pizza intake following the cookie cue or no cue (see Fig. 1). (The same pattern was evident for cookie intake—greater cookie consumption occurred following the cookie cue than after the pizza cue or no cue—but this difference did not reach significance.) The second-order interaction of restraint by cue by food type was significant, Fð2; 119Þ ¼ 7:07; p , 0:01: Because differences between these means were predicted, planned comparisons were performed. Restrained eaters ate significantly more cookies after being cued with cookies than after the pizza cue, tð120Þ ¼ 2:71; p , 0:05; and they ate more than did unrestrained eaters in this condition, tð120Þ ¼ 2:24; p , 0:05: Similarly, restrained eaters’ pizza intake following pizza cues was significantly greater than after cookie cues, tð120Þ ¼ 3:62; p , 0:05; and than was the pizza intake of unrestrained eaters in this condition, tð120Þ ¼ 2:3; p , 0:05: (see Table 1 for means and standard deviations). There was no effect of specific cues for unrestrained eaters. 2
Analyses were also conducted for calories and number of pieces eaten. More calories for cookies than for pizza were consumed by all participants. Apart from this difference, the pattern of results was the same as that for the analyses conducted on grams consumed. Grams were thus used in order to conform to most of the literature.
10
I. Fedoroff et al. / Appetite 41 (2003) 7–13
Unrestrained eaters did not eat more cookies in the cued than in the uncued condition; pizza intake increased nonsignificantly for unrestrained eaters in both food-cue conditions as compared to the no-cue condition.
Table 1 Means and standard deviations for intake ðgÞ of the food served immediately after cue exposure Food served first No cue Cookie Pizza
Subsidiary analyses: second-food intake To determine whether the influence of the food cues continued during the second serving of food, a three-way ANOVA (restraint, cue, food type) was conducted for the intake (in grams) of the food served second. There was a main effect of food type, Fð1; 120Þ ¼ 9:97; p , 0:01; reflecting the greater intake of pizza overall. Furthermore, there was a marginal main effect for food cues, Fð2; 120Þ ¼ 2:13; p , 0:10; such that intake was greater following exposure to the cue conditions than in the no-cue condition. Post-cue VAS ratings Multivariate analyses of variance (MANOVAs) with the factors restraint and cue condition were conducted for the VAS ratings of desire to eat, liking and craving for cookies and pizza after cue exposure (see Table 2 for craving ratings). The MANOVA for cookie ratings on all of these variables was significant for cue, Fð6; 248Þ ¼ 5:60; p , 0:001; this effect was qualified by a significant interaction of restraint and cue, Fð6; 248Þ ¼ 2:47; p , 0:05: Univariate analyses indicated significant effects for cues on desire to eat, liking and craving for cookies, Fð2; 126Þ ¼ 12:48; p , 0:001; Fð2; 126Þ ¼ 3:87; p , 0:05; and Fð2; 126Þ ¼ 15:84; p , 0:001; respectively. Post-hoc tests revealed elevated ratings following the cookie cue relative to the pizza cue and no-cue condition. The individual univariate analyses for the restraint-by-cue interaction did not reach significance, although there was a marginal interaction on ratings of liking for cookies, Fð2; 126Þ ¼ 2:45; p , 0:10:3 The MANOVA for ratings for pizza was significant for cue condition, Fð6; 248Þ ¼ 5:85; p , 0:001: The univariate analyses revealed significant effects of cue for desire to eat and craving for pizza, Fð2; 126Þ ¼ 5:89; p , 0:001 and Fð2; 126Þ ¼ 11:51; p , 0:001; respectively. Ratings for pizza were significantly higher following the pizza cue than following the cookie cue or no cue. Correlations of VAS ratings and intake Pearson Correlations were computed between intake and ratings of craving (post-cue) for cookies and pizza by restraint (see Table 3). The table shows that for restrained 3
The pattern of means for liking, desire to eat, and craving for cookies was consistent: restrained eaters and unrestrained eaters were the same in the no-cue condition, but restrained eaters’ ratings were higher than were unrestrained eaters’ in the cookie-cue condition and lower in the pizza-cue condition.
Cookie cue
Pizza cue
Cookie
Cookie Pizza
Unrestrained M SD n
78.96b 35.99 12
95.12b 44.11 10
Restrained M SD n
58.58b 17.38 10
85.18b 107.96a 40.32 52.62 9 11
Pizza
71.74b 113.82ab 78.70b 28.91 47.93 29.70 11 10 10 84.71b 45.14 10
63.10b 26.19 10
111.58b 35.29 19 146.03a 37.30 10
Note: Means that share the same superscript are not significantly different from each other. Differences are based on planned comparisons (t-tests, p , 0:05). Table 2 Post cue VAS ratings of craving for cookies and pizza Craving for cookies No cue
Cookie cue
Craving for pizza Pizza cue
No cue
Cookie cue
Pizza cue
Unrestrained M 46.68 74.14 SD 37.45 25.64 n 22 21
55.34 29.21 29
6.86 27.09 22
53.59 32.64 21
75.59 21.97 29
Restrained M 43.22 SD 33.34 n 18
49.90 33.45 21
45.33 35.43 18
40.81 29.20 21
68.38 27.15 21
85.05 11.68 21
eaters craving for pizza was significantly correlated with pizza intake (r 2 ¼ 0:57; p , 0:001) and there was a trend for craving for cookies to be correlated with cookie intake (r 2 ¼ 0:32; p , 0:07). Intake and craving ratings were not correlated for unrestrained eaters. VAS ratings over time Repeated measures ANOVAs with the between-subjects factors restraint, cue, and food served first, measured over time (pre-cue, post-cue, and after eating the first food) were conducted for the subjective ratings of hunger and desire to eat. The main effect of time was significant for hunger ratings, Fð2; 240Þ ¼ 77:32; p , 0:001; indicating that hunger was greater after the cue manipulation than before it and than after eating, Ms ¼ 51:17; 63:87; 37:90; respectively. The time by cue interaction was significant, Fð4; 240Þ ¼ 5:20; p , 0:001; indicating that post-cue hunger was greater in the cued conditions than in the no-cue condition. The time by restraint interaction was significant Fð2; 240Þ ¼ 4:18; p , 0:05: Post-hoc tests indicated that at pre-cue the unrestrained eaters rated themselves as hungrier than restrained eaters do. There were no differences at
I. Fedoroff et al. / Appetite 41 (2003) 7–13 Table 3 Pearson correlations: intake and post cue VAS ratings by restraint Craving for cookies
Craving for pizza
Unrestrained Cookie intake Pizza intake
þ 0.12 þ 0.03
þ 0.18 þ 0.09
Restrained Cookie intake Pizza intake
þ 0.32† 20.14
20.04 þ 0.57*
*p , 0:001; †p , 0:07:
the other time points. The second-order interaction of time, cue, and food-served-first was significant, Fð4; 240Þ ¼ 2:60; p , 0:05; reflecting the fact that hunger was reduced after eating when participants had first received the food with which they had been cued. (Of course, participants ate more when they received the food with which they had been cued.) Ratings of desire to eat showed a similar pattern. A main effect of time (Fð2; 240Þ ¼ 68:11; p , 0:001) reflects the fact that desire to eat was greatest after the cue manipulation, Ms ¼ 49:31; 63:56; 40:06; respectively. This main effect is qualified by a significant interaction between time and restraint, Fð2; 240Þ ¼ 3:75; p , 0:05; unrestrained eaters reported a greater desire to eat than did restrained eaters after exposure to either food cue. A significant time by cue interaction, Fð4; 240Þ ¼ 8:55; p , 0:001; indicates that desire to eat was greatest in the cued conditions (as compared to no-cue condition) immediately after the cue manipulation. Taste-test ratings Subjective ratings from the taste-test were divided into two categories: sensory, including the ratings of pleasantness of taste, smell and texture; and appetite, including the ratings of desire to eat and craving. MANOVAs with restraint and cue were performed on ratings of cookies and pizza. Sensory ratings: The MANOVA for ratings for cookies revealed a marginally significant interaction of restraint and cue, Fð6; 112Þ ¼ 2:16; p , 0:06: The univariate analyses were significant for texture, Fð2; 58Þ ¼ 3:88; p , 0:05; and smell, Fð2; 58Þ ¼ 3:46; p , 0:05; with a slight effect for taste, Fð2; 58Þ ¼ 2:08; p , 0:20: The post-hoc tests were significant only for ratings of the smell, such that restrained eaters rated the pleasantness of the smell of cookies higher following the cookie cue than in the no-cue condition, although a similar pattern of means was obtained for taste and texture as well. The MANOVA for the sensory ratings for pizza was similar but with a weaker interaction of restraint and cue, Fð6; 120Þ ¼ 1:70; p , 0:20; the univariate analyses were not significant.
11
Appetite ratings: The MANOVA for cookie ratings was significant for cue, Fð4; 114Þ ¼ 3:14; p , 0:05: The univariate analyses for cue were significant for desire to eat, Fð2; 58Þ ¼ 6:31; p , 0:01; and craving for cookies, Fð2; 58Þ ¼ 5:63; p , 0:01: Desire to eat and craving for cookies were significantly greater following the cookie cue than following the pizza cue or no cue. The MANOVA for ratings of appetite for pizza was not significant.
Discussion As in Fedoroff et al. (1997), restrained eaters in the present study ate more following exposure to food cues, for both the same food used by Fedoroff et al. (pizza) and a different food (cookies). These results replicate the enhanced response of restrained eaters to food cues found by Fedoroff et al. and extends this finding. As predicted, restrained eaters showed cue specificity, eating more only when they had previously been cued with that food. If they had not been cued with the food before it was made available to them to eat, they actually ate less than did the unrestrained eaters. Moreover, restrained eaters experienced a corresponding increase in appetite (hunger, desire and craving) for cookies post-cue, and judged the sensory qualities of the food (pleasantness of smell, and to a lesser extent, texture) more positively only when the food they were eating was the same as the preceding cue (cookies). The pizza cue generated a greater self-reported appetite for pizza in both restrained and unrestrained eaters. Restrained eaters’ ratings of craving for pizza and cookies were correlated with their respective intakes. Unrestrained eaters showed little differential food intake response to the cue conditions. The self-report ratings for cookies showed more consistent differences between restrained and unrestrained eaters than did those for the pizza. Fedoroff et al. (1997) found that restrained (but not unrestrained) eaters reported greater appetite and sensory appreciation for pizza following exposure to a pizza cue, as well as increased pizza consumption. In the present study, both restrained and unrestrained eaters reported increased appetite for pizza following the pizza cue (as compared to the cookie cue or no cue), but only the restrained eaters went on to eat more pizza. Although, restrained eaters’ rating of craving for pizza was significantly correlated with pizza intake. Thus, the effect of cues on eating was similar in the present study and the earlier study, but the effect on subjective reports was somewhat mixed in that unrestrained eaters also reported more appetite for pizza but the correlation between craving and pizza intake was not significant. Weingarten (1985) proposed that cues conditioned to a particular food produce an appetitive motivational state that may manifest itself as a craving for that food, and may be specific to the particular food. The subjective ratings of restrained eaters in the present study lend some support to
12
I. Fedoroff et al. / Appetite 41 (2003) 7–13
this proposal; restrained eaters reported a somewhat greater appetite (liking, desire to eat, and craving) for cookies following the cookie cue, and like unrestrained eaters, increased their appetite for pizza only after the pizza cue. They also perceived the sensory qualities of the cookies to be more pleasant after exposure to the cookie cue. Restrained eaters thus found the food to be more desirable only in the conditions wherein the prior cues matched the food presented. They then went on to eat more of the cued foods. Thus, for restrained eaters, the food cue may indeed act as both a signal to eat and an indicator of what to eat, as Weingarten proposed. Alternatively, a cognitive theory originally developed to explain drug urges and drug use behavior (Tiffany, 1990, 1992) may be useful to account for dieters’ increased eating following exposure to food cues. Tiffany also utilizes a model of incentive-induced appetite but it does not depend on a history of deprivation. Tiffany proposes that drug use in the addict is largely controlled by automatic processes and involves no urges or cravings. However, urges are invoked under certain circumstances. For example, when an addict tries to stop drug use, an urge or craving develops which is associated with relapse. It is only when the automatic behavior (to use drugs) is being resisted that the action becomes nonautomatic and urges are likely to develop. Moreover, such urges are particularly likely to occur in the presence of cues associated with taking the substance, in this case food cues (Woods, 1991). A dieter who is consciously attempting to reduce her food intake, when faced with the very type of food she is trying to resist eating—pizza and cookies are both rated as forbidden foods by dieters (Knight & Boland, 1989)—develops an urge or an appetite to eat which undermines her diet and promotes even greater eating. A nondieter, on the other hand, when presented with an appetizing food, does not experience the urges or appetite to the same extent because she is not attempting to deny herself those foods. The nondieter’s automatic response would be to eat in accord with hunger and satiety signals and not to experience intense urges which could push eating beyond the normal limits of satiety. This theory offers an alternative framework to investigate cognitive processing and key characteristics of behaviors such as drug use and perhaps eating. The evidence from the present studies showing restrained eaters to have greater urges to eat particularly after pre-exposure to specific food cues suggests that this explanation may have some merit. However, it is necessary to design experiments to provide critical tests of the theory before any conclusions can be drawn about restrained eating. Understanding the influence of food cues on hunger, craving, and eating may have implications for eating disorders. There is abundant evidence that eating disordered patients are more sensitive to food-, weight-, and shaperelated cues in the environment (Ward, Bulik, & Johnston, 1996). Vitousek and Hollon (1990) hypothesized that individuals with eating disorders develop organized
cognitive structures around the issues of food, weight and shape. There is a growing literature demonstrating that anorexic and bulimic patients exhibit an attentional bias towards food-, weight-, and shape-related information (Carter, Bulik, McIntosh, & Joyce, 2000; Flynn, 1999; Green, Corr, & DeSilva, 1999), as well as a memory bias for such information (Hermans, Pieter, & Eelen, 1998; King, Polivy, & Herman, 1991). This heightened sensitivity toward food-related information may place eating disordered individuals in what amounts to a perpetual state of exposure to food cues, producing craving, which may contribute to triggering eating binges in bulimic patients. In conclusion, the present study demonstrates that exposure to food cues increases consumption by restrained eaters only when the food cues match the food available to be consumed. There was also evidence for specificity in appetite and sensory ratings of the food (particularly the cookies), suggesting that food-specific cravings can be triggered by food-specific cues.
Acknowledgements This research comprised part of the requirements for a PhD degree for the first author. Support was provided by a SSHRC grant to JP and CPH. Address correspondence to Ingrid C. Fedoroff, Department of Psychiatry, St Paul’s Hospital, 1081 Burrard St, Vancouver, BC, V6Z 1Y6.
References Carter, F. A., Bulik, C., McIntosh, V. V., & Joyce, P. R. (2000). Changes on the stroop test following treatment: Relation to word type, treatment condition, and treatment outcome among women with bulimia nervosa. International Journal of Eating Disorders, 28, 349– 355. Cornell, C. E., Rodin, J., & Weingarten, H. (1989). Stimulus-induced eating when satiated. Physiology and Behavior, 45, 695–704. Fedoroff, I., Polivy, J., & Herman, C. P. (1997). The effect of pre-exposure to food cues on the eating behavior of restrained and unrestrained eaters. Appetite, 28, 33–47. Flynn, S. V. (1999). Do disorder-relevant cognitive biases endure in recovered bulimics? Behavior Therapy, 30, 541–553. Grant, D. P., & Milgram, N. W. (1973). Plasticity of normal feeding: situational and individual factors. Canadian Journal of Psychology, 24, 305 –316. Green, M., Corr, P., & DeSilva, L. (1999). Impaired color naming of body shape-related words in anorexia nervosa: affective valence or associative priming? Cognitive Therapy and Research, 23, 413–422. Hermans, D., Pieters, G., & Eelen, P. (1998). Implicit and explicit memory for shape, body weight, and food-related words in patients with anorexia nervosa and nondieting controls. Journal of Abnormal Psychology, 107, 193 –202. Jansen, A., & Van den Hout, M. (1991). On being led into temptation: counterregulation of dieters after smelling a preload. Addictive Behaviors, 16, 247 –253. King, G., Polivy, J., & Herman, C. P. (1991). Cognitive aspects of dietary restraint: effects on person memory. International Journal of Eating Disorders, 10, 313 –322.
I. Fedoroff et al. / Appetite 41 (2003) 7–13 Klajner, F., Herman, C. P., Polivy, J., & Chhabra, R. (1981). Human obesity, dieting, and anticipatory salivation to food. Physiology and Behavior, 27, 195–198. Knight, L., & Boland, F. (1989). Restrained eating: An experimental disentanglement of disinhibiting variables of calories and food type. Journal of Abnormal Psychology, 98, 412 –420. Legoff, D. B., & Spigelman, M. N. (1987). Salivary response to olfactory food stimuli as a function of dietary restraint and body weight. Appetite, 8, 29– 36. Osborne, S. R. (1977). The free food (contrafreeloading) phenomenon: A review and analysis. Animal Learning and Behavior, 5, 221 –235. Polivy, J. (1996). Psychological consequences of food restriction. Journal of the American Dietetic Association, 96, 589–594. Polivy, J., Herman, C. P., & Howard, K. I. (1988). The restraint scale: assessment of dieting. In M. Hersen, & A. S. Bellack (Eds.), Dictionary of behavioral assessment techniques (pp. 377 –380). New York: Pergamon Press. Rogers, P. J., & Hill, A. (1989). Breakdown of dietary restraint following mere exposure to food stimuli: interrelationships between restraint,
13
hunger, salivation and food intake. Addictive Behaviors, 14, 387– 397. Tiffany, S. T. (1990). A cognitive model of drug urges and drug-use behavior: role of automatic and nonautomatic processes. Psychological Review, 97, 147– 168. Tiffany, S. T. (1992). A critique of contemporary urge and craving research: methodological, psychometric, and theoretical issues. Advanced Behavior Research Therapy, 14, 123–139. Vitousek, K., & Hollon, S. D. (1990). The investigation of schematic content and processing in eating disorders. Cognitive Therapy and Research, 14, 135–143. Ward, T., Bulik, C. M., & Johnston, L. (1996). Return of the suppressed: mental control and bulimia nervosa. Behaviour Change, 13, 79–90. Weingarten, H. P. (1985). Stimulus control of eating: implications for a two-factor theory of hunger. Appetite, 6, 387– 401. Weingarten, H., & Elston, D. (1990). The phenomenology of food cravings. Appetite, 15, 231–246. Woods, S. C. (1991). The eating paradox: how we tolerate food. Psychological Review, 98, 488–505.
Research Report
The specificity of restrained versus unrestrained eaters’ responses to food cues: general desire to eat, or craving for the cued food? Ingrid Fedoroffa,*, Janet Polivyb, C. Peter Hermanb a
Department of Psychiatry, University of British Columbia, St Paul’s Hospital, 1081 Burrard Street, Vancouver, BC, Canada V6Z 1Y6 b University of Toronto, Toronto, Canada Received 23 November 2001; revised 14 November 2002; accepted 30 November 2002
Abstract Previous research has shown that exposure to food cues increases eating, especially in restrained eaters. The present study attempted to determine whether this elevated consumption reflects a general desire to eat in response to food cues, or specific desire/craving for the cued food. Restrained and unrestrained eaters were exposed to the smell of either pizza, cookies, or no smell for 10 minutes, were asked to write their thoughts (corresponding to the smell cue) about pizza, cookies or in general, and were then presented with either pizza or cookies to ‘taste and rate’. Results indicated that restrained eaters are more responsive than are unrestrained eaters to pre-eating exposure to smell and thought cues, eating significantly more after such cues. An interaction with food type, however, indicated that restrained eaters ate more only when the food presented to eat was the same as the prior food cues. Self-reported desire to eat, liking, and craving for a particular food increased somewhat for restrained eaters after exposure to the smell and thought of that food. Restrained eaters’ intake of a food that differed from the pre-eating food cues did not differ from their intake following no prior exposure to food cues. Restrained eaters thus showed a highly specific response to exposure to food cues. q 2003 Elsevier Ltd. All rights reserved. Keywords: Cued food; Restrained eaters; Craving
Who can resist the lure of freshly baked chocolate chip cookies, or savory food, just out of the oven? Attractive food cues have been shown to influence consumption even in the laboratory. Fedoroff, Polivy, and Herman (1997) found increased eating following exposure to the odor of pizza and to thinking about pizza. This elevated consumption, however, was confined to restrained eaters (chronic dieters). Unrestrained eaters in that study did not significantly increase their intake following the food cues. In fact, it seems that restrained eaters are particularly responsive to exposure to food cues. For example, restrained eaters salivate more in response to attractive food cues (Klajner, Herman, Polivy, & Chhabra, 1981; Legoff & Spigelman, 1987), and both a 5-minute and a 10-minute exposure to food cues before eating stimulated more food ingestion in dieters than in nondieters (Jansen & van den Hout, 1991; Rogers & Hill, 1989). * Corresponding author. E-mail address: [email protected] (I. Fedoroff). 0195-6663/03/$ - see front matter q 2003 Elsevier Ltd. All rights reserved. doi:10.1016/S0195-6663(03)00026-6
The present authors previously investigated the effects of single and multimodal food cues (smell or thoughts of attractive food) on food intake, hunger, craving, and desire to eat (Fedoroff et al., 1997) in restrained and unrestrained eaters. After either smelling pizza cooking for 10 min, or thinking about pizza for 10 min (or both), restrained eaters ate more pizza than after no cues, and more than unrestrained eaters did in each of these conditions (except the no cues condition). Moreover, after exposure to the cue manipulations, restrained eaters reported more desire to eat, hunger, liking and craving for pizza, and desire to eat pizza than did unrestrained eaters (Fedoroff et al., 1997). The pattern of results found by Fedoroff et al. (1997) raises questions about how food cues exert their effect on consumption. Do food cues generate a specific desire for the cued food, or a more general desire to consume something/ anything? Weingarten (1985) speculated that exposure to a food cue evokes a craving for that particular food, in rats at least. Similarly, when previously conditioned food cues are presented to an animal, the animal’s eating is directed
8
I. Fedoroff et al. / Appetite 41 (2003) 7–13
specifically at the previously conditioned source of that food (e.g. the food chute in an operant chamber), even if that food is currently available elsewhere in the chamber (e.g. food dish; Osborne, 1977; Weingarten, 1985). Moreover, animals conditioned with food as a reinforcer do not substitute an alternative consummatory behavior (such as drinking) in a later test situation (Grant & Milgram, 1973). Food cue specificity in humans has also been reported (Cornell, Rodin, & Weingarten, 1989). Males who had already eaten to satiety were given a priming taste of either ice cream or pizza (or no prime), and were then presented with large portions of both foods to eat ad lib. The participants ate more of the food that they had previously tasted than of the food that they had not tasted. In line with these findings, Weingarten has proposed that external food stimuli produce a motivational state (incentive-induced hunger) that may be manifested as a food craving, which may, in turn, be specific to the food expected (Cornell et al., 1989; Weingarten, 1985; Weingarten & Elston, 1990). On the other hand, if cravings reflect a physiological response, it may not be as important that precisely the craved item be supplied; Weingarten and Elston (1990, p. 236) state ‘items which affect a common brain neurochemical system should be capable of satisfying cravings evoked by any other item whose biological activity is mediated by that same system’. Thus, in order to satisfy a food cue-induced craving it may be sufficient for a different but equally attractive food to be offered. The present study was designed to test whether food cues elicit specific cravings or general desires to eat, and whether such desires are especially elevated in restrained eaters. The Cornell et al. (1989) study did find specificity of response to food cues, but although craving was discussed as a possible mediator of this response, it was not actually measured. Moreover, the role of dieting (or other individual difference) status in response to food cues was not assessed. In the present study, we presented two different food cues (smell of pizza or chocolate chip cookies) and test foods (the pizza and cookies) to restrained and unrestrained eaters. We measured both eating and subjective craving, hunger, and desire to eat each food, thus combining ideas presented in the papers by Fedoroff et al., 1997; Cornell et al., 1989. Participants were presented with a pizza smell, or cookie smell, or no pre-eating food cue, and then were given either pizza or cookies to eat ad lib. We predicted that, as in our earlier study, exposure to food cues would result in more eating by restrained eaters than would no exposure. On the premise that a food cue may act as both a signal to eat and a message indicating what to eat (Weingarten, 1985) and that restrained eaters are more sensitive to such messages (see e.g. Polivy, 1996, for a review), we hypothesized that restrained eaters would eat significantly more of the food only when it was the same as the cued food. We expected unrestrained eaters to be less influenced by exposure to food cues and not to eat significantly more food of either type after
pre-eating exposure to food cues, responding primarily to internal signals such as increased insulin secretion or salivation (nonspecific appetitive responses to food cues).
Method Participants. Participants were 132 female first year psychology undergraduate students, aged 17 – 47 years (M ¼ 21:2; SD ¼ 2.2). Participants received course credit for introductory psychology, and gave their written consent to participate in a study of food preferences. The 10-item Revised Restraint Scale (Polivy, Herman, & Howard, 1988) was used to divide participants into restrained ðn ¼ 60Þand unrestrained ðn ¼ 72Þ eaters. The restraint scores ranged from 5 to 28. All subjects were classified as either restrained or unrestrained eaters as determined by scores on the Revised Restraint Scale (Polivy et al., 1988). Those with scores of 14 and below were classified as unrestrained and those who scored 15 and above were classified as restrained. Dieting status was not evaluated. Ethical guidelines as recommended by the American Psychological Association were adhered to. Experimental design. Participants were randomly assigned to one of three cue conditions: no cue, cookie cue, or pizza cue, and were presented with two foods, pizza and chocolate chip cookies, in counterbalanced order. All conditions were crossed with restraint. This yielded a three (cue condition) by two (order of food presented—pizza first versus cookies first) by two (restraint) design. Procedure. Participants were scheduled for a 60 min session between the hours of 11:00 am and 6:00 pm. They were instructed not to eat for 2 h prior to their appointment. All participants were tested individually. The experiment was presented as an investigation into food preferences and taste perception. Students were told that they would be asked to give their opinion on various foods, and that there would be some food-tasting, but they were not informed as to what food would be sampled. Upon arrival at the lab, the participant was seated at a table in a private room, where she signed a consent form and completed a series of visual analogue scales (VAS) rating ‘hunger’, ‘fullness’, and ‘desire to eat’. She also rated her ‘liking’, ‘desire to eat’, and ‘craving’1 for four different foods: roast beef sandwiches, chocolate-chip cookies, pizza, and ice cream. The scales were anchored by ‘not at all’ and ‘extremely’ on a 100 mm line. The ratings for nontargeted foods were included in order to divert the participant’s attention from the experimenter’s interest in the target foods, pizza and chocolate-chip cookies. After completing the VAS, the participant was exposed to one of the three cue conditions (pizza, cookie, no cue) for 10 min. During this 1
Craving was not defined for participants. There are many definitions of craving and it was decided to allow participants to define craving for themselves.
I. Fedoroff et al. / Appetite 41 (2003) 7–13
time she was given written and verbal instructions to write her thoughts about chocolate chip cookies (cookie cue condition) or pizza (the pizza cue condition) or her thoughts in general (no cue condition). The smell of pizza or chocolate-chip cookies came from an oven, positioned in the room but behind and out of sight of the participant. Odor stimulation was controlled so participants were exposed to the odor for a constant amount of time (10 min) with the same amount of food being cooked (one mini pizza or six chocolate chip cookies). The door to the testing room was edged with weatherstripping so as to contain the odor as much as possible. Following cue exposure, the participant completed a second set of VAS as above. She was then given a large plate of freshly baked pizza or chocolate-chip cookies to taste and rate. The pizza plate consisted of three freshly cooked individual frozen pizzas cut into quarters. The cookie portion consisted of a plate of 12 warm chocolate-chip cookies, each about three inches in diameter. The participant was instructed to help herself to the food, and asked to rate it on several sensory variables, including pleasantness of taste and smell, intensity of spicy, sweet and salty tastes, as well as her desire to eat and her craving for this food. She was told that she had 10 min to complete the ratings, that she should take as many pieces as she needed to complete her ratings accurately, and that if she finished her ratings before the time was up, she could help herself to more of the food if she wished, as there was plenty in the lab. After the taste test, she was given more VASs (assessing hunger, fullness, etc., as above) to complete. The participant was given a glass of water (approximately 450 g) to drink from ad libitum. The participant was then given a second plate of food, either pizza or chocolate-chip cookies, whichever she had not received first. The same instructions were given for the second taste test as for the first. Thus, a student in the pizza cue/pizza condition would be exposed to the pizza cue and would then receive pizza first to taste and rate, getting chocolate-chip cookies second. Participants had no reason to expect that a second plate of food would be presented, and would presumably have eaten their fill of the first food. Eating of the second food was thus, in some sense, completely gratuitous. Finally, a last set of VASs, the Revised Restraint Scale, and a post-experimental questionnaire were administered. Participants were then debriefed and asked not to discuss the experiment with other students.
Results The data were analyzed by 3 £ 2 £ 2 ANOVAs. When significant main or interaction effects were found, post-hoc Newman – Keuls tests were performed to determine which means differed significantly. All differences between means
9
Fig. 1. Mean consumption of pizza or chocolate chip cookies as a function of cue conditions and restraint.
presented below are supported by ðp , 0:05Þ significant Newman – Keuls post-hoc tests, unless otherwise stated. Main analysis: first-food intake A three-way ANOVA involving restraint, cue (pizza, cookie, no cue) and food type (pizza, cookie) was conducted for the intake of whichever food the participant received first (immediately after the food cue). Food intake was measured in grams.2 Because analyses of VAS ratings of initial hunger indicated a difference for restraint, an analysis of covariance was computed with pre-cue hunger as the covariate (see the section VAS Ratings Over Time). There was a main effect of cue, Fð2; 119Þ ¼ 3:30; p , 0:05; indicating greater intake in the cued conditions than the nocues condition. There was also a significant main effect for food type, Fð1; 119Þ ¼ 20:71; p , 0:001; pizza intake was greater than was cookie intake. A food type by cue interaction, Fð2; 119Þ ¼ 5:16; p , 0:01; shows elevated intake of pizza following the pizza cue relative to pizza intake following the cookie cue or no cue (see Fig. 1). (The same pattern was evident for cookie intake—greater cookie consumption occurred following the cookie cue than after the pizza cue or no cue—but this difference did not reach significance.) The second-order interaction of restraint by cue by food type was significant, Fð2; 119Þ ¼ 7:07; p , 0:01: Because differences between these means were predicted, planned comparisons were performed. Restrained eaters ate significantly more cookies after being cued with cookies than after the pizza cue, tð120Þ ¼ 2:71; p , 0:05; and they ate more than did unrestrained eaters in this condition, tð120Þ ¼ 2:24; p , 0:05: Similarly, restrained eaters’ pizza intake following pizza cues was significantly greater than after cookie cues, tð120Þ ¼ 3:62; p , 0:05; and than was the pizza intake of unrestrained eaters in this condition, tð120Þ ¼ 2:3; p , 0:05: (see Table 1 for means and standard deviations). There was no effect of specific cues for unrestrained eaters. 2
Analyses were also conducted for calories and number of pieces eaten. More calories for cookies than for pizza were consumed by all participants. Apart from this difference, the pattern of results was the same as that for the analyses conducted on grams consumed. Grams were thus used in order to conform to most of the literature.
10
I. Fedoroff et al. / Appetite 41 (2003) 7–13
Unrestrained eaters did not eat more cookies in the cued than in the uncued condition; pizza intake increased nonsignificantly for unrestrained eaters in both food-cue conditions as compared to the no-cue condition.
Table 1 Means and standard deviations for intake ðgÞ of the food served immediately after cue exposure Food served first No cue Cookie Pizza
Subsidiary analyses: second-food intake To determine whether the influence of the food cues continued during the second serving of food, a three-way ANOVA (restraint, cue, food type) was conducted for the intake (in grams) of the food served second. There was a main effect of food type, Fð1; 120Þ ¼ 9:97; p , 0:01; reflecting the greater intake of pizza overall. Furthermore, there was a marginal main effect for food cues, Fð2; 120Þ ¼ 2:13; p , 0:10; such that intake was greater following exposure to the cue conditions than in the no-cue condition. Post-cue VAS ratings Multivariate analyses of variance (MANOVAs) with the factors restraint and cue condition were conducted for the VAS ratings of desire to eat, liking and craving for cookies and pizza after cue exposure (see Table 2 for craving ratings). The MANOVA for cookie ratings on all of these variables was significant for cue, Fð6; 248Þ ¼ 5:60; p , 0:001; this effect was qualified by a significant interaction of restraint and cue, Fð6; 248Þ ¼ 2:47; p , 0:05: Univariate analyses indicated significant effects for cues on desire to eat, liking and craving for cookies, Fð2; 126Þ ¼ 12:48; p , 0:001; Fð2; 126Þ ¼ 3:87; p , 0:05; and Fð2; 126Þ ¼ 15:84; p , 0:001; respectively. Post-hoc tests revealed elevated ratings following the cookie cue relative to the pizza cue and no-cue condition. The individual univariate analyses for the restraint-by-cue interaction did not reach significance, although there was a marginal interaction on ratings of liking for cookies, Fð2; 126Þ ¼ 2:45; p , 0:10:3 The MANOVA for ratings for pizza was significant for cue condition, Fð6; 248Þ ¼ 5:85; p , 0:001: The univariate analyses revealed significant effects of cue for desire to eat and craving for pizza, Fð2; 126Þ ¼ 5:89; p , 0:001 and Fð2; 126Þ ¼ 11:51; p , 0:001; respectively. Ratings for pizza were significantly higher following the pizza cue than following the cookie cue or no cue. Correlations of VAS ratings and intake Pearson Correlations were computed between intake and ratings of craving (post-cue) for cookies and pizza by restraint (see Table 3). The table shows that for restrained 3
The pattern of means for liking, desire to eat, and craving for cookies was consistent: restrained eaters and unrestrained eaters were the same in the no-cue condition, but restrained eaters’ ratings were higher than were unrestrained eaters’ in the cookie-cue condition and lower in the pizza-cue condition.
Cookie cue
Pizza cue
Cookie
Cookie Pizza
Unrestrained M SD n
78.96b 35.99 12
95.12b 44.11 10
Restrained M SD n
58.58b 17.38 10
85.18b 107.96a 40.32 52.62 9 11
Pizza
71.74b 113.82ab 78.70b 28.91 47.93 29.70 11 10 10 84.71b 45.14 10
63.10b 26.19 10
111.58b 35.29 19 146.03a 37.30 10
Note: Means that share the same superscript are not significantly different from each other. Differences are based on planned comparisons (t-tests, p , 0:05). Table 2 Post cue VAS ratings of craving for cookies and pizza Craving for cookies No cue
Cookie cue
Craving for pizza Pizza cue
No cue
Cookie cue
Pizza cue
Unrestrained M 46.68 74.14 SD 37.45 25.64 n 22 21
55.34 29.21 29
6.86 27.09 22
53.59 32.64 21
75.59 21.97 29
Restrained M 43.22 SD 33.34 n 18
49.90 33.45 21
45.33 35.43 18
40.81 29.20 21
68.38 27.15 21
85.05 11.68 21
eaters craving for pizza was significantly correlated with pizza intake (r 2 ¼ 0:57; p , 0:001) and there was a trend for craving for cookies to be correlated with cookie intake (r 2 ¼ 0:32; p , 0:07). Intake and craving ratings were not correlated for unrestrained eaters. VAS ratings over time Repeated measures ANOVAs with the between-subjects factors restraint, cue, and food served first, measured over time (pre-cue, post-cue, and after eating the first food) were conducted for the subjective ratings of hunger and desire to eat. The main effect of time was significant for hunger ratings, Fð2; 240Þ ¼ 77:32; p , 0:001; indicating that hunger was greater after the cue manipulation than before it and than after eating, Ms ¼ 51:17; 63:87; 37:90; respectively. The time by cue interaction was significant, Fð4; 240Þ ¼ 5:20; p , 0:001; indicating that post-cue hunger was greater in the cued conditions than in the no-cue condition. The time by restraint interaction was significant Fð2; 240Þ ¼ 4:18; p , 0:05: Post-hoc tests indicated that at pre-cue the unrestrained eaters rated themselves as hungrier than restrained eaters do. There were no differences at
I. Fedoroff et al. / Appetite 41 (2003) 7–13 Table 3 Pearson correlations: intake and post cue VAS ratings by restraint Craving for cookies
Craving for pizza
Unrestrained Cookie intake Pizza intake
þ 0.12 þ 0.03
þ 0.18 þ 0.09
Restrained Cookie intake Pizza intake
þ 0.32† 20.14
20.04 þ 0.57*
*p , 0:001; †p , 0:07:
the other time points. The second-order interaction of time, cue, and food-served-first was significant, Fð4; 240Þ ¼ 2:60; p , 0:05; reflecting the fact that hunger was reduced after eating when participants had first received the food with which they had been cued. (Of course, participants ate more when they received the food with which they had been cued.) Ratings of desire to eat showed a similar pattern. A main effect of time (Fð2; 240Þ ¼ 68:11; p , 0:001) reflects the fact that desire to eat was greatest after the cue manipulation, Ms ¼ 49:31; 63:56; 40:06; respectively. This main effect is qualified by a significant interaction between time and restraint, Fð2; 240Þ ¼ 3:75; p , 0:05; unrestrained eaters reported a greater desire to eat than did restrained eaters after exposure to either food cue. A significant time by cue interaction, Fð4; 240Þ ¼ 8:55; p , 0:001; indicates that desire to eat was greatest in the cued conditions (as compared to no-cue condition) immediately after the cue manipulation. Taste-test ratings Subjective ratings from the taste-test were divided into two categories: sensory, including the ratings of pleasantness of taste, smell and texture; and appetite, including the ratings of desire to eat and craving. MANOVAs with restraint and cue were performed on ratings of cookies and pizza. Sensory ratings: The MANOVA for ratings for cookies revealed a marginally significant interaction of restraint and cue, Fð6; 112Þ ¼ 2:16; p , 0:06: The univariate analyses were significant for texture, Fð2; 58Þ ¼ 3:88; p , 0:05; and smell, Fð2; 58Þ ¼ 3:46; p , 0:05; with a slight effect for taste, Fð2; 58Þ ¼ 2:08; p , 0:20: The post-hoc tests were significant only for ratings of the smell, such that restrained eaters rated the pleasantness of the smell of cookies higher following the cookie cue than in the no-cue condition, although a similar pattern of means was obtained for taste and texture as well. The MANOVA for the sensory ratings for pizza was similar but with a weaker interaction of restraint and cue, Fð6; 120Þ ¼ 1:70; p , 0:20; the univariate analyses were not significant.
11
Appetite ratings: The MANOVA for cookie ratings was significant for cue, Fð4; 114Þ ¼ 3:14; p , 0:05: The univariate analyses for cue were significant for desire to eat, Fð2; 58Þ ¼ 6:31; p , 0:01; and craving for cookies, Fð2; 58Þ ¼ 5:63; p , 0:01: Desire to eat and craving for cookies were significantly greater following the cookie cue than following the pizza cue or no cue. The MANOVA for ratings of appetite for pizza was not significant.
Discussion As in Fedoroff et al. (1997), restrained eaters in the present study ate more following exposure to food cues, for both the same food used by Fedoroff et al. (pizza) and a different food (cookies). These results replicate the enhanced response of restrained eaters to food cues found by Fedoroff et al. and extends this finding. As predicted, restrained eaters showed cue specificity, eating more only when they had previously been cued with that food. If they had not been cued with the food before it was made available to them to eat, they actually ate less than did the unrestrained eaters. Moreover, restrained eaters experienced a corresponding increase in appetite (hunger, desire and craving) for cookies post-cue, and judged the sensory qualities of the food (pleasantness of smell, and to a lesser extent, texture) more positively only when the food they were eating was the same as the preceding cue (cookies). The pizza cue generated a greater self-reported appetite for pizza in both restrained and unrestrained eaters. Restrained eaters’ ratings of craving for pizza and cookies were correlated with their respective intakes. Unrestrained eaters showed little differential food intake response to the cue conditions. The self-report ratings for cookies showed more consistent differences between restrained and unrestrained eaters than did those for the pizza. Fedoroff et al. (1997) found that restrained (but not unrestrained) eaters reported greater appetite and sensory appreciation for pizza following exposure to a pizza cue, as well as increased pizza consumption. In the present study, both restrained and unrestrained eaters reported increased appetite for pizza following the pizza cue (as compared to the cookie cue or no cue), but only the restrained eaters went on to eat more pizza. Although, restrained eaters’ rating of craving for pizza was significantly correlated with pizza intake. Thus, the effect of cues on eating was similar in the present study and the earlier study, but the effect on subjective reports was somewhat mixed in that unrestrained eaters also reported more appetite for pizza but the correlation between craving and pizza intake was not significant. Weingarten (1985) proposed that cues conditioned to a particular food produce an appetitive motivational state that may manifest itself as a craving for that food, and may be specific to the particular food. The subjective ratings of restrained eaters in the present study lend some support to
12
I. Fedoroff et al. / Appetite 41 (2003) 7–13
this proposal; restrained eaters reported a somewhat greater appetite (liking, desire to eat, and craving) for cookies following the cookie cue, and like unrestrained eaters, increased their appetite for pizza only after the pizza cue. They also perceived the sensory qualities of the cookies to be more pleasant after exposure to the cookie cue. Restrained eaters thus found the food to be more desirable only in the conditions wherein the prior cues matched the food presented. They then went on to eat more of the cued foods. Thus, for restrained eaters, the food cue may indeed act as both a signal to eat and an indicator of what to eat, as Weingarten proposed. Alternatively, a cognitive theory originally developed to explain drug urges and drug use behavior (Tiffany, 1990, 1992) may be useful to account for dieters’ increased eating following exposure to food cues. Tiffany also utilizes a model of incentive-induced appetite but it does not depend on a history of deprivation. Tiffany proposes that drug use in the addict is largely controlled by automatic processes and involves no urges or cravings. However, urges are invoked under certain circumstances. For example, when an addict tries to stop drug use, an urge or craving develops which is associated with relapse. It is only when the automatic behavior (to use drugs) is being resisted that the action becomes nonautomatic and urges are likely to develop. Moreover, such urges are particularly likely to occur in the presence of cues associated with taking the substance, in this case food cues (Woods, 1991). A dieter who is consciously attempting to reduce her food intake, when faced with the very type of food she is trying to resist eating—pizza and cookies are both rated as forbidden foods by dieters (Knight & Boland, 1989)—develops an urge or an appetite to eat which undermines her diet and promotes even greater eating. A nondieter, on the other hand, when presented with an appetizing food, does not experience the urges or appetite to the same extent because she is not attempting to deny herself those foods. The nondieter’s automatic response would be to eat in accord with hunger and satiety signals and not to experience intense urges which could push eating beyond the normal limits of satiety. This theory offers an alternative framework to investigate cognitive processing and key characteristics of behaviors such as drug use and perhaps eating. The evidence from the present studies showing restrained eaters to have greater urges to eat particularly after pre-exposure to specific food cues suggests that this explanation may have some merit. However, it is necessary to design experiments to provide critical tests of the theory before any conclusions can be drawn about restrained eating. Understanding the influence of food cues on hunger, craving, and eating may have implications for eating disorders. There is abundant evidence that eating disordered patients are more sensitive to food-, weight-, and shaperelated cues in the environment (Ward, Bulik, & Johnston, 1996). Vitousek and Hollon (1990) hypothesized that individuals with eating disorders develop organized
cognitive structures around the issues of food, weight and shape. There is a growing literature demonstrating that anorexic and bulimic patients exhibit an attentional bias towards food-, weight-, and shape-related information (Carter, Bulik, McIntosh, & Joyce, 2000; Flynn, 1999; Green, Corr, & DeSilva, 1999), as well as a memory bias for such information (Hermans, Pieter, & Eelen, 1998; King, Polivy, & Herman, 1991). This heightened sensitivity toward food-related information may place eating disordered individuals in what amounts to a perpetual state of exposure to food cues, producing craving, which may contribute to triggering eating binges in bulimic patients. In conclusion, the present study demonstrates that exposure to food cues increases consumption by restrained eaters only when the food cues match the food available to be consumed. There was also evidence for specificity in appetite and sensory ratings of the food (particularly the cookies), suggesting that food-specific cravings can be triggered by food-specific cues.
Acknowledgements This research comprised part of the requirements for a PhD degree for the first author. Support was provided by a SSHRC grant to JP and CPH. Address correspondence to Ingrid C. Fedoroff, Department of Psychiatry, St Paul’s Hospital, 1081 Burrard St, Vancouver, BC, V6Z 1Y6.
References Carter, F. A., Bulik, C., McIntosh, V. V., & Joyce, P. R. (2000). Changes on the stroop test following treatment: Relation to word type, treatment condition, and treatment outcome among women with bulimia nervosa. International Journal of Eating Disorders, 28, 349– 355. Cornell, C. E., Rodin, J., & Weingarten, H. (1989). Stimulus-induced eating when satiated. Physiology and Behavior, 45, 695–704. Fedoroff, I., Polivy, J., & Herman, C. P. (1997). The effect of pre-exposure to food cues on the eating behavior of restrained and unrestrained eaters. Appetite, 28, 33–47. Flynn, S. V. (1999). Do disorder-relevant cognitive biases endure in recovered bulimics? Behavior Therapy, 30, 541–553. Grant, D. P., & Milgram, N. W. (1973). Plasticity of normal feeding: situational and individual factors. Canadian Journal of Psychology, 24, 305 –316. Green, M., Corr, P., & DeSilva, L. (1999). Impaired color naming of body shape-related words in anorexia nervosa: affective valence or associative priming? Cognitive Therapy and Research, 23, 413–422. Hermans, D., Pieters, G., & Eelen, P. (1998). Implicit and explicit memory for shape, body weight, and food-related words in patients with anorexia nervosa and nondieting controls. Journal of Abnormal Psychology, 107, 193 –202. Jansen, A., & Van den Hout, M. (1991). On being led into temptation: counterregulation of dieters after smelling a preload. Addictive Behaviors, 16, 247 –253. King, G., Polivy, J., & Herman, C. P. (1991). Cognitive aspects of dietary restraint: effects on person memory. International Journal of Eating Disorders, 10, 313 –322.
I. Fedoroff et al. / Appetite 41 (2003) 7–13 Klajner, F., Herman, C. P., Polivy, J., & Chhabra, R. (1981). Human obesity, dieting, and anticipatory salivation to food. Physiology and Behavior, 27, 195–198. Knight, L., & Boland, F. (1989). Restrained eating: An experimental disentanglement of disinhibiting variables of calories and food type. Journal of Abnormal Psychology, 98, 412 –420. Legoff, D. B., & Spigelman, M. N. (1987). Salivary response to olfactory food stimuli as a function of dietary restraint and body weight. Appetite, 8, 29– 36. Osborne, S. R. (1977). The free food (contrafreeloading) phenomenon: A review and analysis. Animal Learning and Behavior, 5, 221 –235. Polivy, J. (1996). Psychological consequences of food restriction. Journal of the American Dietetic Association, 96, 589–594. Polivy, J., Herman, C. P., & Howard, K. I. (1988). The restraint scale: assessment of dieting. In M. Hersen, & A. S. Bellack (Eds.), Dictionary of behavioral assessment techniques (pp. 377 –380). New York: Pergamon Press. Rogers, P. J., & Hill, A. (1989). Breakdown of dietary restraint following mere exposure to food stimuli: interrelationships between restraint,
13
hunger, salivation and food intake. Addictive Behaviors, 14, 387– 397. Tiffany, S. T. (1990). A cognitive model of drug urges and drug-use behavior: role of automatic and nonautomatic processes. Psychological Review, 97, 147– 168. Tiffany, S. T. (1992). A critique of contemporary urge and craving research: methodological, psychometric, and theoretical issues. Advanced Behavior Research Therapy, 14, 123–139. Vitousek, K., & Hollon, S. D. (1990). The investigation of schematic content and processing in eating disorders. Cognitive Therapy and Research, 14, 135–143. Ward, T., Bulik, C. M., & Johnston, L. (1996). Return of the suppressed: mental control and bulimia nervosa. Behaviour Change, 13, 79–90. Weingarten, H. P. (1985). Stimulus control of eating: implications for a two-factor theory of hunger. Appetite, 6, 387– 401. Weingarten, H., & Elston, D. (1990). The phenomenology of food cravings. Appetite, 15, 231–246. Woods, S. C. (1991). The eating paradox: how we tolerate food. Psychological Review, 98, 488–505.