- Email: [email protected]
Cognitive restraint and sensitivity to cues for hunger and satiety
Physiology& Behavior,Vol. 47, pp. 477--481. ©PergamonPress pie, 1990. Printedin the U.S.A.
0031-9384/90$3.00 + .00
Cognitive Restraint and Sensitivity to Cues for Hunger and Satiety JANE OGDEN AND JANE WARDLE x
Psychology Department, Institute o f Psychiatry, University o f London, UK Received 15 A u g u s t 1989
OGDEN, J. AND J. WARDLE. Cognitive restraint and sensitivity to cuesfor hunger and satiety. PHYSIOL BEHAV 47(3) 477--481, 1990.--The present study provides experimental evidence for the effects of cognitive restraint on sensitivity to internal and external cues of hunger. The design involved four experimental conditions. In each condition the subjects were given either a high calorie or a low calorie early morning drink, and either correctly or incorrectly told its calorie content. The subjects, 10 restrained and 10 nonrestrained normal weight women, then rated their subjective hunger and satiety responses and had a test meal. All subjects reported lower sensations of hunger and higher sensations of fullness after the high calorie drink than the low calorie drink, indicating sensitivity to internal cues. However, the restrained eaters' ratings of hunger were also influenced by the believed content of the drink, indicating sensitivity to external cues. The implications of the results are analysed in terms of current theories of restrained eating and their relevance to further research into pathological eating regulation is discussed. Restraint
Intemai cues
External cues
Hunger
Fullness
THE relative contribution of internal (physiological) cues and external (cognitive and emotional) cues in controlling eating behaviour was originally investigated specifically in connection with the problem of obesity. The work of Schachter and his colleagues (14,15) implicated both relative insensitivity to internal cues and relative over-responsiveness to external cues, in the psychopathology of obesity. This elegant and imaginative theory of obesity has stimulated a wealth of research into the mechanisms of food intake regulation, although there has not been consistent support for the original theory. Wooley and Wooley and their colleagues (23-25) reported a series of studies investigating the issue of responsiveness to internal cues. They were among the first to recognise the importance of disguising caloric loads in order to identify the independent contribution of purely internal factors. Their work suggested that few people of any weight level were able to discriminate disguised high and low calorie meals in the short term. In other words, relative insensitivity to internal signals of caloric loads (at least with novel stimuli) appeared to be the norm. However, in contrast with these results, Wardle (21) was able to demonstrate significant responsiveness to physiological (internal) cues in normal weight women. The difference in these results may be explained by the methods of analysis used. Wooley, Wooley and Dunham (23) based their conclusions on a series of univariate tests at different time points, and although the differences between the high and low calorie meals were in the expected direction, they failed to reach statistical significance. Wardle (21) used a repeated measures analysis of variance which was sensitive to the consistent differences across time in the two conditions. Responsiveness to external cues such as food palatability, food
variety, or food availability (the other facet of Schachter and Rodin's model) has increasingly been found to be a widespread characteristic, not restricted to obese people (13). Using a method of giving meals of varying caloric density, and asking subjects (obese and normal weight) to guess the caloric content, Wooley, Wooley and Dunham (23) reported a significant relationship between beliefs about food intake and subsequent hunger in both groups of subjects. Wardle (21), however, found that beliefs about caloric content had little influence on hunger. The major difference between these two studies was a methodological one; Wooley et al. allocated subjects to groups post hoc, with respect to their guesses about what they had consumed, i.e., at the end of the study period subjects were asked whether they "guessed" that they had had the low or the high calorie meal. As these "guesses" were made after the subjective ratings it was not surprising that "guesses" and subjective ratings were related. Wardle (21) used a design in which the subjective and physiological factors were manipulated independently and could find no evidence that telling subjects about the amount that they were going to eat influenced their subsequent feelings of hunger and fullness. Another possible explanation for the diverse results derived from studies of sensitivity to food cues might be that relative responsiveness to external and internal cues is related to cognitive restraint (22). This would parallel the work of Hibscher and Herman (7) who reevaluated Schachter's model of obesity and found that restraint rather than weight level could account for variations in the behavioural responses to preloads of different sizes. The role of restraint in the control of eating has been elaborated in Herman and Polivy's "boundary model" (6,12). Cognitive
1Requests for reprints should be addressed to Dr. Jane Wardle, Department of Psychology, Institute of Psychiatry, De Crespigny Park, London SE5 8AF.
477
478
OGDEN AND WARDLE
control of food intake has been specifically used by chronic dieters (restrained eaters). The model describes food intake as being regulated by physiologically determined hunger and satiety boundaries in " n o r m a l " eaters, but by cognitively determined " d i e t " boundaries in restrained eaters. It suggests that restrained eaters have a lowered hunger boundary and a raised satiety boundary as a consequence of decreased sensitivity to internai cues of hunger and satiety. Few investigations have compared responsiveness to internal and external cues directly. Lowered sensitivity to internal cues in dieters is mainly deduced from the observation that dieters do not show compensatory regulation in the laboratory setting (4, 5, 7). However, this observation does not unambiguously indicate sensitivity to internal cues, since a variety of internal and external cues could be salient in these situations. The methodology of the study used by Wardle (21) offers one way of investigating responsiveness to internal and external cues. It involved independently manipulating both the actual caloric content (internal cues) and the believed caloric content (external cues) of a drink and tracking hunger and satiety over time. Restraint theory predicts that restrained eaters would show lower sensitivity to the manipulation of the actual caloric content of the drinks and higher sensitivity to the "believed" caloric content than nonrestrained eaters. The present study was designed to compare sensitivity to internal and external cues in restrained and unrestrained eaters. METHOD
Subjects The subjects were 20 female nursing and administrative staff, aged between 20 and 58, who volunteered for a study of hunger and satiety. Restraint was assessed after all the experimental sessions were complete, using the Restrained Eating Scale of the Dutch Eating Behaviour Questionnaire (DEBQ) (19), the validity of which has been tested, and which appears to be easier to complete than other restraint scales (20).
Design The design involved four experimental conditions. In each condition the subjects were given either a high calorie drink or a low calorie drink, and either correctly or incorrectly told its calorie content. The subjects' belief was, therefore, either consistent or inconsistent with the true calorie content of the drink. This generated four conditions: actual high/apparent high calorie; actual high/apparent low calorie; actual low/apparent low calorie; actual low/apparent high calorie. Following the drink, hunger and satiety were recorded at regular intervals and after two hours food intake was assessed in a test meal (lunch). A within-subject design was used to reduce subject variation; each subject took part in the experiment four times. The order of the conditions was counterbalanced.
Procedure At the first session subjects were told that they were going to be given a midmorning orange drink on four different days which would either be high calorie or low calorie. They were told that the high calorie drink had a small amount of tasteless and colourless starch additive added to it to increase the caloric content to 300 kcai and that the low calorie drink was made to match it in texture with gelatine, but had a caloric content of only 50 kcal. The aim of the experiment was explained as being to determine the
effectiveness of the high calorie additive in the treatment of eating disorders. The subjects were asked to eat the same breakfast on each day of the experiment and not to eat between the drink and lunch. Each subject had her drink at the same time for each condition. The procedure was the same for all four conditions. First the subject completed a set of subjective ratings and was given the midmorning drink. Depending on the condition assigned, she was then either correctly or incorrectly informed of its caloric content. This was also written on the drink container. The high calorie drink consisted of 125 ml of orange juice and 62.5 g of a starch additive Maxijul (Vestric Foods, Ltd) giving a total caloric content of 300 kcal. The low calorie drink consisted of 125 ml of orange juice with half a teaspoon of gelatine added to make the two drinks of equal consistency. This had a total caloric content of 50 kcal. A pilot study indicated that the two drinks were closely comparable in taste and texture. The subject then completed a set of subjective ratings each half hour, the final rating being completed two hours after the drink, when she came for her lunch. This gave a total of five sampling points. Lunch was arranged to coincide with the subjects' usual lunch break, and the food provided had been previously selected by the subject from a limited menu. It consisted either of ham or cheese sandwiches made either with brown or white bread followed by biscuits. The four varieties of sandwiches were equicaloric (80 kcal per sandwich quarter) as were the biscuits (70 kcal per biscuit). Ten sandwich quarters and nine biscuits were provided, as this appeared to exceed any usual lunch intake, and was considered an ample amount of food which would allow the subjects to eat as much as they wished. The subjects were also provided with a drink of water or low calorie orange or lemon. Each subject had the same lunch on each day of the experiment. Magazines were provided with all the food pictures removed and subjects were left alone for 20 minutes to eat their lunch. After they had left, the remaining food was counted and weighed to calculate food intake. To check on the effectiveness of the belief manipulation the subjects were asked after the meal to remember whether they had been given a low or high calorie preload that morning. After all four conditions had been completed the subjects were given a further personal questionnaire to complete. This included questions about weight, height, age and dietary history. The restrained eating scale of the DEBQ was also given.
Measures 1. Subjective ratings. These were paper and pen tests, completed by the subject. a) Hunger rating. A 100 mm visual analogue scale was used. The subject rated herself on a scale from "not at all hungry" (0 mm) to "as hungry as I've ever been" (100 mm). b) Satiety rating. This was similarly measured on a 100 mm visual analogue scale with the extremes labelled " n o t at all full" (0 mm) and "as full as I've ever been" (100 mm). 2. Behavioural measures. Food intake was measured by weighing and counting the sandwiches and biscuits remaining after a test meal and calculating the amount that the subject had eaten. RESULTS
The subjects were divided into restrained and unrestrained eaters using a median split on the Restrained Eating scale of the DEBQ. Subject characteristics are shown in Table 1. All subjects accurately remembered the calorie content of the drink they had been given, indicating the effectiveness of the briefing.
RESTRAINT AND HUNGER
479
80
TABLE 1
_ _
low
....
high
preload
GROUP CHARACTERISTICS
Age (years) Weight (kg) Height (m) Restraint score
Restrainers
Nonrestralners
(n = 10)
(n = 10)
33 60 1.58 3.5
-+ ± ± ±
13.6 4.9 0.74 0.39
26 57.3 1.65 1.89
± ± _ ±
4.0 6.9 0.63 0.54
preload
70
P~
60
50
40
The data were analysed using the Multivariate Analysis of Variance (MANOVA) procedure (SPSS.pc) with restraint level (high restraint versus low restraint) as the between-subject factor, and externai cues (believed-high calorie versus believed-low calorie), internal cues (actual-high calorie versus actual-low calorie), and time, as the within-subject factors.
30
20
a
,
0 Time
q
30 since
i
i
60 90 120 preload (rain )
Subjective Measurements The subjective ratings showed a significant effect of time with a change in hunger, F(4,76)=28.17, p<0.001, and fullness, F(4,76) = 17.84, p<0.001, indicating that the subjects responded to an increased period of food deprivation from the morning drink to lunch, with an increase in hunger and a decrease in satiety. There was no time by restraint interaction, which suggested that the overall profile of change was similar in both the restrained and unrestrained eaters. There was a significant effect of the actual caloric content of the drink (internal cue manipulation) with a calorie level by time interaction, both in the hunger ratings, F(4,76) = 3.11, p < 0 . 0 5 , and the fullness ratings, F(4,76) = 2.62, p<0.05. These results are illustrated in Figs. 1 and 2. There was no three-way interaction with restraint. Examination of the means showed that all subjects were appropriately sensitive to the manipulation of actual caloric content, giving higher hunger ratings and lower fullness ratings after the low calorie drink than after the high calorie drink.
O0
FIG. 2. Fullness ratings following drinks of different caloric content. In the case of the believed calorie level (external cue manipulation) there was a significant time by believed calorie level by restraint level interaction for the hunger ratings, F(4,76)= 4.68, p<0.01. Comparisons of the means showed that the restrained eaters gave higher hunger ratings in the "believed-low" condition than in the "believed-high" condition, whereas the unrestrained eaters had similar profile of hunger ratings in the two conditions. Fullness ratings were not shown to be sensitive to the external cue manipulation although the means were in the comparable direction. The results of the hunger ratings for the two groups are illustrated in Figs. 3 and 4.
Behavioural Measurement Results from the test meal were analysed for food intake overall
80 - -
low
....
high
preload preload
70
_ _
Believed
lo~r
....
Believed
high
7O
oo
60
50
50
/
~s
/ / s
40
4o
30
~0
/
S_
'
30
i
0 Time
i
30
since
t
60
preload
i
90
i
120 (rain )
0 Time
30 since
60 preload
90
120 (rain
)
FIG. 3. H u n g e r ratings following drinks believed to be o f different caloric
FIG. 1. Hunger ratings following drinks of different caloric content.
content in restrained eaters.
480
OGDEN AND WARDLE
60 - -
Believed
low
....
Believed
high
70
60 / // e-,
/
~50 //
///
40
30
50
i
J
i
i
o
30
6o
9o
Time
since
preload
12o (rain )
FIG. 4. Hunger ratings following drinks believed to be of different caloric content in unrestrained eaters.
and for biscuit and sandwich intake separately. Neither showed any significant effect either of the external cue or internal cue manipulation. However, the nonrestrained eaters ate significantly more sandwiches than the restrained eaters (160.25-+52.61 g as opposed to only 120.75 -+ 33.2 g), resulting in a significant effect of restraint level, F(1,18) = 4.5, p <0.05. They ate slightly more biscuits (34.50 -+ 21.59 g as opposed to 28.88 -+ 29.03 g), but this failed to reach significance. DISCUSSION The results of this study indicated that all subjects, regardless of restraint level, were sensitive to variation in internal cues in their ratings of hunger and satiety, reporting a greater decrease in hunger and increase in fullness after the high calorie drink than after the low calorie drink. This coincides with the results of Wardle's (21) study, with good evidence for sensitivity in hunger and fullness ratings to the manipulation of internal cues. There were no group differences in sensitivity, and, therefore, no support for the view that restrained eaters are less sensitive to changes in this aspect of the "internal milieu." The results give little support to the predictions made in the boundary model (6,12), which
suggests that dieters may "lose the ability to detect the subtle precursors of satiety . . . " [(12), p. 639]. As predicted by restraint theory, the restrained eaters showed evidence of greater sensitivity to the manipulation of external cues, reporting less hunger if they were told that the drink was high in calories, and more if they were told it was low in calories. Cognitive restraint, therefore, appeared to increase subjects' susceptibility to external influences on their subjective experience of hunger. This is consistent with studies showing disinhibition of food intake after preloads which are believed to be high in calories (17). It also corresponds with the suggestion from the boundary model that restrained eaters utilise a cognitively defined "dietboundary" to determine food intake, in what was described as the "peculiar cognitive calculus of the dieter" [(6), p. 147]. The restrained eaters consumed less at the test meal than the unrestrained eaters irrespective of condition, suggesting that in the present study the restrained eaters were successfully restricting their food intake. However, sensitivity to internal and external cues identified in the subjective ratings did not extend to food intake in the test meal. The subjects ate much the same on each day of the experiment, irrespective of either the believed or the actual content of the preload. Short-term caloric regulation has been an area where research findings have been conflicting. Some studies have found a clear effect of preload size on food intake (2, 8, 10), others, have found no immediate intake compensation for foods of different caloric content (15, 18, 21). There are no obvious systematic differences between these studies, although variability of factors, such as the nutritional content of the preload and test meal or the duration of the interval between the two, could explain the discrepancy of the results [see Hill et al. (8) for discussion]. The individual consistency in food intake, demonstrated both in the present study and the previous one (21), suggests that habitual factors may be important determinants of eating, especially if the meal is given at a usual meal time. Restraint has been proposed as a precursor to pathological eating regulation, in particular in association with binge eating (22). Patients with bulimia nervosa and restrained eaters have both been hypothesised to show decreased sensitivity to internal cues. The results of the present study suggest that in terms of subjective experiences both restrained and unrestrained subjects are sensitive to variation in caloric load, with restrained eaters showing additional sensitivity to external cues. They also suggest that patients with disorders of eating control might show a readjustment of the balance between internal and external cue sensitivity. ACKNOWLEDGEMENT The assistance of a grant from the Medical Research Council is gratefully acknowledged.
REFERENCES
1. Blundell, J. E.; Rogers, P. J. Effects of anorexic drugs on food intake, food selection and preferences, and hunger motivation and subjective experiences. Appetite 1:151-165; 1980. 2. Geliebter, A. A. Effects of equicaloric loads of protein, fat and carbohydrate on food intake in the rat and man. Physiol. Behav. 22:267-273; 1979. 3. Herman, C. P.; Polivy, J.; Esses, V. M. The illusion of counterregulation. Appetite 9:161-169; 1987. 4. Herman, C. P.; Mack, D. Restrained and unrestrained eating. J. Pers. 43:647-660; 1975. 5. Herman, C. P.; Polivy, J.; Silver, R. The effects of an observer on eating behaviour: The induction of "sensible eating". J. Pers. 47:85-99; 1979. 6. Herman, C. P.; Polivy, J. A boundary model for the regulation of eating. In: Stunkard, A. J.; Stellar, E., eds. Eating and its disorders. New York: Raven Press; 1984.
7. Hibscher, J. A.; Herman, C. P. Obesity, dieting and the expression of "obese" characteristics. J. Comp. Physiol. Psychol. 91:374-380; 1977. 8. Hill, A. J.; Leathwood, P. D.; Blundell, J. E. Some evidence for short-term caloric compensation in normal weight human subjects. Hum. Nutr. App. Nutr. 41A:244-257; 1987. 9. Monello, L. F.; Mayer, J. Hunger and satiety sensations in men, women, boys and girls. Am. J. Clin. Nutr. 20:253-261; 1967. 10. Pliner, P. Effect of liquid and solid preloads on eating behaviour of obese and normal persons. Physiol. Behav. 11:285-290; 1973. 11. Polivy, J. Perception of calories and regulation of intake in restrained and unrestrained subjects. Addict. Behav. 1:237-244; 1976. 12. Polivy, J.; Herman, C. P. Diagnosis and treatment of normal eating. J. Consult. Clin. Psychol. 55:635-644; 1987. 13. Rodin, J. The externality theory today. In: Stunkard, A. J., ed. Obesity. Philadelphia: Saunders; 1980.
RESTRAINT AND HUNGER
14. Schachter, S.; Rodin, J. Obese humans and rats. Potomac, MD: Erlbaum; 1974. 15. Schachter, S. Obesity and eating. Science 161:751-756; 1968. 16. Speigel, T. A. Caloric regulation of food intake in man. J. Comp. Physiol. Psychol. 84:24--37; 1973. 17. Spencer, J. A.; Fremouw, M. J. Binge eating as a function of restraint and weight classification. J. Abnorm. Psychol. 88:262-267; 1979. 18. Sylvestre, J. L.; Tournier, A.; Verger, P.; Chabert, M.; Delarme, B.; Hossenlopp, J. Learned caloric adjustment of human intake. Appetite 12:95-104; 1989. 19. Van Strien, J.; Frijters, J. E.; Bergers, G. P.; Defares, P. B. Dutch eating behaviour. Questionnaire for assessment of restrained, emotional and external eating behaviour. Int. J. Eat. Dis. 5:295-315; 1985.
481
20. Wardle, J. The assessment of restrained eating. Behav. Res. Tber. 24:213-215; 1986. 21. Wardle, J. Hunger and satiety: A multidimensional assessment of response to caloric loads. Physiol. Behav. 40:577-582; 1987. 22. Wardle, J. Compulsive eating and dietary restraint. Br. J. Clin. Psychol. 26:47-55; 1987. 23. Wooley, S. C.; Wooley, O. W.; Dunham, R. B. Can calories be perceived and do they affect hunger in obese and non obese humans. J. Comp. Physiol. Psychol. 80:250-258; 1972. 24. Wooley, S. C. Physiologic versus cognitive factors in short-term food regulation in the obese and non obese. Psychosom. Med. 34:62-68; 1972. 25. Wooley, O. W.; Wooley, S. C.; Woods, W. A. Effect of calories on appetite for palatable food in obese and non-obese humans. J. Comp. Physiol. Psychol. 89:619-625; 1975.
0031-9384/90$3.00 + .00
Cognitive Restraint and Sensitivity to Cues for Hunger and Satiety JANE OGDEN AND JANE WARDLE x
Psychology Department, Institute o f Psychiatry, University o f London, UK Received 15 A u g u s t 1989
OGDEN, J. AND J. WARDLE. Cognitive restraint and sensitivity to cuesfor hunger and satiety. PHYSIOL BEHAV 47(3) 477--481, 1990.--The present study provides experimental evidence for the effects of cognitive restraint on sensitivity to internal and external cues of hunger. The design involved four experimental conditions. In each condition the subjects were given either a high calorie or a low calorie early morning drink, and either correctly or incorrectly told its calorie content. The subjects, 10 restrained and 10 nonrestrained normal weight women, then rated their subjective hunger and satiety responses and had a test meal. All subjects reported lower sensations of hunger and higher sensations of fullness after the high calorie drink than the low calorie drink, indicating sensitivity to internal cues. However, the restrained eaters' ratings of hunger were also influenced by the believed content of the drink, indicating sensitivity to external cues. The implications of the results are analysed in terms of current theories of restrained eating and their relevance to further research into pathological eating regulation is discussed. Restraint
Intemai cues
External cues
Hunger
Fullness
THE relative contribution of internal (physiological) cues and external (cognitive and emotional) cues in controlling eating behaviour was originally investigated specifically in connection with the problem of obesity. The work of Schachter and his colleagues (14,15) implicated both relative insensitivity to internal cues and relative over-responsiveness to external cues, in the psychopathology of obesity. This elegant and imaginative theory of obesity has stimulated a wealth of research into the mechanisms of food intake regulation, although there has not been consistent support for the original theory. Wooley and Wooley and their colleagues (23-25) reported a series of studies investigating the issue of responsiveness to internal cues. They were among the first to recognise the importance of disguising caloric loads in order to identify the independent contribution of purely internal factors. Their work suggested that few people of any weight level were able to discriminate disguised high and low calorie meals in the short term. In other words, relative insensitivity to internal signals of caloric loads (at least with novel stimuli) appeared to be the norm. However, in contrast with these results, Wardle (21) was able to demonstrate significant responsiveness to physiological (internal) cues in normal weight women. The difference in these results may be explained by the methods of analysis used. Wooley, Wooley and Dunham (23) based their conclusions on a series of univariate tests at different time points, and although the differences between the high and low calorie meals were in the expected direction, they failed to reach statistical significance. Wardle (21) used a repeated measures analysis of variance which was sensitive to the consistent differences across time in the two conditions. Responsiveness to external cues such as food palatability, food
variety, or food availability (the other facet of Schachter and Rodin's model) has increasingly been found to be a widespread characteristic, not restricted to obese people (13). Using a method of giving meals of varying caloric density, and asking subjects (obese and normal weight) to guess the caloric content, Wooley, Wooley and Dunham (23) reported a significant relationship between beliefs about food intake and subsequent hunger in both groups of subjects. Wardle (21), however, found that beliefs about caloric content had little influence on hunger. The major difference between these two studies was a methodological one; Wooley et al. allocated subjects to groups post hoc, with respect to their guesses about what they had consumed, i.e., at the end of the study period subjects were asked whether they "guessed" that they had had the low or the high calorie meal. As these "guesses" were made after the subjective ratings it was not surprising that "guesses" and subjective ratings were related. Wardle (21) used a design in which the subjective and physiological factors were manipulated independently and could find no evidence that telling subjects about the amount that they were going to eat influenced their subsequent feelings of hunger and fullness. Another possible explanation for the diverse results derived from studies of sensitivity to food cues might be that relative responsiveness to external and internal cues is related to cognitive restraint (22). This would parallel the work of Hibscher and Herman (7) who reevaluated Schachter's model of obesity and found that restraint rather than weight level could account for variations in the behavioural responses to preloads of different sizes. The role of restraint in the control of eating has been elaborated in Herman and Polivy's "boundary model" (6,12). Cognitive
1Requests for reprints should be addressed to Dr. Jane Wardle, Department of Psychology, Institute of Psychiatry, De Crespigny Park, London SE5 8AF.
477
478
OGDEN AND WARDLE
control of food intake has been specifically used by chronic dieters (restrained eaters). The model describes food intake as being regulated by physiologically determined hunger and satiety boundaries in " n o r m a l " eaters, but by cognitively determined " d i e t " boundaries in restrained eaters. It suggests that restrained eaters have a lowered hunger boundary and a raised satiety boundary as a consequence of decreased sensitivity to internai cues of hunger and satiety. Few investigations have compared responsiveness to internal and external cues directly. Lowered sensitivity to internal cues in dieters is mainly deduced from the observation that dieters do not show compensatory regulation in the laboratory setting (4, 5, 7). However, this observation does not unambiguously indicate sensitivity to internal cues, since a variety of internal and external cues could be salient in these situations. The methodology of the study used by Wardle (21) offers one way of investigating responsiveness to internal and external cues. It involved independently manipulating both the actual caloric content (internal cues) and the believed caloric content (external cues) of a drink and tracking hunger and satiety over time. Restraint theory predicts that restrained eaters would show lower sensitivity to the manipulation of the actual caloric content of the drinks and higher sensitivity to the "believed" caloric content than nonrestrained eaters. The present study was designed to compare sensitivity to internal and external cues in restrained and unrestrained eaters. METHOD
Subjects The subjects were 20 female nursing and administrative staff, aged between 20 and 58, who volunteered for a study of hunger and satiety. Restraint was assessed after all the experimental sessions were complete, using the Restrained Eating Scale of the Dutch Eating Behaviour Questionnaire (DEBQ) (19), the validity of which has been tested, and which appears to be easier to complete than other restraint scales (20).
Design The design involved four experimental conditions. In each condition the subjects were given either a high calorie drink or a low calorie drink, and either correctly or incorrectly told its calorie content. The subjects' belief was, therefore, either consistent or inconsistent with the true calorie content of the drink. This generated four conditions: actual high/apparent high calorie; actual high/apparent low calorie; actual low/apparent low calorie; actual low/apparent high calorie. Following the drink, hunger and satiety were recorded at regular intervals and after two hours food intake was assessed in a test meal (lunch). A within-subject design was used to reduce subject variation; each subject took part in the experiment four times. The order of the conditions was counterbalanced.
Procedure At the first session subjects were told that they were going to be given a midmorning orange drink on four different days which would either be high calorie or low calorie. They were told that the high calorie drink had a small amount of tasteless and colourless starch additive added to it to increase the caloric content to 300 kcai and that the low calorie drink was made to match it in texture with gelatine, but had a caloric content of only 50 kcal. The aim of the experiment was explained as being to determine the
effectiveness of the high calorie additive in the treatment of eating disorders. The subjects were asked to eat the same breakfast on each day of the experiment and not to eat between the drink and lunch. Each subject had her drink at the same time for each condition. The procedure was the same for all four conditions. First the subject completed a set of subjective ratings and was given the midmorning drink. Depending on the condition assigned, she was then either correctly or incorrectly informed of its caloric content. This was also written on the drink container. The high calorie drink consisted of 125 ml of orange juice and 62.5 g of a starch additive Maxijul (Vestric Foods, Ltd) giving a total caloric content of 300 kcal. The low calorie drink consisted of 125 ml of orange juice with half a teaspoon of gelatine added to make the two drinks of equal consistency. This had a total caloric content of 50 kcal. A pilot study indicated that the two drinks were closely comparable in taste and texture. The subject then completed a set of subjective ratings each half hour, the final rating being completed two hours after the drink, when she came for her lunch. This gave a total of five sampling points. Lunch was arranged to coincide with the subjects' usual lunch break, and the food provided had been previously selected by the subject from a limited menu. It consisted either of ham or cheese sandwiches made either with brown or white bread followed by biscuits. The four varieties of sandwiches were equicaloric (80 kcal per sandwich quarter) as were the biscuits (70 kcal per biscuit). Ten sandwich quarters and nine biscuits were provided, as this appeared to exceed any usual lunch intake, and was considered an ample amount of food which would allow the subjects to eat as much as they wished. The subjects were also provided with a drink of water or low calorie orange or lemon. Each subject had the same lunch on each day of the experiment. Magazines were provided with all the food pictures removed and subjects were left alone for 20 minutes to eat their lunch. After they had left, the remaining food was counted and weighed to calculate food intake. To check on the effectiveness of the belief manipulation the subjects were asked after the meal to remember whether they had been given a low or high calorie preload that morning. After all four conditions had been completed the subjects were given a further personal questionnaire to complete. This included questions about weight, height, age and dietary history. The restrained eating scale of the DEBQ was also given.
Measures 1. Subjective ratings. These were paper and pen tests, completed by the subject. a) Hunger rating. A 100 mm visual analogue scale was used. The subject rated herself on a scale from "not at all hungry" (0 mm) to "as hungry as I've ever been" (100 mm). b) Satiety rating. This was similarly measured on a 100 mm visual analogue scale with the extremes labelled " n o t at all full" (0 mm) and "as full as I've ever been" (100 mm). 2. Behavioural measures. Food intake was measured by weighing and counting the sandwiches and biscuits remaining after a test meal and calculating the amount that the subject had eaten. RESULTS
The subjects were divided into restrained and unrestrained eaters using a median split on the Restrained Eating scale of the DEBQ. Subject characteristics are shown in Table 1. All subjects accurately remembered the calorie content of the drink they had been given, indicating the effectiveness of the briefing.
RESTRAINT AND HUNGER
479
80
TABLE 1
_ _
low
....
high
preload
GROUP CHARACTERISTICS
Age (years) Weight (kg) Height (m) Restraint score
Restrainers
Nonrestralners
(n = 10)
(n = 10)
33 60 1.58 3.5
-+ ± ± ±
13.6 4.9 0.74 0.39
26 57.3 1.65 1.89
± ± _ ±
4.0 6.9 0.63 0.54
preload
70
P~
60
50
40
The data were analysed using the Multivariate Analysis of Variance (MANOVA) procedure (SPSS.pc) with restraint level (high restraint versus low restraint) as the between-subject factor, and externai cues (believed-high calorie versus believed-low calorie), internal cues (actual-high calorie versus actual-low calorie), and time, as the within-subject factors.
30
20
a
,
0 Time
q
30 since
i
i
60 90 120 preload (rain )
Subjective Measurements The subjective ratings showed a significant effect of time with a change in hunger, F(4,76)=28.17, p<0.001, and fullness, F(4,76) = 17.84, p<0.001, indicating that the subjects responded to an increased period of food deprivation from the morning drink to lunch, with an increase in hunger and a decrease in satiety. There was no time by restraint interaction, which suggested that the overall profile of change was similar in both the restrained and unrestrained eaters. There was a significant effect of the actual caloric content of the drink (internal cue manipulation) with a calorie level by time interaction, both in the hunger ratings, F(4,76) = 3.11, p < 0 . 0 5 , and the fullness ratings, F(4,76) = 2.62, p<0.05. These results are illustrated in Figs. 1 and 2. There was no three-way interaction with restraint. Examination of the means showed that all subjects were appropriately sensitive to the manipulation of actual caloric content, giving higher hunger ratings and lower fullness ratings after the low calorie drink than after the high calorie drink.
O0
FIG. 2. Fullness ratings following drinks of different caloric content. In the case of the believed calorie level (external cue manipulation) there was a significant time by believed calorie level by restraint level interaction for the hunger ratings, F(4,76)= 4.68, p<0.01. Comparisons of the means showed that the restrained eaters gave higher hunger ratings in the "believed-low" condition than in the "believed-high" condition, whereas the unrestrained eaters had similar profile of hunger ratings in the two conditions. Fullness ratings were not shown to be sensitive to the external cue manipulation although the means were in the comparable direction. The results of the hunger ratings for the two groups are illustrated in Figs. 3 and 4.
Behavioural Measurement Results from the test meal were analysed for food intake overall
80 - -
low
....
high
preload preload
70
_ _
Believed
lo~r
....
Believed
high
7O
oo
60
50
50
/
~s
/ / s
40
4o
30
~0
/
S_
'
30
i
0 Time
i
30
since
t
60
preload
i
90
i
120 (rain )
0 Time
30 since
60 preload
90
120 (rain
)
FIG. 3. H u n g e r ratings following drinks believed to be o f different caloric
FIG. 1. Hunger ratings following drinks of different caloric content.
content in restrained eaters.
480
OGDEN AND WARDLE
60 - -
Believed
low
....
Believed
high
70
60 / // e-,
/
~50 //
///
40
30
50
i
J
i
i
o
30
6o
9o
Time
since
preload
12o (rain )
FIG. 4. Hunger ratings following drinks believed to be of different caloric content in unrestrained eaters.
and for biscuit and sandwich intake separately. Neither showed any significant effect either of the external cue or internal cue manipulation. However, the nonrestrained eaters ate significantly more sandwiches than the restrained eaters (160.25-+52.61 g as opposed to only 120.75 -+ 33.2 g), resulting in a significant effect of restraint level, F(1,18) = 4.5, p <0.05. They ate slightly more biscuits (34.50 -+ 21.59 g as opposed to 28.88 -+ 29.03 g), but this failed to reach significance. DISCUSSION The results of this study indicated that all subjects, regardless of restraint level, were sensitive to variation in internal cues in their ratings of hunger and satiety, reporting a greater decrease in hunger and increase in fullness after the high calorie drink than after the low calorie drink. This coincides with the results of Wardle's (21) study, with good evidence for sensitivity in hunger and fullness ratings to the manipulation of internal cues. There were no group differences in sensitivity, and, therefore, no support for the view that restrained eaters are less sensitive to changes in this aspect of the "internal milieu." The results give little support to the predictions made in the boundary model (6,12), which
suggests that dieters may "lose the ability to detect the subtle precursors of satiety . . . " [(12), p. 639]. As predicted by restraint theory, the restrained eaters showed evidence of greater sensitivity to the manipulation of external cues, reporting less hunger if they were told that the drink was high in calories, and more if they were told it was low in calories. Cognitive restraint, therefore, appeared to increase subjects' susceptibility to external influences on their subjective experience of hunger. This is consistent with studies showing disinhibition of food intake after preloads which are believed to be high in calories (17). It also corresponds with the suggestion from the boundary model that restrained eaters utilise a cognitively defined "dietboundary" to determine food intake, in what was described as the "peculiar cognitive calculus of the dieter" [(6), p. 147]. The restrained eaters consumed less at the test meal than the unrestrained eaters irrespective of condition, suggesting that in the present study the restrained eaters were successfully restricting their food intake. However, sensitivity to internal and external cues identified in the subjective ratings did not extend to food intake in the test meal. The subjects ate much the same on each day of the experiment, irrespective of either the believed or the actual content of the preload. Short-term caloric regulation has been an area where research findings have been conflicting. Some studies have found a clear effect of preload size on food intake (2, 8, 10), others, have found no immediate intake compensation for foods of different caloric content (15, 18, 21). There are no obvious systematic differences between these studies, although variability of factors, such as the nutritional content of the preload and test meal or the duration of the interval between the two, could explain the discrepancy of the results [see Hill et al. (8) for discussion]. The individual consistency in food intake, demonstrated both in the present study and the previous one (21), suggests that habitual factors may be important determinants of eating, especially if the meal is given at a usual meal time. Restraint has been proposed as a precursor to pathological eating regulation, in particular in association with binge eating (22). Patients with bulimia nervosa and restrained eaters have both been hypothesised to show decreased sensitivity to internal cues. The results of the present study suggest that in terms of subjective experiences both restrained and unrestrained subjects are sensitive to variation in caloric load, with restrained eaters showing additional sensitivity to external cues. They also suggest that patients with disorders of eating control might show a readjustment of the balance between internal and external cue sensitivity. ACKNOWLEDGEMENT The assistance of a grant from the Medical Research Council is gratefully acknowledged.
REFERENCES
1. Blundell, J. E.; Rogers, P. J. Effects of anorexic drugs on food intake, food selection and preferences, and hunger motivation and subjective experiences. Appetite 1:151-165; 1980. 2. Geliebter, A. A. Effects of equicaloric loads of protein, fat and carbohydrate on food intake in the rat and man. Physiol. Behav. 22:267-273; 1979. 3. Herman, C. P.; Polivy, J.; Esses, V. M. The illusion of counterregulation. Appetite 9:161-169; 1987. 4. Herman, C. P.; Mack, D. Restrained and unrestrained eating. J. Pers. 43:647-660; 1975. 5. Herman, C. P.; Polivy, J.; Silver, R. The effects of an observer on eating behaviour: The induction of "sensible eating". J. Pers. 47:85-99; 1979. 6. Herman, C. P.; Polivy, J. A boundary model for the regulation of eating. In: Stunkard, A. J.; Stellar, E., eds. Eating and its disorders. New York: Raven Press; 1984.
7. Hibscher, J. A.; Herman, C. P. Obesity, dieting and the expression of "obese" characteristics. J. Comp. Physiol. Psychol. 91:374-380; 1977. 8. Hill, A. J.; Leathwood, P. D.; Blundell, J. E. Some evidence for short-term caloric compensation in normal weight human subjects. Hum. Nutr. App. Nutr. 41A:244-257; 1987. 9. Monello, L. F.; Mayer, J. Hunger and satiety sensations in men, women, boys and girls. Am. J. Clin. Nutr. 20:253-261; 1967. 10. Pliner, P. Effect of liquid and solid preloads on eating behaviour of obese and normal persons. Physiol. Behav. 11:285-290; 1973. 11. Polivy, J. Perception of calories and regulation of intake in restrained and unrestrained subjects. Addict. Behav. 1:237-244; 1976. 12. Polivy, J.; Herman, C. P. Diagnosis and treatment of normal eating. J. Consult. Clin. Psychol. 55:635-644; 1987. 13. Rodin, J. The externality theory today. In: Stunkard, A. J., ed. Obesity. Philadelphia: Saunders; 1980.
RESTRAINT AND HUNGER
14. Schachter, S.; Rodin, J. Obese humans and rats. Potomac, MD: Erlbaum; 1974. 15. Schachter, S. Obesity and eating. Science 161:751-756; 1968. 16. Speigel, T. A. Caloric regulation of food intake in man. J. Comp. Physiol. Psychol. 84:24--37; 1973. 17. Spencer, J. A.; Fremouw, M. J. Binge eating as a function of restraint and weight classification. J. Abnorm. Psychol. 88:262-267; 1979. 18. Sylvestre, J. L.; Tournier, A.; Verger, P.; Chabert, M.; Delarme, B.; Hossenlopp, J. Learned caloric adjustment of human intake. Appetite 12:95-104; 1989. 19. Van Strien, J.; Frijters, J. E.; Bergers, G. P.; Defares, P. B. Dutch eating behaviour. Questionnaire for assessment of restrained, emotional and external eating behaviour. Int. J. Eat. Dis. 5:295-315; 1985.
481
20. Wardle, J. The assessment of restrained eating. Behav. Res. Tber. 24:213-215; 1986. 21. Wardle, J. Hunger and satiety: A multidimensional assessment of response to caloric loads. Physiol. Behav. 40:577-582; 1987. 22. Wardle, J. Compulsive eating and dietary restraint. Br. J. Clin. Psychol. 26:47-55; 1987. 23. Wooley, S. C.; Wooley, O. W.; Dunham, R. B. Can calories be perceived and do they affect hunger in obese and non obese humans. J. Comp. Physiol. Psychol. 80:250-258; 1972. 24. Wooley, S. C. Physiologic versus cognitive factors in short-term food regulation in the obese and non obese. Psychosom. Med. 34:62-68; 1972. 25. Wooley, O. W.; Wooley, S. C.; Woods, W. A. Effect of calories on appetite for palatable food in obese and non-obese humans. J. Comp. Physiol. Psychol. 89:619-625; 1975.