Binge Size Increases with Body Mass Index in Women with Binge-Eating Disorder


Obesity Research Center, St. Luke's/Roosevelt Hospital Center, Babcock Building, Room 1018, 1111 Amsterdam Avenue, New York, NY 10025. E-mail:


Objective: To determine whether meal size is related to body mass index (BMI) in obese subjects with binge-eating disorder (BED).

Research Methods and Procedures: Five groups of subjects each consumed two laboratory-test meals on nonconsecutive days. Forty-two women, categorized by BMI and BED diagnosis, were instructed to “binge” during one meal and to eat “normally” during another. Eighteen women had BMI values >38 kg/m2 (more-obese) and 17 had BMI values between 28 to 32 kg/m2 (less-obese). Twelve of the more-obese and nine of the less-obese individuals met Diagnostic and Statistical Manual (DSM)-IV criteria for BED. Seven normal-weight women also participated as controls.

Results: Subjects with BED ate significantly more in both meals than subjects without BED. Binge meals were significantly larger than normal meals only among subjects with BED. The more-obese subjects with BED ate significantly more than the less-obese subjects with BED, but only when they were asked to binge. Intake of the binge meal was significantly, positively correlated with BMI among subjects with BED. Subjects with BED reported significantly higher satiety ratings after the binge than after the normal meal, but subjects without BED reported similar ratings after both meals. Regardless of instructions and diagnosis, obese subjects consumed a significantly higher percentage of energy from fat (38.5%) than did normal-weight subjects (30.8%).

Discussion: During binge meals, the energy intake of subjects with BED is greater than that of individuals of similar body weight without BED and is positively correlated with BMI.


Binge-eating disorder (BED) is a provisional eating disorder included in the fourth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) (1). Like individuals with bulimia nervosa (BN), individuals with BED engage in recurrent binge eating. The definition of a binge episode in DSM-IV is identical for BN and for BED, and includes two components: 1) “eating, in a discrete period of time (e.g., within any 2-hour period), an amount of food that is definitely larger than most people would eat in a similar period of time under similar circumstances”; and 2) “a sense of lack of control over eating during the episode” (1). The key distinction between BN and BED is that the binges of subjects with BED are not followed by inappropriate compensatory behaviors, such as self-induced vomiting, which are required for the diagnosis of BN.

Previous work from our group (2), and from investigators at NIH (3), has attempted to obtain objective information on the eating behavior of subjects with BED by asking subjects to consume meals in a structured laboratory setting. Yanovski et al. (3) reported that the average binge meal of subjects with BED (2964 kcal) was significantly greater than that of weight-matched obese subjects without BED and was similar to the size of binge meals observed among subjects with BN (2680 kcal) (4). In contrast, Goldfein et al. (2) reported that the average binge meal of subjects with BED, although greater than that of weight-matched obese subjects without BED, was only approximately half that observed in subjects with BN (1515 kcal vs. 2964 kcal, respectively). Thus, the binge meals of subjects with BED reported by Goldfein et al. (2) differed dramatically from those reported by Yanovski et al. (3). Both laboratories used an identical paradigm (including the test meal) that was developed to characterize the eating behavior of subjects with BN (5).

Yanovski et al. (3) also reported that the normal meal of subjects with BED was significantly greater, on average, than that of subjects of similar body weight without BED. This result is strikingly different from studies of BN, which found, using identical procedures, that the size of a non-binge (i.e., normal) meal among individuals with BN is substantially smaller than that of subjects without BN (6). These unexpected differences in energy intake of both binge and normal meals, across populations who were both classified with the same eating behavioral abnormality, led us to consider potential differences between the subject populations that could have led to such results.

One notable difference between the subjects with BED in the studies by Yanovski et al. (3) and Goldfein et al. (2) was their average body mass index (BMI). The mean BMI of BED subjects in the study by Yanovski et al. was 41.2 kg/m2, whereas the mean BMI of BED subjects in the study by Goldfein et al. was only 33.4 kg/m2. This difference raised the possibility that subjects with BED were eating in proportion to their body size, particularly during binge episodes. A relationship between energy intake and body size would be predicted thermodynamically, but is seldom observed in humans within an individual meal. If such a relationship is found among subjects with BED, it could impact both our understanding of the relationship between binge eating and obesity and of the mechanisms underlying this disorder. This study was undertaken to compare, in a single protocol, subjects with and without BED whose body mass indices were similar to those of the populations studied by Yanovski and Goldfein. The objectives of this study were as follows: 1) to determine whether the differences in BMI between the two previous studies could account for differences in amounts eaten among subjects with BED, and 2) to examine the relationship between BMI and meal size among obese subjects with and without BED.

Research Methods and Procedures


Forty-two women, between the ages of 18 and 45, participated in three multiple-item test meals. To ensure that subjects’ body mass indices could be compared with participants from the studies of Goldfein et al. (2) and Yanovski et al. (3), overweight subjects were recruited in two weight ranges. One overweight group had a BMI >38 kg/m2 (referred to as the more-obese group), and the other group had BMI values between 28 and 32 kg/m2 (the less-obese group). The participants included 21 overweight women who met DSM-IV criteria for BED (9 less-obese and 12 more-obese) and 14 overweight women (8 less-obese, 6 more-obese) without any history of eating disorders. Seven control subjects, within normal weight for height (BMI, 19 kg/m2 to 23 kg/m2) (7) for all of their adult life, excluding pregnancy, also completed the study. The original intent was to have half the subjects within each BMI category be individuals meeting criteria for BED. However, it was not possible to find such a diverse group of obese subjects during the recruitment period, and the numbers of obese subjects with and without BED for each BMI group were therefore unequal.

Subject Recruitment.

All subjects were recruited from advertisements in the local media and from posters on the medical center campus. Obese subjects with BED were offered short-term psychotherapy for BED, and obese subjects without BED were offered short-term behavior modification treatment for weight loss. Normal-weight controls were offered financial compensation. This study was approved by the Institutional Review Boards of the St. Luke's–Roosevelt Hospital Center and the Columbia University/New York State Psychiatric Institute. All subjects provided written, signed consent and were informed during this consent process that their eating behavior would be monitored at all times with a closed-circuit television monitor.

Clinical Evaluation.

The Structured Clinical Interview for DSM-IV Axis I Disorders (SCID) (7), including questions derived from the Questionnaire on Eating and Weight Patterns (9), was used to determine lifetime eating-disorder history and psychiatric status in all participants. Participants who met SCID criteria for an eating disorder were further interviewed to evaluate whether they met standard criteria for BED. The diagnosis of BED was determined with the Eating Disorder Examination (10), which is a semi-structured, investigator-based interview that generates operational eating disorder diagnoses based on DSM-IV criteria (11). The format allowed assessors to rigorously evaluate whether potential subjects met DSM-IV (1) standards for the definition of BED (12). All BED subjects met DSM-IV criteria for the disorder (1) and had no history of BN.

Obese subjects without BED never met DSM-IV criteria for any eating disorder and were only included if they had not engaged in any form of binge eating for at least the past 6 months. None of the obese subjects was currently enrolled in a weight-loss program. Control subjects had no history of psychiatric illness, binge eating, or self-induced vomiting. All groups of subjects were physically healthy, not taking medication, and had no recent history of alcohol or drug abuse. All subjects completed the Three-Factor Eating Questionnaire (TFEQ) (13), which measures dietary restraint, disinhibition, and perceived hunger.

Study Design

Three test sessions took place on nonconsecutive days. The first test day served as an adaptation session (see below), and the second and third sessions served as the experimental test-meal days. The two experimental test-meal conditions were instructions to “binge eat,” and instructions to “eat normally” (see “Meal Instructions” below). These instructions were given just before subjects consumed each of the test meals. Subjects were not told which instruction they would receive until immediately before the start of their meal. The instruction to binge or to eat normally was assigned to each subject in random order.

Adaptation Session.

On the first day of participation, subjects arrived at the laboratory between 8:00 am and 10:00 am after an overnight fast. They then participated in a brief taste test (14) to ensure that none of the participants disliked the foods that would be offered during the multiple-item meals. Subjects were required to rate the foods that were to be offered in the study as at least a six (“like moderately”) on a nine-point category scale of liking (15). After the taste test, subjects consumed a 300 kcal, standardized breakfast consisting of a buttered (6 g) English muffin and apple juice (249 g). They were instructed to abstain from eating or drinking anything except for water until they returned to the laboratory 6 hours later. On their return, subjects were asked to consume “as much as you would like” of the multiple-item test meal (4). This test meal served to adapt all subjects to eating a meal in the laboratory setting without biasing their future responses to specific instructions regarding eating.

Experimental Day Procedures.

Subjects were given preportioned servings of the standardized breakfast (see “Adaptation Session” below) to prepare at home for each experimental day. Subjects consumed the breakfast 6 hours before their test-meal session at the laboratory, which took place between 4:00 and 6:00 pm. On arrival, they were seated in a private room in front of a table that contained an array of foods arranged as described previously (16). A laboratory assistant escorted them to the room, but left the room immediately after the instructions were given for each test meal. The assistant observed the subject from an adjacent room that contained a closed-circuit television monitor that was connected to a concealed video camera within the eating room.

Meal Instructions.

Immediately before the start of each meal, subjects listened to tape-recorded instructions that on one test day stated the following: “Let yourself go and binge eat. If you are someone who binge eats regularly, we would like your behavior here to resemble your behavior at home as much as possible. If you are someone who does not binge eat, we would like you to let yourself go and overeat.” On the other test day, each subject was instructed to “eat as much as you would eat in a normal meal in a relaxed atmosphere.”

Meal-Related Ratings.

Before and after each meal, subjects were asked to indicate responses to the following questions: “how hungry are you?” and “how satiated are you?” by placing marks on 150-mm lines (Visual Analog Scales, or VAS). Each line was anchored at the ends by the words “not at all” (corresponding to 0 mm) and “extremely” (corresponding to 150 mm) (17). Responses were scored by measuring the subject's mark placement along the 150-mm line. After completing each meal, subjects also rated its palatability on a nine-point category scale of liking (15).

Meal Composition.

The multiple-item meal was identical to that served in previous studies (2, 16). The foods served were typical of those consumed in normal meals (e.g. chicken or fish, lettuce, rice, apples, etc.), as well as foods that had been reported by patients with BN to be typical of binge episodes (e.g. ice cream, cookies, potato chips). Total energy intake was calculated by taking the sum of the differences, in weight, for every food item offered in the test meal, before and after the meal was served. Energy and macronutrient values were obtained from either the manufacturer's information or Nutritionist IV software (18).


A one-way ANOVA was used to assess differences among the five groups on the demographic data and on subscores of the TFEQ. Post hoc Duncan tests were then used to evaluate differences among groups. The remaining dependent variables were analyzed using a two-factor, mixed-measures ANOVA, with group (i.e., more-obese BED, more-obese non-BED, less-obese BED, less-obese non-BED, and normal-weight controls) serving as the between-subject factor and instruction (binge eat vs. eat normally) as the within-subjects factor.

Because there was no group of non-obese subjects with BED, a complete (i.e., 3 × 2 × 2) factorial design could not be employed for the data analysis. Therefore, the two-way (group x instruction) ANOVA was followed by post hoc contrast tests using appropriate error terms for comparisons (i.e., contrasts) of interest. This analysis allowed us to compare the following: (1) effects of weight group and of diagnosis; (2) effects of instruction; (3) effects of instruction within weight and diagnostic groups; (4) interactions between the aforementioned comparisons; and (5) simple effects between groups (e.g. more-obese BED vs. less-obese BED) for a given instruction.

The “subjects within-groups” error term from the ANOVA was used for planned comparisons between groups (e.g. more-obese BED vs. less-obese BED) collapsed over both instructions (e.g. binge and normal). The “subjects within-groups x instruction” error term was used for comparisons between groups (e.g. BED groups vs. non-BED groups) between instructions (e.g. binge vs. normal). The pooled error term (i.e., “subjects within groups + subjects within-groups x instruction”) was used for comparison between any one level of instruction across more than one group (19). Degrees of freedom were adjusted using the Satterthwaite approximation (19).

Four separate regression analyses of intake on BMI were performed: one for each combination of instruction and diagnostic category (i.e., BED/binge, BED/normal, non-BED/binge, non-BED/normal). Statistical values reported in this study were generated with the SAS/STAT software package for Windows, using Procedures General Linear Model (GLM) and Mixed, from Versions 6.03 (20) and 8.01 (21), respectively.



There were no significant differences in BMI between subjects with and without BED within each of the obese groups (see Table 1 for demographic characteristics of each group).

Table 1.  Means and SEs of dependent variables
 More obeseLess obese 
VariableBED (n = 12)Non-BED (n = 6)BED (n = 9)Non-BED (n = 8)Controls (n = 7)
  1. BED, binge-eating disorder; BMI, body mass index; TFEQ, Three-Factor Eating Questionnaire.

BMI (kg/m2)41.53 ± 0.9340.39 ± 0.4631.05 ± 0.4930.40 ± 0.4820.99 ± 0.70
Age (years)31.67 ± 1.3235.00 ± 2.8333.67 ± 2.2432.50 ± 1.8331.14 ± 3.36
TFEQ: restraint7.9 ± 1.056.7 ± 0.997.3 ± 1.089.7 ± 1.584.6 ± 0.57
TFEQ: disinhibition13.4 ± 0.576.3 ± 0.7613.6 ± 0.504.4 ± 0.903.6 ± 0.81
TFEQ: hunger9.6 ± 0.925.7 ± 1.4510.6 ± 0.603.4 ± 0.733.1 ± 0.63
Binge intake (kcal)2388 ± 192.91239 ± 164.51898 ± 190.41109 ± 88.51323 ± 118.6
Binge protein (% kcal)20.1 ± 1.6120.1 ± 2.5116.4 ± 1.6723.0 ± 3.3923.7 ± 2.62
Binge carbohydrate (% kcal)42.4 ± 1.5639.8 ± 5.3943.7 ± 1.6039.5 ± 2.9345.2 ± 1.39
Binge fat (% kcal)37.4 ± 1.8340.1 ± 3.5639.9 ± 1.2737.8 ± 1.4231.3 ± 2.82
Binge satiety rating (mm)108.6 ± 21.479.2 ± 11.3126.9 ± 26.2103.1 ± 18.4103.8 ± 20.5
Normal intake (kcal)1539 ± 141.71091 ± 107.81405 ± 214.51007 ± 179.31146 ± 101.7
Normal carbohydrate (% kcal)37.9 ± 2.5036.2 ± 4.9340.3 ± 3.1841.1 ± 4.1547.1 ± 2.18
Normal fat (% kcal)36.4 ± 2.9340.6 ± 2.7039.4 ± 2.6736.4 ± 2.0029.4 ± 2.57
Normal protein (% kcal)25.5 ± 2.2723.2 ± 2.9220.7 ± 2.5520.1 ± 4.5423.5 ± 3.01
Normal satiety rating (mm)89.3 ± 12.5101.8 ± 20.699.6 ± 15.3104.3 ± 22.0104.1 ± 18.8

The Three-Factor Eating Questionnaire

Significant differences were found among the five groups for the disinhibition (F4,37 = 51.14, p < 0.0001) and hunger (F4,37 = 15.48, p < 0.0001) subscales of the TFEQ, and there was a trend toward a significant difference among the groups for the restraint subscale (F4,37 = 2.4, p = 0.06) (see Table 1 for group means). The mean disinhibition score of obese subjects with BED was significantly higher than that of obese subjects without BED. The mean disinhibition score of the normal-weight control group, however, was significantly lower than that of all obese groups (critical range = 2.16). The mean score of the hunger subscale of the TFEQ was also significantly higher among obese subjects with BED than among obese groups without BED, as well as higher than the mean hunger score found for the normal-weight control group (critical range = 2.67). The mean score for the restraint subscale among normal-weight controls was significantly lower than the mean restraint score for less-obese subjects without BED (4.6 vs. 9.5; critical range = 3.7); there were no other significant differences among the groups.


Mean energy intake differed significantly among the five subject groups (F4,37 = 9.86, p < 0.0001; see Figure 1 and Table 1 for mean intakes for each group and meal). Differences between diagnostic and weight groups, (within and between meals, when subjects were instructed to either binge or to eat normally) are described below.

Figure 1.

Mean amount consumed (intake) for each group; ○ and •, represent more-obese subjects; ▵ and ▴, represent less-obese subjects; open symbols represent subjects with BED; closed symbols represent subjects without BED. ▪, represent normal-weight controls. LSD, least-significant difference.

Intake: Effects of Diagnosis (BED vs. non-BED).

The average energy intake, across meals with binge and normal instructions, of obese subjects (across more and less-obese groups) with BED (1807 kcal) was significantly greater [difference = 696 kcal ± 121 SE of the difference (SED); t1,37 = 5.74, p < 0.0001] than that of obese subjects without BED (1111 kcal) and than that of normal-weight controls (1235 kcal; difference = 573 kcal ± 140 SED; t1,37 = 4.09, p = 0.0002).

Intake: Effects of Instructions (Binge vs. Normal).

Meals on which subjects were instructed to binge (1591 kcal) were significantly larger (difference = 353.6 kcal ± 101.5 SED; t1,37 = 3.48, p = 0.001) than meals on which subjects were instructed to eat normally (1238 kcal). However, this instruction effect was only significant among subjects with BED, resulting in a significant diagnosis by instruction interaction (t1,37 = 2.45, p = 0.019). Differences between groups in the instruction effects are described below.

Following the instructions to binge, both obese groups with BED (2143 kcal) ate significantly more than both obese groups without BED (1174 kcal) (difference = 969 kcal ± 171 SED; t1,37 = 5.65, p < 0.0001). The less-obese BED subjects ate 789 kcal (± 239 SED; t1,37 = 3.30, p = 0.001) more than the weight-matched subjects without BED, and the more-obese BED subjects ate 1149 kcal (± 246 SED; t1,37 = 4.67, p < 0.0001) more than weight-matched subjects without BED.

Following the instructions to eat normally, both obese groups with BED ate (1472 kcal) significantly more than both obese groups without BED (1049 kcal) (difference = 423 kcal ± 171 SED; t1,37 = 2.46, p = 0.016). This 546-kcal (± 223 SED) difference between BED and non-BED groups, in the incremental increase of intake during the binge vs. the normal instruction, accounts for the significant diagnosis by instruction interaction reported above.

Another way of stating this interaction is that subjects with BED ate 671 kcal (± 141 SED) more during the binge meal than during the normal meal (t1,37 = 4.76, p < 0.0001), whereas subjects without BED ate only 125 kcal (± 173 SED) more (not significant) during the binge meal than during the normal meal. Hence, only subjects with BED ate significantly more when they were asked to binge than when they were asked to eat normally.

Intake: Effects of Weight (More Obese, Less Obese, Normal Weight).

There were no overall differences in intake among groups attributable to weight group alone, nor was there a significant weight-group by diagnosis interaction.

Intake: Effect of Weight within Diagnostic Group.

The mean energy intake of more-obese subjects with BED (2388 kcal) was significantly greater than that of less-obese subjects with BED (1898 kcal) during meals preceded by the binge instruction (difference = 490 kcal ± 216.9 SED; t1,72 = 2.26, p = 0.026), but not during meals preceded by instructions to eat normally (difference = 134.3 kcal ± 216.9 SED; t1,72 = 0.62, p = 0.54).

Despite the fact that intakes differed significantly between more- and less-obese BED subjects only after the binge (difference = 490 kcal), but not after the normal (difference = 134.3 kcal) instruction, the weight-group x instruction interaction was not significant (difference = 360 kcal ± 343 SED; t1,37 = 1.05, p = 0.297). That is, the difference between intake after instructions to binge and intake after instructions to eat normally did not differ significantly between more- and less-obese BED subjects. Mean intakes were not significantly different among the obese and normal-weight groups without BED following either instruction. Furthermore, among the subjects without BED, intakes did not differ between meals preceded by instructions to eat normally or to binge eat.

Correlation of Intake and BMI

Among the subjects with BED, there was a significant, positive correlation between meal size and BMI (r2 = 0.2, slope = 49.9, p = 0.04; see Figure 2) when subjects were instructed to binge. This relationship between BMI and energy intake was not present when BED subjects were instructed to eat normally, nor was it significant among subjects without BED (either obese or normal weight) for either instruction (see Figure 2).

Figure 2.

Correlation between binge-meal intake and BMI among obese subjects with (•) and without (▴) BED.

Macronutrient Consumption

Both obese groups consumed a significantly greater percentage of their total energy from fat (38.5%) than did normal-weight controls (30.8%), across instructions and diagnostic groups (average difference = 8.09% ± 2.6 SED more energy from fat consumed by obese groups compared with normal-weight controls; t1,37 = 2.78, p = 0.008). There was no significant difference in the percentage of fat consumed among all four groups of overweight subjects, neither between the two BMI groups nor between subjects with BED vs. those without BED. Meal instruction also had no effect on the percentage of total energy consumed from fat (see Table 1).

The proportions of energy consumed from carbohydrate and protein sources did not differ between lean and obese subjects, nor did it differ between meals that were preceded by different instructions. On average, subjects consumed 21.5% of their total energy from protein and 41.25% of their total energy from carbohydrate (all groups and meals combined).

Meal-Related Ratings

There were no differences between groups in ratings of liking of the test meals, nor in ratings of hunger or satiety before the meals began. Among the obese subjects with BED, the mean satiety rating after the meal (117.7 mm) preceded by binge instructions was significantly higher than that after the meal preceded by instructions to eat normally (94.4 mm; difference = 23.3 mm ± 9.5 SED; t1,37 = 2.45, p = 0.01). In contrast, among obese subjects without BED, the mean satiety rating after the meal (91.15 mm) preceded by binge instructions was actually slightly lower than that after the meal preceded by instructions to eat normally (103.05 mm; difference = −11.9 mm ± 11.7 SED; not significant). This 35.3-mm (± 15.1 SED) binge vs. non-binge instruction difference in post-meal satiety ratings, between the BED and non-BED groups, resulted in a significant diagnosis by instruction interaction (t1,37 = 2.34, p = 0.025). This interaction, between instruction and diagnosis, occurred in the same direction, for the same meals, as the interaction observed for differences in energy intake.


The major findings of this study were the following: 1) when asked to binge, the more-obese subjects with BED ate more than the less-obese subjects with BED; 2) when asked to eat normally or to binge, more- and less-obese subjects with BED ate more than did weight-matched or normal-weight subjects without BED; and 3) only subjects with BED ate more when they were asked to binge eat than when they were asked to eat normally. Equally obese and normal-weight subjects without BED did not eat more when they were asked to overeat. In addition, this study found that meal size was related to BMI, but only among subjects with BED and only when they had been instructed to binge. These findings confirm and extend previous work and raise challenging new questions about the eating behavior of subjects with BED. Specifically, what mechanisms account for the overeating that characterizes BED? Why is the meal size of these individuals proportional to their body size, but only when they are asked to binge?

Mechanisms That May Account for Overeating in Subjects with BED

One hypothesis that might help account for the difference in energy intake between obese subjects with and without BED is that the total energy expenditure (TEE) of subjects with BED is higher than that of equally obese individuals without BED. If this were the case, then the excessive intakes of subjects with BED would simply be a behavioral consequence of a physiological requirement for higher energy intake. However, several studies have considered this possibility and have failed to detect differences in any metabolic parameter (22, 23). In one study (22), 86 obese women were assessed for binge eating status using a standardized questionnaire (24). Metabolic parameters, including thyroid function, resting metabolic rate, and fat-free mass were compared between binge eaters and subjects who did not engage in such behavior. No differences on any metabolic parameter were found. In a more recent study, Alger et al. (23) compared food consumption and energy expenditure between nine obese women with BED and nine obese women without BED during an inpatient stay on a metabolic ward. No differences in metabolic parameters or in average daily energy intake were observed between subjects with and without BED. Thus, these studies suggest that a difference in energy requirements is unlikely to account for the differences in energy intake in laboratory meals between subjects with and without BED.

Because there is no evidence of a metabolic disturbance in subjects with BED, their average 24-hour energy intakes probably do not differ from those of subjects without BED who have a similar BMI. To compensate energetically for binge eating, subjects with BED may eat fewer, smaller meals when not overeating. Such a phenomenon is suggested by the data of Alger et al. (23), who found that daily energy intake among subjects with BED varied much more than it did among subjects without BED.

Another possible explanation for the binge eating in subjects with BED is that they lack certain inhibitory factors that typically terminate an eating episode before it becomes very large. This lack of inhibition could be physiological, psychological, or both. The significantly elevated disinhibition scores on the TFEQ reported by subjects with BED in this study, and in others (3, 25), suggest that a loss of inhibition over eating is a cardinal feature of this eating disorder (26, 27).

How is a lack of inhibition to be understood? One possibility is that these individuals have some physiological deficiency that results in a lack of inhibitory control over meal size. The physiological mechanisms known to mediate such sensations in humans include distention of the stomach (28), activation of chemoreceptors in the intestine, levels of hormones and metabolites in the plasma, and the release of neurotransmitters in the brain (29, 30). Disturbances in any of these physiological processes could result in subjects’ requiring more food before they could experience sensations of fullness that inhibit further consumption. Patients with other disorders of binge eating also exhibit disturbances in such systems; for example, gastric emptying is delayed, and postprandial release of cholecystokinin is diminished in patients with BN (31, 32, 33). Examination of such postprandial physiological mediators of satiety may be warranted among patients with BED.

Alternatively, the lack of inhibition over the eating behavior of subjects with BED may be related to the presence of psychological disturbances. It is well established that subjects with BED have more symptoms of anxiety and depression than comparably obese subjects without BED (34, 35, 36). Several studies have argued that binge eating might simply be a form of self-medication (37, 38). Others have suggested that some obese subjects may experience a greater reward value from food, overriding potential inhibitory influences on intake, and rendering such individuals easily disinhibited (39). Thus, elevated disinhibition scores may result from a combination of physiological and psychological disturbances that together result in the marked overeating observed among subjects with BED.

Mechanisms to Account for the Relationship between Binge Size and BMI in BED

The loss of inhibition over the eating behavior described above, which is so characteristic of subjects with BED (particularly during binge episodes), may permit the underlying, predicted relationship between body size and food intake to be observed in a single laboratory meal. That is, the instruction to binge may act to further disinhibit subjects with BED, whose eating is, at baseline, disinhibited. From a physiological or thermodynamic perspective, it is not surprising that energy intake is proportional to body size. Such a relationship has been extremely well documented among animal species (40), and the most accurate equations to calculate human total energy expenditure are dependent on measures of body size (41). Telch and Agras (9) also reported a significant relationship between BMI and meals consumed in a laboratory setting, although that study did not distinguish between binge and normal meals. This hypothesis predicts that if suitable means were found to remove inhibition from subjects without BED (who are not disinhibited), then they too would show a relationship between BMI and meal size.

Why Don't Meal Size and BMI Correlate in Individuals without BED?

Because meal size and BMI were only related among disinhibited subjects during binge meals, the lack of a correlation among subjects without BED may be due to basal inhibitory influences that terminate individual meals before they comprise a significant proportion of one's daily energy needs. This inhibition could be similarly due to both physiological and psychological factors, some of which may have been enhanced by the laboratory environment. For example, individuals who are not easily disinhibited might only prefer to eat very large meals (i.e., binges) when they have the opportunity to select or prepare the foods themselves or be in the company of family and friends, such as at a holiday feast.

Another possibility is that individuals without BED are more sensitive to sensations of satiety than are subjects with BED, conceivably because of differences in the physiological consequences of food ingestion. Differences in both environmental and physiological feeding-inhibitory factors may therefore account for the difference in meal size among the obese subjects without BED, as well as account for the absence of a clear relationship between BMI and intake among these subjects.

Macronutrient Composition

It was not surprising to find that obese individuals consumed a greater proportion of energy from fat compared with normal-weight individuals. These data are consistent with numerous epidemiological surveys in which associations between fat intake (or some correlate) and adiposity are often reported (42, 43). The similar proportions of macronutrients comprising both binge and non-binge meals, for both subjects with and without BED, confirm previous reports (44) and indicate that differences in macronutrient selection of meals do not relate to the abnormal eating associated with BED.

Satiety Ratings

As noted in the results, the obese subjects with BED reported significantly higher satiety ratings after consuming significantly larger meals (i.e., those on which they were instructed to binge). The normal-weight subjects, whose intakes did not differ between test meals, reported similar satiety ratings after each test meal. In contrast, the obese subjects without BED, whose intakes also did not differ between test meals, reported significantly lower satiety ratings after the meal in which they were asked to binge, than after that in which they were asked to eat normally. It seems that more-obese subjects without BED may have gauged their post-meal satiety responses according to how satiated they anticipated feeling after binge vs. normal meals. In other words, the more-obese non-BED subjects may have expected to feel more satiated after eating the meal in which they were instructed to binge. However, because they had not actually consumed any more than normal, they felt less satisfied than after they had been instructed to eat normally.

Another possibility is that this group was restricting their intake more than the other groups, such that when they were asked to “let go and binge eat,” they felt compelled to behave even more restrictively. This restrictive behavior could explain why they did not feel as satisfied by the binge meal compared with the normal meal, despite the fact that they had eaten similar amounts during both meals.


This study demonstrates that subjects with BED eat abnormally large meals in a laboratory setting, particularly when they are asked to binge. Among subjects with BED, the more-obese the subject, the larger the size of her binge. This study therefore confirms that the diagnosis of BED is consistently associated with disturbances in eating behavior that can be quantified in a laboratory setting. Laboratory studies of the eating behavior among subjects with this disorder may be a useful method of examining novel treatment interventions for this population and may be used to further investigate the mechanisms that underlie disturbances in eating behavior.


The assistance of Julie Goldfein, Ph.D., and Sara Wolk, Ph.D., in the recruitment of subjects is gratefully acknowledged. This work was supported by MH42206 (to B.T.W.) and in part by Obesity Core Center NIH Grant OX26687.