Physiology & Behavior 77 (2002) 45 – 54

A new animal model of binge eating:

Key synergistic role of past caloric restriction and stress
M.M. Hagan*, P.K. Wauford, P.C. Chandler, L.A. Jarrett, R.J. Rybak, K. Blackburn
Department of Psychology, Behavioral Neuroscience Division, 415 Campbell Hall, 1300 University Boulevard,
University of Alabama at Birmingham, Birmingham, AL 35294-1170, USA
Received 10 December 2001; received in revised form 25 March 2002; accepted 9 May 2002

Abstract

Dieting and stress are important in the etiology and maintenance of eating disorders, and dieting strongly predicts stress-induced
overeating in humans. We hypothesized that caloric restriction and stress interact in a unique manner to promote binge eating. To test this
hypothesis, a group of young female rats were cycled through a restriction period (4 days of 66% of control food intake) followed by 6 days
of free feeding prior to being stressed by acute foot shock. After three of these cycles, the food intake of rats exposed only to restriction (R),
or only to stress (S), did not differ from controls. However, R + S rats that were restricted and refed, despite normal body weight and food
intake after free feeding, engaged in a powerful bout of hyperphagia when stressed (Experiment 1). The R + S effect was replicated in an
older group of rats (Experiment 2). The hyperphagia was characteristically binge-like, it constituted a 40% selective increase in highly
palatable (HP) food ( P < .001) over a discrete period of time (within 24 h post-stress), and reflected feeding for reward (higher HP:chow
ratio) over metabolic need as occurred after restriction (higher chow:HP ratio). Subsequent experiments revealed that binge eating did not
occur if only chow was available (Experiment 3) or if restriction – refeeding (R – R) did not proximally precede stress (Experiment 4).
Experiment 5 revealed that a history of R – R cycles followed by only one stress episode was sufficient to increase intake to 53% above
controls as early as 2 h after stress ( P < .001). This animal model of binge eating should facilitate investigations into the neurochemical
changes induced by dieting and environmental stress to produce disordered eating and provide a preclinical tool to test preventive strategies
and treatments more relevant to bulimia nervosa, multiple cases of binge eating disorder (BED) and binge-purge type anorexia nervosa.
D 2002 Elsevier Science Inc. All rights reserved.

Keywords: Eating disorders; Bulimia; Binge eating disorder; Stress-induced eating; Rats; Hyperphagia; Reward; Foot shock; Opioids; Behavior; Refeeding;
Food intake; Female; Dieting

1. Introduction stress in their precipitation and maintenance [1,6 – 9].

Binge eating is a central feature of bulimia nervosa,
binge eating disorder (BED) and many cases of anorexia
nervosa [1]. These disorders exact a significant toll on the
physical and psychological health of millions, and medical
consequences are substantially multiplied when binge eating
leads to obesity and diabetes [2]. Much is unknown regard-
ing the etiology, psychophysiology and optimal course of
treatment of BEDs. However, considerable evidence points
to a history of food restriction [1,3 – 5] and of environmental
* Corresponding author. Tel.: +1-205-934-2439; fax: +1-205-975-
6110.
E-mail address: mhagan@uab.edu (M.M. Hagan).
Among the nonclinical population, food restriction is con-
sistently the strongest predictor of overeating in response to
stress [2,10 – 17]. These observations suggest that some
variants of binge eating may be caused by a unique
interaction between dieting and stress.
Animal models afford tight experimental control over
variables hypothesized to cause a behavior. Of the few
available models of eating disorders, two critical factors
emerge in the production of binge eating: a history of food
restriction [18,19] and access to highly palatable (HP) food
[19 –21]. However, no animal model to date has considered
the potentially unique interaction between experience with
dieting and environmental stress on food intake. Such a
model would allow us greater understanding of the behav-
ioral contingencies by which dieting and stress predispose

0031-9384/02/$ – see front matter D 2002 Elsevier Science Inc. All rights reserved.
PII: S 0 0 3 1 - 9 3 8 4 ( 0 2 ) 0 0 8 0 9 - 0
46 M.M. Hagan et al. / Physiology & Behavior 77 (2002) 45–54
or precipitate binge eating and would provide a window into
the underlying neurochemical changes mediating this effect.
Animal stress paradigms have not typically included food
restriction. In one exception, food intake was measured in
rats while at 80% of normal body weight. Deprived, foot-
shocked rats increased their intake relative to deprived,
nonshocked rats; however, this intake was less than that of
sated, nonshocked (or control) rats [22]. In a second study,
foot shock enhanced food intake of calorie-restricted rats,
but this intake was measured while rats were rebounding
from food restriction and weight loss [23]. In general, most
environmental stressors (e.g. immobilization, cold or water
exposure, social defeat, noise, emotional stress) decrease
food intake in rats. Electric shock also typically suppresses
food intake, but in some instances increases or has no affect
on intake [11,24,25]. We propose that a veridical model of
clinical binge eating should incorporate environmental
stress but demonstrate hyperphagia in animals that are not
in energy deficit because satiation and normal body weight
are typical of binge-eaters [1,26 – 29].
Also typical of clinical binge eating is the salience of
HP food that is energy dense in fat and sugar. The binge-
triggering effect, craving, preferential selection and ulti-
mate overconsumption of HP food is omnipresent in BEDs
[1,26,28 –30]. HP food is also preferentially consumed in
nonclinical, stress-induced overeaters [17,31 – 34]. In ani-
mals, access to HP food modifies stress-induced feeding
behavior [35 –39] and is necessary in evoking hyperphagia
observed in other animal models of bulimia and BED
[19 – 21]. Hence, access to HP food may be an important
factor in binge eating produced by an interaction be-
tween dieting and stress.
The goal of this study, therefore, was to test the hypo-
thesis that dieting and environmental stress interact in a
synergistic manner to produce binge eating and that access
to HP food is critical to this response. To test this hypo-
thesis, rats were subjected to brief cycles of caloric restric-
tion followed by ad lib refeeding prior to being exposed to
an acute episode of stress via foot shock. In keeping with the
prevalence of adolescent and young adult females [1] in
BEDs, young female rats were used. To rationally test for a
significant interaction between dieting and stress, an intens-
ity and duration of both food restriction and stress were used
that alone did not affect food intake relative to control rats.
Subsequent tests were designed to elucidate some contin-
gencies required to evoke the binge-like response.
2. Experiments
2.1. Experiment 1: effect of caloric restriction and stress on
food intake in young post-pubescent rats
2.1.1. Rationale
Experiment 1 was conducted to assess the effect of
experience with caloric restriction alone, of shock-induced
stress alone and of the combination of both on food intake
of rats.
2.1.2. Subjects
A total of N = 32, 52-day-old female Sprague – Dawley
rats were acclimated to individual bedded cages under a
12-h light/dark cycle (lights on at 0600 h) with ad lib
chow and water for 2 weeks prior to the experiment. The
chow was the same used throughout the feeding tests
(Harlan Teklad Global Diets; 3.3 kcal/g; 69.8%, 3.5%,
16.7% and 10% of kcal from carbohydrate, fat, protein and
moisture, respectively). The rats were then weight-matched
into one of four groups (n = 8 per group): a nonrestricted
with no stress group (NR + NS), a nonrestricted with stress
group (NR + S), a restricted with no stress group (R + NS)
and a restricted with stress group (R + S). Once assigned to
one of these groups, the rats remained in that group
throughout the study.
2.1.3. Restriction phase of restriction – refeeding
(R – R)/stress cycling procedures
Rats in the restricted groups (R + NS and R + S) were
given 66% of the mean daily chow intake of rats in the
nonrestricted groups for 4 consecutive days. Regular rat
chow was placed in the cages at 10:00 every morning (4 h
after lights on). Body weights and food intake were recorded
daily. By the last day of restriction, rats weighed within 91–
93% of controls’ (nonrestricted rats’) body weight. Immedi-
ately following the last day of restriction, the restricted
groups were allowed to ‘refeed’ for 6 consecutive days on
ad lib chow. By the last day of refeeding, the body weight
and food intake of restricted rats were within 95 –105% of
nonrestricted rats (a nonsignificant difference). Within the
two restricted subgroups (R + S and R + NS), rats were
within 99% body weight and food intake of each other. This
precluded potentially confounding effects of hunger or
energy deficit on observed differences in intake following
stress. Following the last day of refeeding, rats were exposed
to the shock environment (described below). Together, the
R – R episode followed by acute stress constituted a ‘R – R/S
cycle.’ This protocol of food-restriction and refeeding back
to normal body weight was guided by an effort to simulate
cyclic dieting and disinhibition of food among binge eaters
and bulimics whose body weight is also typically normal to
overweight, as opposed to anorexia nervosa patients who are
underweight.
2.1.4. Stress phase of R – R/S cycling procedures
Immediately after the last day of R –R, all chow was
removed from the cages at 09:00 h. At 11:00 h, the rats were
transported from the animal colony in their individual cages
to a separate room housing the shock apparatus. This
apparatus consisted of four 70 ! 9.5 cm closed runways
with metal bar floors (Coulbourn Instruments Habitest
System, Allentown, PA). Two animals belonging to the
same group were each placed in a separate runway at the
 M.M. Hagan et al. / Physiology & Behavior 77 (2002) 45–54 47

same time. For rats in the stressed groups (NR + S and

R + S), electric shock was delivered through the metal bar
floor via a precision regulated shocker programmed to
deliver 0.6 mA over 3 s, four different times, with 15 s of
no shock in between each time. The total time spent in the
alley was 87 s. This included a 15-s delay prior to the first
shock delivery and a 15-s delay at the end of the fourth
shock delivery. The 0.6-mA level of intensity was found in
prior pilot tests to evoke clear signs of stress including an
increase in high-pitched vocalization, motor activity and
defecation. With repetition and anticipation, shock produced
fear, manifested by freezing behavior. These behaviors and
concurrent activation of the HPA axis have verified foot
shock as a stressful experience [40 –42]. Rats in the no-
stress groups (NR + NS and R + NS) were placed in deo-
dorized shock-free runways for the same amount of time.
Leaving these rats in the remote animal colony while rats in
the stress groups were being shocked minimized the poten-
tial for NS rats to be psychologically stressed by overheard
vocalizations.

2.1.5. Feeding test following stress

Immediately following time in the shock apparatus, rats
were returned to the animal colony. A premeasured amount
of chow and HP food was placed inside each cage and was
measured 1, 2, 3, 4, 24 and 48 h after, taking care to collect
all spillage. Water was available at all times. Food intake is
expressed as the group mean in kcal ± S.E.M.
2.1.6. Introduction of HP food into R – R/S cycling
After the last record of food intake was taken following
stress, the rats immediately were started on a second R –R/S
cycle according to their group assignment. Because no
effect on food intake was observed after the second R –R/S
cycle, HP food was introduced [21]. The rats were again
cycled through the restriction and refeeding phase, for a
third time, but this time, on the first 2 days of refeeding, the
rats were given access to ad lib chow and ad lib HP food
(Nabisco Oreo cookies; 4.7 kcal/g; 57%, 40% and 3% of
kcal from carbohydrate, fat and protein, respectively), while
the nonrestricted groups were also given ad lib HP food in
addition to chow. During this time, as expected after caloric
restriction, the restricted groups were hyperphagic. The
subsequent 4 days of refeeding consisted of ad lib chow
only (no HP food). Following the last day of refeeding, rats
were placed into the shock runway for shock-induced stress
(or no stress according to the assigned group conditions) as
described above.
Fig. 1. The effect of experience with R or NR on food intake in response to S
or NS. Panel A: 24-h intake after the second R – R/S cycle that included only
chow (no differences between groups). Panel B: 24-h intake after the third
R – R/S cycle that included HP food (cookies) and chow together for the first
time. All groups were tested under sated conditions; restricted rats (R + NS
and R + S group) having access to ad lib food for 6 days prior to S/NS.
* * P < .01 with Tukey’s HSD (total kcal different from other groups).
they ate significantly more than the other groups (Fig. 1B).
This increase in intake was due entirely to a 33% and 22–
28% greater intake of HP food than that of nonstressed
restricted rats and nonrestricted rats, respectively ( P < .01;
Fig. 1B, dark bars). Chow intake (Fig. 1B, hatched bars) did
not differ among groups. Despite accessibility to HP food,
stress alone (NR + S) and restriction alone (R + NS) had no
influence on intake, supporting a unique interaction between
restriction and stress to elicit overeating. The overeating of
(R + S) rats reflected binge-like behavior in that it occurred in
a discrete period of time (not lasting beyond 24 h) and was
comprised of HP food.
2.2. Experiment 2: effect of early introduction of HP food in
R – R/S cycling on food intake in young adult rats

2.1.7. Results 2.2.1. Rationale

The food intake of rats subjected to R – R alone, stress
alone or both did not differ after the first or second R –R/S
cycles, when only rat chow was available after stress
(Fig. 1A). However, after the third R –R/S cycle, at which
time HP food was introduced, the (R + S) group began to
steadily increase their food intake so that by 24 h after stress,
Experiment 2 addressed whether the binge-like eating
found in Experiment 1 would (1) replicate in a group of
older female rats; (2) whether the effect was solely contin-
gent on availability of HP food; or (3) whether, despite
access to HP food, at least two R –R/S cycles were still
required for the binge-like response.
48 M.M. Hagan et al. / Physiology & Behavior 77 (2002) 45–54

2.2.2. Subjects
A total of N = 25, 97-day-old female Sprague – Dawley
rats were acclimated to the same environmental conditions as
rats in Experiment 1 and were likewise weight-matched into
one of the same four type of groups (n = 6– 7 per group): a
nonrestricted with no stress group (NR + NS), a nonrestricted
with stress group (NR + S), a restricted with no stress group
(R + NS) and a restricted with stress group (R + S).

2.2.3. R –R/S cycling procedures

The rats were cycled three different times through their
corresponding R – R and stress conditions as described for
the last cycle of Experiment 1 that included HP food. Hence,
ad lib HP food was present during the first 2 days of
refeeding and after stress in all three cycles. With these
older rats, the restriction phase was 5 days long, as opposed
to 4 days long in the younger rats. However, as was the case
with the younger rats in Experiment 1, body weights never
fell below 90% of controls.

2.2.4. Results
Surprisingly, as with the younger rats in Experiment 1,
and despite access to HP food from the first R – R/S cycle,
no difference in intake emerged in (R + S) rats after the
first or second R – R/S cycle (not shown). However, and
mirroring the behavior of younger rats in Experiment 1,
when rats were subjected to a third experience with R –R,
they became markedly hyperphagic in response to stress.
This time, the increase in food intake of (R + S) rats was
evident as early as 3 h ( P < .01; not shown) and 4 h after
stress (Fig. 2A), and continued for 24 h (Fig. 2B).
Replicating the hyperphagia in younger rats (Experiment
1), the increased feeding at all of these time points was
again due entirely to a preferential increase in HP food
intake that was at least 40% higher at 4 h ( P < .001) and
24 h ( P < .01) after stress than that of all other groups,
none of which differed from each other. Again, the binge-
like effect lasted for a discrete period of time, not
persisting beyond 24 h after stress. In summary, experi-
ence with three R –R/S cycles, despite HP food from the
onset, was required to evoke binge eating in response to
stress.
2.3. Experiment 3: effect of R –R/S cycling on food intake
when only chow and no HP food is available after stress
2.3.1. Rationale
Experiment 3 was conducted to examine (1) whether the
binge-like effect would be expressed if only chow, and no
HP food, was available after stress; and (2) whether the
motivation to overeat would persist beyond 24 h after stress
if HP food were again made available at that time.
2.3.2. Subjects
The same rats that served in Experiment 2 served in
this experiment.
Fig. 2. The effect of experience with R or NR on food intake in response
to S or NS. Panel A: 4-h intake after the third R – R/S cycle that included
HP food. Panel B: 24-h intake after the third R – R/S cycle that included
HP food. All groups were tested under sated conditions; restricted rats
(R + NS and R + S group) having access to ad lib food for 6 days prior to
S/NS. * * * P < .001, * * P < .01 with Tukey’s HSD (total kcal different from
othergroups).
2.3.3. R – R/S cycling procedures
Immediately following the post-stress feeding test of
Experiment 2, the rats were once more cycled through their
corresponding R – R/S regimen, but this time only chow was
available during the post-stress feeding test. Food intake
was recorded for up to 24 h; HP food was then added to the
chow and intake recorded for a second 24-h period.
2.3.4. Results
As shown in Fig. 3A, when only chow was available after
stress, rats with R – R experience failed to overeat in response
to stress. There was a nonsignificant trend for restricted
groups to increase chow intake 4 h post-stress (not shown),
but the effect dissipated with all groups eating the same
amount of chow within 24 h (Fig. 3A). Surprisingly, when HP
food was reintroduced 24– 48 h after stress (Fig. 3B), there
was again no difference in intake of chow (14.4 ± 4.7 vs.
14.8 ± 3.6) or HP food (71.2 ± 3.2 vs. 74.2 ± 5.2 kcal)
between the (R + S) group and the NR + NS group, respect-
ively, or with any of the other groups. The lack of differences
in chow intake between groups served as further confirmation
that restricted rats had enough time to normalize their food
 M.M. Hagan et al. / Physiology & Behavior 77 (2002) 45–54
Fig. 3. The effect of experience (four cycles) with R or NR on 24-h food
intake in response to S or NS when only chow was available after stress
(Panel A) and when HP food was reintroduced during the second 24-h
period (Panel B). All groups were tested under sated conditions with
restricted rats (R + NS and R + S group) having access to ad lib food for 6
days prior to S/NS. No differences between groups was observed in chow
or HP food intake with Tukey’s HSD.
intake during refeeding prior to stress. Secondly, these results
indicated that the binge-like effect was contingent upon close
temporal proximity between stress and access to HP food.
2.4. Experiment 4: effect of omitting restriction from R –R/S
cycling on food intake
2.4.1. Rationale
Experiment 4 was conducted to assess whether stress
after restriction-induced hyperphagia, despite previous
experience with repeated R – R/S cycles, was contingent
on R –R always preceding stress.
2.4.2. Subjects
The same animals that served in Experiments 2 and 3
served in this experiment.
2.4.3. R –R/S cycling procedures
Immediately after the last post-stress recording of food
intake in Experiment 3, all of the groups including the
restriction groups (R + NS and R + S) were given free access
to ad lib chow for 4 days until they were again exposed to
the stress protocol.
2.4.4. Results
As was the case when no HP food was available after
stress, the absence of restriction and restriction-induced
49
refeeding preceding stress, despite availability of HP food
after stress, precluded binge-like eating in (R + S) rats. No
differences in intake were observed shortly after and 24 h
post-stress (Fig. 4). There was a trend for the (R + S) group to
eat less chow after this time, but this decrease was not
significant (Fig. 4, hatched bars). Surprisingly, the (R + S)
group did not differ in intake of HP food (Fig. 4, dark bars),
weakening support for the possibility that the rats had formed
a learned association between stress and HP feeding. In
summary, Experiments 1 and 2 show that at least three R – R/S
cycles were necessary to produce the binge-like effect.
Results from Experiment 4 indicate that restriction is critical
and must continue to precede stress to produce this effect.
2.5. Experiment 5: effect of repeated restriction bouts but
only a single experience of stress on food intake
2.5.1. Rationale
Experiment 5 addressed whether repeated experience
with stress was as critical as repeated experience with
restriction (i.e. were at least three experiences with stress
necessary to produce the binge-like effect or could just a
single stress experience elicit the effect in restricting rats).
2.5.2. Subjects
The same animals were used that served in Experiments
2 –4 as they had prior history with restriction and/or stress.
2.5.3. R – R/S cycling procedures
By the end of Experiment 4, all rats had been cycled
through their corresponding R – R/S regimen (which
included HP food at refeeding) five times. They were then
cycled through their corresponding regimens two more times
for a history of seven R– R/S cycles. Up until this point, the
nonstressed groups (NR + NS and R + NS) were naı̈ve to foot
shock, but after the seventh cycle they were shocked for the
first time. In contrast, the stress-experienced rats (NR + S and
R + S), with a history of seven foot shock episodes, were now
Fig. 4. Effect of a history of experience (five cycles) with R or NR on 24-h
food intake in response to S or NS when no R – R period immediately
preceded the stress event. No differences between groups was found in
chow or HP food intake with Tukey’s HSD.
50 M.M. Hagan et al. / Physiology & Behavior 77 (2002) 45–54

values relative to controls). Data from Experiment 2 is

presented to illustrate this pattern.

2.6.2. Results
On the first day of refeeding following a period of caloric
restriction, rats ate proportionally more chow (66% of total
intake) than HP food (34% of total intake). This was in
contrast to nonrestricted rats who ate only 16% of their total
intake from chow and 84% from HP food at the same time
the restricted rats were allowed to refeed on HP food and
chow ( P < .001). This increased preference for chow by the
restricted rats compared to the nonrestricted rats is depicted
in Fig. 6A. Since there was no difference within the two
restricted groups and within the two nonrestricted groups,
the data were collapsed into single bars (Fig. 6A). However,
following refeeding and normalization of body weight and
food intake, the high chow:low HP food ratio of restricted
rats shifted to a high HP food:low chow intake pattern, a
pattern that did not differ from that of the nonrestricted rats

Fig. 5. Effect of a history (seven cycles) of R or NR on food intake in

response to first time stress or first time being spared of stress. Panel A: 2-h
intake. Panel B: 24-h intake. All groups were tested under sated conditions;
restricted rats (R + NS and R + S group) having access to ad lib food for 6
days prior to S/NS. * * * P < .001 (total kcal different from other groups);
chow intake did not differ between groups with Tukey’s HSD.

being spared from shock for the first time. While the NS
groups were being shocked for the first time, the stress-
experienced rats were left in their home cages to avoid
anxiety associated with the trip to the shock apparatus.

2.5.4. Results
Rats with a history of seven R – R cycles but with no prior
experience with shock (R + NS) engaged in a powerful 53%
increase of HP food as early as 2 h after their first exposure to
foot shock stress (Fig. 5A, P < .001). This hyperphagia was
sustained across 4 h ( P < .001, not shown) and up to 24 h
(Fig. 5B, P < .001) following acute stress. Interestingly,
despite a long history of R + S, these rats failed to exhibit
binge-like intake when not stressed (Fig. 5B, third bar).

2.6. Results of food selection pattern comparisons within

restricted groups during refeeding and after stress
2.6.1. Rationale for analysis of food selection patterns
Throughout the aforementioned experiments, it was
noted that the pattern of food selection (% of chow vs.
HP food) among restricted rats (R + NS and R + S groups)
differed depending on whether the rats were refeeding in
response to caloric-restriction (when hungry) or were ad lib
feeding after a period of refeeding (when sated as assumed
from normalization of their body weight and food intake
Fig. 6. The effect of experience with R or NR on food type partitioning in
24-h intake during the first day of refeeding (Panel A) and after 6 days of
refeeding following stress procedures (Panel B). * * * P < .001 difference
between chow:HP food ratios of nonrestricted and restricted groups;
#
P < .001 difference in chow:HP food ratio of restricted groups following
caloric restriction (Panel A) compared to chow:HP food ratio following
refeeding and stress (Panel B). Data represent collapsed means between the
two restriction subgroups (R + NS and R + S) and the two nonrestriction
subgroups (NR + NS and NR + S) as chow:HP food ratios did not differ
between them.
 M.M. Hagan et al. / Physiology & Behavior 77 (2002) 45–54
(Fig. 6B). The shift from a low percent of HP food intake
(16%) to a higher percent of HP food intake (79%) in the
restricted rats was significant ( P < 0.001). Because there
was never a difference in food selection between restricted
rats who were shocked and restricted rats who were not
shocked, the shift in food selection following caloric
deprivation and that following refeeding suggested a main
effect of caloric deficit in food selection patterns. In sum-
mary, when rats were in energy deficit they ate proportion-
ately more chow than HP food; when rats were sated, they
ate proportionately more HP food than chow.
2.7. Statistical analyses for all experiments
To examine the effect of R –R and stress, alone and in
combination, separate between groups analyses of variance
(ANOVAs) were conducted using post-stress food intake as
the dependent variable and group (NR + NS, NR + S, R + NS
and R + S) as the independent variable. A significant inter-
action between stress and experience with restriction was
determined if the (R + S) group mean differed significantly
from the means of all other groups using Tukey’s highly
significant difference post-hoc test, P < .05. Repeated meas-
ures ANOVAs were used to examine differences in food
selection ratios (HP:chow) within groups after restriction
and after stress. a level was set at P < .05.
3. General discussion
The purpose of this study was to test the hypothesis that
caloric restriction and stress interact to produce a synergistic
increase in food intake. This hypothesis was guided by
evidence of an important role of dieting and stress in clinical
BEDs including anorexia nervosa, bulimia and BED [1,3 –
9], and by the consistent finding that dieting is the strongest
predictor of stress-induced overeating in the nonclinical
population [2,11– 17]. We report here, for the first time,
that a level of acute stress and a R –R protocol, that alone
have no affect on food intake despite repeated exposure to
each, together interact to produce a potent synergistic
increase in food intake. The hyperphagia occurred in early
post-pubescent rats and in a separate group of young adult
female rats. Two salient factors were observed in the
development of this binge eating model: the need for
sufficient experience with R – R (at least three cycles), and
a pattern of food selection that is consistent with motivation
for reward as opposed to meeting a metabolic need.
First, the hyperphagic response to stress was found to be
contingent upon a minimum of three exposures to R– R,
whether HP food is or is not available from the first cycle
(Experiments 1 and 2). Furthermore, once stress-induced
hyperphagia occurred, it occurred every additional time the
animals were stressed, but R – R was required to precede
stress (Experiment 4). If enough R – R episodes were experi-
enced, the very first exposure to stress was enough to elicit
51
the binge eating effect (Experiment 5). Hence, previous
restriction, despite a lack of energy deficit, was a powerful
and necessary determinant of overeating in response to
stress. In a prior study by Hagan and Moss [21], rats with
a history of similar R – R cycles developed hyperphagia. In
that study, rats were not exposed to environmental stressors,
but were exposed to a much lengthier, 12-week history of
R – R. The presence of environmental stress, therefore, may
serve to expedite the advent of binge-like eating. Prior
exposure to a stressor has been shown to potentiate the stress
response to a second stressor [23,43,44] and, with repeated
stress, food intake and body weight would be expected to
change [45 – 47]. This was not the case. There was no
cumulative effect of restriction experience on food-intake
in rats restricted multiple times but never shocked, nor of
repeated experience with foot shock on food intake. Hence,
the hyperphagia appears to involve a unique interaction
unexplainable by the psychological or physiological con-
sequences of stress alone or restriction alone.
The second prominent observation in the development of
this model was the selective intake of HP food over chow
during the stress after restriction-induced hyperphagia. Rats
cycled through R –R that already demonstrated hyperphagia
in response to stress failed to do so when only chow was
available after stress (Experiment 3). In this way, the present
model of stress-induced overeating differs from tail-pinch-
induced overeating. Had our effect mimicked the nonfeed-
ing specific, goal-directed activation of tail pinch stress [48],
the rats would have overeaten on chow when only chow was
available, but they did not (Experiment 3). Although these
experiments do not definitively rule out the possibility that
experience with more than three R –R cycles might have
produced hyperphagia of chow alone, nor that a contrast
effect from prior exposure to the HP food prohibited chow-
only hyperphagia, there is evidence from prior studies to
suggest that the binge effect may ultimately be contingent
on the presence of HP food. In the 12-week R –R study cited
above, rats with a history of restriction and HP-refeeding
engaged in subsequent binge-like feeding of HP food but
not chow [21]. If contrast effects prohibit overeating of
chow, rats in the 12-week study that were never exposed to
HP food, and hence had nothing to compare chow with,
should have subsequently overeaten chow but this was not
the case. Of course, shock-induced stress, which was used
here but not in the 12-week study, may have altered this
response. Of greater importance is that stress after restric-
tion-induced overeating produces a clear preference for HP
food over chow. This pattern of selection is consistent with
feeding motivated by reward as opposed to metabolic need
or energy-regulation [49 –52]. If ‘hunger’ played a key role
in the stress-induced overeating, restricted rats would have
overeaten chow when only chow was available but they did
not (Experiment 3). Stress after restriction-induced hyper-
phagia of food made more palatable without calories (e.g.
chow with saccharine, or saccharine alone), as well as
determination of central opioid or dopamine involvement
52 M.M. Hagan et al. / Physiology & Behavior 77 (2002) 45–54
in this response will further support an underlying motiva-
tion for reward as has been shown in other food-restriction
models [18,53 –55].
A hypothesis to explain stress after restriction-induced
binge eating is that it is an adaptive response to repetitive
bouts of caloric restriction [56,57] or a homeostatic seques-
tering of surplus energy to cope with the possibility of future
environmental stress amidst reduced food availability. The
most efficient way of recovering calories is via intake of HP
food, which is generally also more caloric. Support for this
hypothesis depends on determining if rats will overeat on
less palatable food (e.g. chow) if they eventually learned
that it was the only food available to them. R –R/S cycles
may increase the incentive value of food, but because these
animals are allowed time to return to normal food intake and
body weight, the increased incentive value is not behavior-
ally expressed until they experience acute stress. Similarly,
rats with a history of deprivation did not overeat until
injected with an opioid agonist at doses without effect in
rats that had never known restriction [18]. Foot shock,
which involves pain, would be expected to release endo-
genous opioids [58], possibly mimicking the exogenous
opioid agonist administration.
The behavioral contingencies tested in this study are only
some of several that will need to be examined to further
define stress after restriction-induced binge eating. For
example, although Experiments 3 and 5 indicated that rats
who previously overate returned to normal levels of food
intake when not restricted just prior to stress, or when not
foot-shocked despite a long history of restriction, additional
rats controlling for number of experiences with restriction
and stress must be concurrently examined to definitively
rule out a potential influence of these accrued manipulations
on the binge response. The effect of other stressors will also
help to further define this model of binge eating. The goal of
this study was to sensitively test for an interaction between
restriction and stress so the focus was not on mode of stress
so much as utilizing levels of stress that alone did not affect
food intake. The assessment of corticosterone levels result-
ing from cyclic R – R/S will help define the specific role of
stress in binge eating.
We have operationally defined the present hyperphagic
response as ‘binge-like’ due to the robustness of the increase
in food intake and its duration within a 24-h period. The
pattern is more similar to the potent and acute hyperphagia
elicited by central peptide YY (PYY) injection [50] than the
‘grazing’ type of overeating over several days elicited by
central injection of agouti-related peptide [59]. More fre-
quent intake readings will determine the exact duration of
the stress after restriction-induced hyperphagia, and it is
very likely that it is considerably shorter than 24 h. The
degree to which the hyperphagic effect is contingent upon
HP food will also be determined by conducting the protocol
without ever introducing HP during refeeding. Rats in
Experiments 3 may have failed to overeat when only chow
was available due to a contrast effect between HP food and
chow. Also of interest will be to assess the effect of the R –
R/S protocol in male rats, especially in light of evidence that
various stressors affect food intake patterns of female and
male rats differentially [38,47,60]. Besides elucidating the
effect of reproductive hormones on the model, it is of
clinical interest to replicate this protocol in male rats since
there is a disproportionately greater incidence of eating
disorders among females [1].
In summary, two important outcomes were obtained from
this study. The first is the finding that caloric-restriction and
acute environmental stress in young female rats interact in a
synergistic manner to produce a powerful increase in food
intake. The second outcome is the development of a useful
animal model of binge eating, one that reflects critical
analogies to human binge eating. In both this animal model
and human BEDs, caloric-restriction and stress are antece-
dents, there is robust overeating despite satiation and normal
energy status, the overeating indicates a high preference for
palatable food that occurs in a discrete period of time and is
consistent with eating for reward, commonly described as
‘addictive-like’ in humans [1,17,26,28 – 34,56,61].
This model will facilitate exploration of the neuroendo-
crine changes that take place during R – R and stress to
develop and maintain binge-like eating. Rational candidates
include alterations in cholecystokinin and central PYY
function, both implicated in bulimia nervosa [62,63]. Regu-
lation of hypothalamic – pituitary – adrenal axis hormones
and of leptin is relevant as their activation affect food intake
under stress [64] and is altered during chronic stress to favor
strong hyperphagia in female rats [65]. As a tool in
understanding the physiological and behavioral contingen-
cies of binge eating, this model should ultimately serve to
guide prevention and treatment strategies that are more
specifically targeted to the large number of BEDs that are
anteceded by dieting and stress.
Acknowledgements
This work was partly funded by a National Eating
Disorders Association (NEDA) Laureate Young Investiga-
tors Award.
References
[1] American Psychiatric Association, editors. Diagnostic and statistical
manual of mental disorders. Washington (DC): APA, 1994.
[2] Crow S, Kendall D, Praus B, Thuras P. Binge eating and other psy-
chopathology in patients with type II diabetes mellitus. Int J Eat Dis-
ord 2001;30:222 – 6.
[3] Mitchell JE, Hatsukami D, Eckert ED, Pyle RL. Characteristics of 275
patients with bulimia. Am J Psychiatry 1985;142:482 – 5.
[4] Polivy J, Herman CP. Dieting and bingeing: a causal analysis. Am
Psychol 1985;40:193 – 201.
[5] Stice E, Agras WS, Telch FC, Halmi KA, Mitchell JE, Wilson T.
Subtyping binge eating-disordered women along dieting and negative
affect dimensions. Int J Eat Disord 2001;30:11 – 27.
 M.M. Hagan et al. / Physiology & Behavior 77 (2002) 45–54
[6] Crowther JH, Sanftner J, Bonifazi DZ, Shepherd KL. The role of daily
hassles in binge eating. Int J Eat Disord 2001;29:449 – 54.
[7] Raffi AR, Rondini M, Grandi S, Fava GA. Life events and prodromal
symptoms in bulimia nervosa. Psychol Med 2000;30:727 – 31.
[8] Vanderlinden J, Vandereycken W, van Dyck R, Vertommen H. Dis-
sociative experiences and trauma in eating disorders. Int J Eat Disord
1993;13:187 – 93.
[9] Wolff GE, Crosby RD, Roberts JA, Wittrock DA. Differences in daily
stress, mood, coping, and eating behavior in binge eating and non-
binge eating college women. Addict Behav 2000;25:205 – 16.
[10] Baucom DH, Aiken PA. Effect of depressed mood on eating among
obese and nonobese dieting and nondieting. J Pers Soc Psychol 1981;
41:577 – 85.
[11] Greeno CG, Wing RR. Stress-induced eating. Psychol Bull 1994;115:
444 – 64.
[12] Herman CP, Polivy J. Anxiety, restraint, and eating behavior. J Ab-
norm Psychol 1975;84:66 – 72.
[13] Macht M. Effects of high- and low-energy meals on hunger, physio-
logical processes and reactions to emotional stress. Appetite 1996;26:
71 – 88.
[14] Oliver G, Wardle J. Perceived effects of stress on food choice. Physiol
Behav 1999;66:511 – 5.
[15] Polivy J, Herman CP. Distress and eating: why do dieters overeat? Int
J Eat Disord 1999;26:153 – 64.
[16] Rutledge T, Linden W. To eat or not to eat: affective and physiological
mechanisms in the stress – eating relationship. J Behav Med 1998;21:
221 – 40.
[17] Wardle J, Steptoe A, Oliver G, Lipsey Z. Stress, dietary restraint and
food intake. J Psychosom Res 2000;48:195 – 202.
[18] Hagan MM, Moss DE. An animal model of bulimia nervosa: opioid
sensitivity to fasting episodes. Pharmacol, Biochem Behav 1991;39:
421 – 2.
[19] Leigh AJ, Stock MJ, Lacey JH, Wilson CA. Diet-induced loss of
cyclic ovarian function at normal body weight in a rodent model for
bulimia nervosa. J Reprod Fertil 1998;112:217 – 23.
[20] Corwin RL, Wojnicki FH, Fisher JO, Dimitriou SG, Rice HB, Young
MA. Limited access to a dietary fat option affects ingestive behav-
ior but not body composition in male rats. Physiol Behav 1998;65:
545 – 53.
[21] Hagan MM, Moss DE. Persistence of binge-eating patterns after a
history of restriction with intermittent bouts of refeeding on palatable
food in rats: implications for bulimia nervosa. Int J Eat Disord 1997;
22:411 – 20.
[22] Sterritt GM. Inhibition and facilitation of eating by electric shock. J
Comp Physiol Psychol 1962;55:226 – 9.
[23] Inoue K, Kiriike N, Okuno M, Fujisaki Y, Kurioka M, Iwasaki S,
Yamagami S. Prefrontal and striatal dopamine metabolism during
enhanced rebound hyperphagia induced by space restriction—a rat
model of binge eating. Biol Psychiatry 1998;44:1329 – 36.
[24] Morley JE, Levine AS, Rowland NE. Minireview: stress induced
eating. Life Sci 1983;32:2169 – 82.
[25] Robbins TW, Fray PJ. Stress-induced eating: fact, fiction or misun-
derstanding? Appetite 1980;1:103 – 33.
[26] Abraham SF, Beumont PJV. How patients describe bulimia or binge
eating. Psychol Med 1982;12:625 – 35.
[27] Hetherington MM, Stoner SA, Andersen AE, Rolls BJ. Effects of
acute food deprivation on eating behavior in eating disorders. Int J
Eat Disord 2000;28:272 – 83.
[28] Hetherington MM, Altemus M, Nelson ML, Bernat AS, Gold PW.
Eating behavior in bulimia nervosa: multiple meal analyses. Am J Clin
Nutr 1994;60:864 – 73.
[29] Hetherington MM, Rolls BJ. Eating behavior in eating disorders:
response to preloads. Physiol Behav 1991;50:101 – 8.
[30] Weltzin TE, Hsu LKG, Pollice C, Kaye WH. Feeding pattern in
bulimia nervosa. Biol Psychiatry 1991;30:1093 – 110.
[31] Oliver G, Wardle J, Gibson EL. Stress and food choice: a laboratory
study. Psychosom Med 2000;62:853 – 65.
53
[32] Epel E, Lapidus R, McEwen B, Brownell K. Stress may add bite to
appetite in women: a laboratory study of stress-induced cortisol and
eating behavior. Psychoneuroendocrinology 2001;26:37 – 49.
[33] Ferber C, Cabanac M. Influence of noise on gustatory affective ratings
and preference for sweet or salt. Appetite 1987;8:229 – 35.
[34] Grunberg NE, Straub RO. The role of gender and taste class in the
effects of stress on eating. Health Psychol 1992;11:97 – 100.
[35] Bertiere MC, Sy TM, Baigts F, Mandenoff A, Apfelbaum M. Stress
and sucrose hyperphagia: role of endogenous opiates. Pharmacol,
Biochem Behav 1984;20:675 – 9.
[36] Dess NK, Choe S, Minor TR. The interaction of diet and stress in rats:
high-energy food and sucrose treatment. J Exp Psychol, Anim Behav
Processes 1998;24:60 – 71.
[37] Marques DM, Fisher AE, Okrutny MS, Rowland NE. Tail pinch
induced fluid ingestion: interaction of taste and deprivation. Physiol
Behav 1979;22:37 – 41.
[38] McIntosh J, Anisman H, Merali Z. Short- and long-periods of neo-
natal maternal separation differentially affect anxiety and feeding in
adult rats: gender-dependent effects. Brain Res Dev Brain Res 1999;
113:97 – 106.
[39] Strongman KT, Coles MGH, Remington RE, Wookey PE. The effect
of shock duration and intensity on the ingestion of food of varying
palatability. Q J Exp Psychol 1970;22:521 – 5.
[40] Anderson SM, Saviolakis GA, Bauman RA, Chu KY, Ghosh S, Kant
GJ. Effects of chronic stress on food acquisition, plasma hormones,
and the estrous cycle of female rats. Physiol Behav 1996;60:325 – 9.
[41] Stam R, van Laar TJ, Akkermans LM, Wiegant VM. Variability fac-
tors in the expression of stress-induced behavioural sensitisation. Be-
hav Brain Res 2002;132:69 – 76.
[42] Takahashi LK, Turner JG, Kalin NH. Development of stress-induced
responses in preweanling rats. Dev Psychobiol 1991;24:341 – 60.
[43] Davis M. Sensitization of the acoustic startle response by foot shock.
Behav Neurosci 1989;103:495 – 503.
[44] Wilson JF, Cantor MB. Noise-induced eating in rats facilitated by
prior tail pinch experience. Physiol Behav 1986;37:523 – 6.
[45] Griffiths J, Shanks N, Anisman H. Strain-specific alterations in con-
sumption of a palatable diet following repeated stress or exposure.
Pharmacol, Biochem Behav 1992;42:219 – 27.
[46] Harris RB, Zhou J, Youngblood BD, Rybkin II, Smagin GN, Ryan
DH. Effect of repeated stress on body weight and body composition of
rats fed low- and high-fat diets. Am J Physiol 1998;275:R1928 – 38.
[47] Zylan KD, Brown SD. Effect of stress and food variety on food intake
in male and female rats. Physiol Behav 1996;59:165 – 9.
[48] Rowland NE, Antelman SM. Stress-induced hyperphagia and obesity
in rats: a possible model for understanding human obesity. Science
1976;191:310 – 2.
[49] Glass MJ, Billington CJ, Levine AS. Naltrexone administered to
central nucleus of amygdala or PVN: neural dissociation of diet
and energy. Am J Physiol, Regul Integr Comp Physiol 2000;279:
R86 – 92.
[50] Hagan MM, Moss DE. Effect of peptide YY (PYY) on food-associ-
ated conflict. Physiol Behav 1995;58:731 – 5.
[51] Hagan MM, Rushing PA, Benoit SC, Woods SC, Seeley RJ. Opioid
receptor involvement in the effect of AgRP-(83 – 132) on food intake
and food selection. Am J Physiol 2001;280:R814 – 21.
[52] Weldon DT, O’Hare E, Cleary J, Billington CJ, Levine AS. Effect of
naloxone on intake of cornstarch, sucrose, and polycose diets in re-
stricted and nonrestricted rats. Am J Physiol 1996;270:R1183 – 8.
[53] Carr KD. Feeding, drug abuse, and the sensitization of reward by
metabolic need. Neurochem Res 1996;21:1455 – 67.
[54] Carroll ME. The role of food deprivation in the maintenance and
reinstatement of cocaine-seeking behavior in rats. Drug Alcohol De-
pend 1985;16:95 – 109.
[55] Carroll ME, Campbell UC, Heideman P. Ketoconazole suppresses
food restriction-induced increases in heroin self-administration in rats:
sex differences. Exp Clin Psychopharmacol 2001;9:307 – 16.
[56] Hagan MM, Whitworth RH, Moss DE. Semistarvation-associated eat-
54 M.M. Hagan et al. / Physiology & Behavior 77 (2002) 45–54
ing behaviors among college binge eaters: a preliminary description
and assessment scale. Behav Med 1999;25:125 – 33.
[57] Laessle RG, Tuschl RJ, Kotthaus BC, Pirke KM. Behavioral and bio-
logical correlates of dietary restraint in normal life. Appetite 1989;12:
83 – 94.
[58] Lewis JW, Cannon JT, Liebeskind JC. Opioid and nonopioid mecha-
nisms of stress analgesia. Science 1980;208:623 – 5.
[59] Hagan MM, Rushing PA, Pritchard LM, Schwartz MW, Strack AM,
Van der Ploeg LHT, Woods SC, Seeley RJ. Long-term orexigenic
effects of AgRP-(83 – 132) involve mechanisms other that melanocor-
tin receptor blockade. Am J Physiol 2000;279:R47 – 52.
[60] Pare WP, Blair GR, Kluczynski J, Tejani-Butt S. Gender differences in
acute and chronic stress in Wistar Kyoto (WKY) rats. Integr Physiol
Behav Sci 1999;34:227 – 41.
[61] Davis C, Claridge G. The eating disorders as addiction: a psycho-
biological perspective. Addict Behav 1998;23:463 – 75.
[62] Geracioti TDJ, Liddle RA. Impaired cholecystokinin secretion in bu-
limia nervosa. N Engl J Med 1988;319:683 – 8.
[63] Kaye WH, Berrettini W, Gwirtsman H, George DT. Altered cerebro-
spinal fluid neuropeptide Y and peptide YY immunoreactivity in ano-
rexia and bulimia nervosa. Arch Gen Psychiatry 1990;47:548 – 56.
[64] Akana SF, Strack AM, Hanson ES, Horsley CJ, Milligan ED, Bhat-
nagar S, Dallman MF. Interactions among chronic cold, corticosterone
and puberty on energy intake and deposition. Stress 1999;3:131 – 46.
[65] Gerozissis K, Panagiotaropoulos T, Papioannou A, Stylianopoulou F.
The dramatic gender-specific shift of corticosterone/leptin ratio fol-
lowing stress is canceled by postnatal handling or testosterone admin-
istration. Soc Neurosci Abstr 2001;31.