International Journal of Obesity (2004) 28, 144–151
& 2004 Nature Publishing Group All rights reserved 0307-0565/04 $25.00
www.nature.com/ijo
PAPER
Effects of manipulated palatability on appetite
depend on restraint and disinhibition scores from
the Three-Factor Eating Questionnaire
MR Yeomans1*, HM Tovey, EM Tinley1 and CJ Haynes1
1
Experimental Psychology, University of Sussex, Brighton, UK
OBJECTIVES: The present study evaluated the effects of dietary restraint on short-term appetite in response to manipulated
palatability.
DESIGN: The effects of palatability on appetite during a lunchtime meal were assessed by contrasting intake of a bland and
palatable version of a simple food (within subject). To test how responses to palatability varied with restraint, these meals
consumed by women were classified according to restraint (R) and disinhibition (D) scores from the Three-Factor Eating
Questionnaire (TFEQ) as high R/high D (HR–HD), high R/low D (HR–LD), low R/high D (LR–HD) and low R/low D (LR–LD).
SUBJECTS: A total of 40 normal-weight women subdivided into four groups based on TFEQ scores.
MEASUREMENTS: The overall intake, appetite and hedonic ratings before, during and after the meal.
RESULTS: All groups ate similar amounts of the bland food, but the LR–HD group ate significantly more of the palatable version
than the other groups, whereas HR–LD did not increase intake in response to palatability. Hunger increased on tasting the
palatable food in all but the HR–LD group, and this group ended both meals more hungry/less full than the others.
CONCLUSION: Women classified as HR–LD were unresponsive to manipulated palatability, whereas those classified as LR–HD
were over-responsive. These findings imply that some individuals are prone to over-respond to palatability and so are at greater
risk of developing obesity, whereas others are able to resist the effects of palatability and so successfully self-restrict their food
intake. Implications for obesity are discussed.
International Journal of Obesity (2004) 28, 144–151. doi:10.1038/sj.ijo.0802483
Published online 21 October 2003
Keywords: palatability; restraint; disinhibition; appetite; overeating
Introduction
The concept of dietary restraint, defined as the cognitively
mediated effort an individual makes to combat the urge to
eat and restrict food intake in order to control body weight,
remains an influential theoretical construct in studies of
human appetite. A key claim of restraint theory is that
women who score high on measures of dietary restraint
develop abnormal eating patterns characterised by periods of
self-restriction combined with periodic overeating.1,2 The
factors that trigger this overeating have been called disinhibitors, based on the idea that these factors break down the
self-imposed inhibition of normal eating. However, although
this classic description of restrained eating remains influen-
*Correspondence: Dr MR Yeomans, Experimental Psychology, University
of Sussex, Brighton BN1 9QG, UK;
E-mail: [email protected]
Received 1 April 2003; revised 8 July 2003; accepted 5 August 2003
tial, a variety of research evidence suggests that this view of
restrained eating is inadequate as a full description of the
effects of attempts to diet on eating behaviour.
While there have been many reports of disinhibited eating
induced by laboratory-based manipulations using the original measure of restraint, the Revised Restraint Scale (RRS3) to
classify participants, studies using two other measures of
restraint (the Dutch Eating Behaviour Questionnaire
(DEBQ)4 and the Three-Factor Eating Questionnaire (TFEQ)5)
have failed to find evidence of disinhibition.6–8 The RRS scale
is known to combine two factors: concern with dieting and
weight fluctuation,9–11 whereas both the TFEQ and DEBQ
appear to measure attempts to restrict eating independent of
weight fluctuation. Thus, women who score high on the RRS
may represent unsuccessful dieters, whereas those scoring
high on the TFEQ and DEBQ combine both successful dieters
and individuals who are likely to diet and periodically
overeat.12,13 It has since been suggested that scores on the
Palatability and restraint
MR Yeomans et al
145
TFEQ disinhibition (D) scale may allow discrimination
between these two types of dieters.14 For example, Westenhoefer et al15 replicated the classic preload study reported by
Herman and Mack,16 with participants categorised according
to both their TFEQ-R and TFEQ-D levels, producing four
subgroups: high R/high D (HR–HD), high R/low D (HR–LD),
low R/high D (LR–HD) and low R/low D (LR–LD). Only the
HR–HD group ate more (counter-regulated) in response to
the preload manipulation, while the other three groups ate
slightly less following the preload. Moreover, the HR–LD
group ate significantly less than the HR–HD group in both
the preload and no-preload conditions, suggesting that HR–
LD represents successful dieters, whereas HR–HD exhibit
eating more characteristic of unsuccessful dieters.
Most laboratory-based studies of restrained eating have
concentrated on the specific effects of situations, which
would be predicted to lead to disinhibited eating. Thus,
studies have shown variously that forced consumption of
preloads that were either high in energy (eg Herman and
Mack16) or were believed to be high in energy17 induced
stress;18 dysphoric mood19 and consumption of alcohol20
can all lead to an increase in food consumption by women
scoring high on the RRS. However, the mere availability of
palatable food may itself act as a stimulus to eat,21 and the
widespread availability of affordable, highly palatable food
in Western society has been discussed as a potential
contributor to the recent increase in obesity.22–24 The
potential for manipulated palatability alone to enhance
short-term food intake in the laboratory is well established in
unrestrained normal-weight individuals,25–28 but we are
unaware of studies testing differential responsivity to
manipulated palatability as a function of dieting history,
and the present study was designed as a preliminary
investigation of this. If manipulated palatability alone is
sufficient to disinhibit eating, then women who can be
classified as unsuccessful dieters (ie HR–LD) may be more
susceptible to overeating in response to manipulated palatability, whereas those who successfully diet may be able to
suppress the normal enhancement of appetite in response to
manipulated palatability. In order to allow us to distinguish
successful and unsuccessful dieters (HR–LD), we adopted the
strategy used successfully by Westenhoefer et al15 of classifying women using both the restraint and disinhibition scales
from the TFEQ, giving four groups of participants: HR–HD,
HR–LD, LR–HD and LR–LD. The prediction was that the HR–
HD group would be most responsive to manipulated
palatability and HR–LD least responsive. We made no
specific predictions differentiating eating between the two
LR groups.
Previously, we have reported consistent effects of altered
palatability in normal-weight unrestrained men, manipulated by modifying the flavour of a standard lunch food,
both on short-term food intake and on changes in rated
appetite within the test meal.27–29 Given the reliability of the
palatability manipulation used in those studies, the same
basic methodology was repeated here. One important
finding from our earlier work was that rated hunger
increased reliably in the early stages of consuming a more
palatable meal, but not a bland meal (the appetiser effect27).
Previously, it has been suggested that restrained individuals
show heightened sensitivity to food cues.30,31 Thus, women
scoring high on the TFEQ restraint scale salivated more when
presented with foods than did women who scored low.
Indeed, mere exposure to palatable food cues can act as a
disinhibitor in restrained women.32 Together, these data
imply that responses to the sensory properties of foods vary
with restraint, and since palatability is known to influence
short-term food intake both in the laboratory and under
naturalistic conditions,33 it is clearly possible that responsivity to palatability may vary with restraint. If so, the HR–
HD group may show a larger differential response to
palatability than the other groups. To test this, we recorded
changes in rated appetite throughout the test meal using the
Sussex Meal Pattern Monitor, an automated system for
recording changes in appetite within a meal developed in
this laboratory.34 This approach allowed us to evaluate fully
differences in response to palatability as a function of R and
D classification.
Method
Design
The overall food intake and differences in rated appetite
during a test lunch were contrasted within participants
between two pasta lunches, one with an unseasoned tomato
sauce (Bland condition) and other with the same sauce with
added seasonings (Palatable condition). To assess the effects
of dieting history, scores on the R and D scales from the
TFEQ were used to classify participants into four groups: HR–
HD, HR–LD, LR–HD and LR–LD, and the effects of the
palatability manipulation were contrasted between these
four groups.
Participants
Participants were 40 female volunteers, recruited from the
staff and students of the University of Sussex. All participants
had previously completed a general recruitment questionnaire that incorporated the TFEQ along with measures of
food and drink preferences and allergies. Women were
categorised as high (H) or low (L) on both TFEQ-R and
TFEQ-D, according to whether they fell above or below the
median on both measures determined from a previous
sample of 150 women from the same population (score of
6 on both restraint and disinhibition). There were 10
Table 1
Age
BMI
TFEQ-R
TFEQ-D
Characteristics (mean7s.e.m.) of the four groups of participants
HR–HD
HR–LD
LR–HD
LR–LD
23.771.3
23.571.0
12.071.0
10.370.8
22.870.9
22.270.7
11.970.8
4.670.3
26.273.1
23.571.0
3.170.4
9.470.6
25.172.1
21.470.7
2.670.6
3.470.5
International Journal of Obesity
Palatability and restraint
MR Yeomans et al
146
participants in each of the four R–D groups (Table 1), and
these groups did not differ significantly in age or body mass
index (BMI), while preselection of participants ensured a
highly significant difference in TFEQ-R score between HR
and LR (F(1,36) ¼ 171.13, Po0.001), and HD and LD
(F(1,36) ¼ 117.00, Po0.001).
Potential participants were informed that the study
explored interactions between mood and appetite, and
anyone self-reporting a previous diagnosis of an eating
disorder, an allergy to any of the ingredients used or diabetes
were excluded. All participants gave their written informed
consent, were paid a small sum on completion of the study,
and were fully debriefed at the end of testing. The study
protocol conformed to British Psychological Society guidelines for ethical human research, and the study protocol was
approved by the University of Sussex Ethics committee.
Foods
Breakfast consisted of a 60 g serving of cereal, 160 g of semiskimmed milk and 200 g of orange juice (1682 kJ total). The
food used for the test meal was the same as that used in
previous studies,35,28 and consisted of pasta shells (Sainsbury’s ‘Conchiglie’), which were prepared by boiling for
10 min in unsalted water, served in either an unseasoned
(Bland) or seasoned (Palatable) tomato and onion sauce
(nutrition/100 g:480 kJ, 4.3 g carbohydrate, 3.3 g fat, 1.2 g
protein for both meals). The flavourings used to season the
Palatable sauce were salt (0.2%), black pepper (0.15%) and
oregano (0.15%). Pasta and sauce were combined in equal
quantities and served at a temperature of 60–651C.
Automated intake and appetite monitor
Subjects were tested individually in laboratory cubicles
equipped with the Sussex Meal Pattern Monitor apparatus
(SMPM: see Yeomans et al,28 Yeomans34 and Gray et al36), our
custom-designed version of the Universal Eating Monitor
(UEM37). Food was weighed automatically by a digital
balance (Sartorius BP 4100), which was built into a bench
top and concealed by a large, white plastic place mat. The
balance was connected through a serial line to an Apple G3
computer, custom-programmed using Future Basic II (Staz
Software) to read the balance weight on stability to an
accuracy of 0.1 g, at 2-s intervals throughout the meal.
Mood and appetite ratings were collected automatically
using the SMPM computer system. Horizontal line scales
were displayed on the monitor, with the question presented
centrally above the line. Each line was end-anchored with
the words ‘Not at all’ in black and ‘Extremely’ in red, with
polarity randomised for each question. A vertical bar
appeared as a pointer in the exact centre of the line, and
could be moved into the desired position by the subject
using a mouse. An on-screen button was clicked to confirm
each response. Mood and appetite ratings were collected at
the start and end of each meal, and consisted of 10 questions
International Journal of Obesity
in the form ‘How {word} do you feel?’, where the word was
‘friendly’, ‘jittery’, ‘hungry’, ‘happy’, ‘thirsty’, ‘clear-headed’,
‘full’, ‘nauseous’, ‘energetic’ or ‘relaxed’, in random order.
The additional ratings of hunger and fullness were made
periodically throughout the meal. Sensory and hedonic
evaluations of each food were obtained after the food was
first tasted with the instruction to ‘Rate the following
property of that food:’ with the word ‘pleasant’, ‘salty’,
‘strong’, ‘sweet’, ‘creamy’ or ‘fruity’ in random order.
Procedure
Participants were instructed in advance to fast (except water)
from 0022 on the evening before each test day. Breakfast was
served in the laboratory between 0830 and 1000 and
participants were instructed to consume nothing but water,
before they returned for lunch. On their return they were
shown into an individual testing cubicle, where they were
served the test lunch. They completed a set of mood and
appetite ratings in the absence of food cues, and then a 500 g
portion of the appropriate pasta dish was served. They were
instructed to take one forkful of the food, complete the
sensory and hedonic ratings and were then instructed to ‘Eat
as much as you like’. The experimenter reminded participants that they would be interrupted ‘from time to time’ by
the computer to complete more appetite ratings, should not
eat while answering questions, and that after several such
interruptions they would be asked to stop eating and
summon the experimenter. This ensured that participants
were unaware that the computer was programmed to
interrupt participants at 50 g intervals, during which additional ratings of hunger, fullness and pleasantness were
made. After the sixth interruption, participants were instructed to call their experimenter and the food bowl was
refilled to approximately its original level to ensure that
participants were not faced with an empty bowl at any time
during the experiment. This process of eating bouts,
interruptions with appetite ratings and food refills, continued until participants indicated that they had finished, after
which they repeated the mood and appetite ratings. After an
interval of at least 1 day, the second test day followed the
same procedure, but with the other food condition. The
order of testing of the two foods was counterbalanced within
each of the four groups. Weight, height and age of all
participants were recorded at the end of testing, which
included debriefing and payment of a small sum for their
participation.
Data analysis
Three-way ANOVA, with restraint and disinhibition classification (between subjects) and meal type (within-subjects) as
factors, was used to analyse overall intake, with the specific
prediction that the HR–HD group would show the largest
response to the palatability manipulation. The prediction
that the HR–HD group may experience a greater enhance-
Palatability and restraint
MR Yeomans et al
147
Appetite ratings at the start and end of eating
Ratings of hunger before food was presented, when the food
was first tasted and at the end of the meal are presented in
Food intake (g)
400
200
0
HR-HD
HR-LD
LR-HD
LR-LD
Group
Figure 1 Overall food intake in the Palatable () and Bland (&) food
conditions for the four subject groups.
a 500
High restraint, high
disinhibition
Rated hunger
400
b 500
300
200
400
300
200
0
0
Start Taste
Time
c 500
High restraint, low
disinhibition
100
100
Start Taste
Time
End
Low restraint, high
disinhibition
400
d 500
Rated hunger
Overall food intake
As predicted, the overall food intake was greater in the
Palatable than Bland condition (F(1,36) ¼ 22.44, Po0.001).
However, this effect depended both on restraint (restraint food interaction, F(1,36) ¼ 5.07, Po0.05) and disinhibition (disinhibition food interaction, F(1,36) ¼ 6.14,
Po0.025). The overall prediction was that the HR–HD group
would show enhanced response to the palatability manipulation. However, as can be seen in Figure 1, the group
showing the greatest response to palatability was the LR–HD
group, who ate 153 g more of the Palatable than Bland food,
significantly (Po0.05 or greater) more than any of the other
three groups. LR–LD and HR–HD showed very similar
responses to manipulated palatability, with both groups
showing a small but significant increase in intake (both
Po0.05). Intake in the HR–LD group, in contrast, did not
differ significantly between conditions. Overall, HD was
associated with a tendency to increase intake in response to
palatability, whereas HR was associated with a tendency to
show reduced sensitivity to palatability. The combination of
these two effects made the LR–HD group more responsive to
palatability than any other group.
600
Rated nhunger
Results
800
Rated hunger
ment of subjective appetite in response to palatability was
tested in two ways. Firstly, appetite (hunger and fullness) was
rated at the start of each meal, after first tasting the food, and
at the end of eating, were contrasted between conditions
using a similar ANOVA, but with time of rating as an
additional factor. Secondly, we fitted quadratic functions
relating ratings of hunger and fullness against actual intake
within each meal for each participant and then contrasted
the best-fit coefficients from these quadratic functions
between meals and groups using three-way ANOVA. We
have used the same technique in many previous studies (see
Yeomans et al28 and Yeomans34), and have found the linear
coefficient from these quadratic functions to be sensitive to
changes in palatability. Here, the specific prediction was that
the HR–HD group would exhibit a larger difference in linear
coefficient between the two meals than would the other
groups.
The success of the flavour manipulation depended firstly
on higher pleasantness ratings for the Palatable than Bland
food, and secondly on the four groups showing similar
hedonic evaluations of the two foods. To test this, we
contrasted pleasantness ratings for the two foods made at the
start and end of both meals between meal types and groups
using four-way ANOVA, with time of rating and meal type
within subject, and restraint and disinhibition between
subjects.
300
200
End
Low restraint, low
disinhibition
400
300
200
100
100
0
0
Start Taste
Time
End
Start Taste
Time
End
Figure 2 Rated hunger at the start, on tasting the food and at the end of the
Palatable (KFK) and Bland (J?J) meals for the four subject groups.
Figure 2. As expected, there was an overall effect of time
(F(2,72) ¼ 362.94, Po0.001), with hunger high at the start
and low at the end of the meal. However, there were also
significant food time (F(2,72) ¼ 3.60, Po0.001) and timeInternational Journal of Obesity
Palatability and restraint
MR Yeomans et al
148
High restraint, high
disinhibition
400
b 500
Rated fullness
Rated fullness
a 500
300
200
400
300
200
0
0
Start Taste End
Time
Start Taste End
Time
Low restraint, high
disinhibition
c 500
Rated fullness
400
d 500
300
200
Low restraint, low
disinhibition
400
300
200
100
100
0
0
Changes in hunger and fullness within a meal
In line with our previous work (see Yeomans34), quadratic
functions were an adequate description of the pattern of
change in hunger and fullness, accounting for on average
84.5% of the variance in hunger and 86.7% of variance in
fullness ratings. We therefore contrasted the coefficients
from the individual best-fit quadratic functions between
conditions (see Table 2).
The averaged best-fit functions for hunger, along with
representative data at intervals throughout each meal, are
presented in Figure 4. Contrasts of the best-fit coefficients
found a significantly higher linear coefficient with the
Palatable than Bland food, consistent with the observation
High restraint, low
disinhibition
100
100
Rated fullness
restraint disinhibition (F(2,72) ¼ 3.34, Po0.05) interactions. Hunger did not differ between groups or conditions
prior to the food being presented, but hunger was significantly higher after tasting the Palatable than Bland food
(F(1,36) ¼ 4.70, Po0.05), regardless of R or D scores. Hunger
ratings at the end of the meal did not differ between foods,
but did vary depending on restraint and disinhibition
classification
(restraint disinhibition
interaction,
F(1,36) ¼ 6.31, Po0.05). The HR–LD group ended both meals
feeling significantly more hungry than the HR–HD or LR–LD
groups, with the LR–HD group intermediate.
As expected, fullness was low at the start and high at the
end of the meal (main effect of time, F(2,72) ¼ 530.76,
Po0.001); however, there was also a significant time restraint disinhibition interaction (F(2,72) ¼ 3.52, Po0.05).
There were no significant effects of food, unlike hunger
ratings (Figure 3), and there were no significant differences
in fullness ratings between groups either at the start of the
meal or on tasting the food. However, the HR–LD group
ended their meals with significantly lower fullness ratings
than the other three groups (Po0.005 or greater).
Start Taste End
Time
Start Taste End
Time
Figure 3 Rated fullness at the start, on tasting the food and at the end of the
Palatable (KFK) and Bland (J?J) meals for the four subject groups.
that hunger tended to increase during the early stages of the
meal with the Palatable food only. No other significant
effects were found from these analyses, although Figure 3
suggests that whereas the best-fit curves for Palatable and
Bland conditions showed clear separation in the HR–HD, LR–
HD and LR–LD groups, this distinction was not evident in
the HR–LD group. No significant effects of food or restraint
classification were seen on best-fit functions for fullness.
Table 2 Components of the best-fit quadratic functions relating hunger and fullness to intake within the two test meals in the four restraint groups. All data are
mean7s.e.m.
Group
Component
Hunger ratings
Y-intercept
Linear coefficient
Quadratic coefficient
Fullness ratings
Y-intercept
Linear coefficient
Quadratic coefficient
International Journal of Obesity
Food condition
HR–HD
HR–LD
LR–HD
LR–LD
Bland
Palatable
Bland
Palatable
Bland
Palatable
392.2720.3
408.1721.7
0.20570.230
0.18170.159
0.001770.0007
0.001870.0005
385.7723.7
373.8721.3
0.06470.076
0.04170.101
0.002270.0006
0.002170.0003
372.1742.7
394.3731.8
0.18970.321
0.18970.124
0.001670.0010
0.001670.0003
371.9714.4
389.7711.6
0.04670.147
0.18570.084
0.002270.0006
0.002170.0004
Bland
Palatable
Bland
Palatable
Bland
Palatable
60.5719.4
73.1715.4
0.47170.157
0.36670.125
0.000870.0002
0.000970.0004
81.3721.6
80.2718.3
0.37770.134
0.29070.171
0.000970.0003
0.000970.0005
69.2720.6
64.7723.5
0.53670.254
0.35770.185
0.000870.0006
0.000570.0003
79.1718.4
57.2713.8
0.36970.164
0.41070.168
0.001070.0004
0.000870.0003
Palatability and restraint
MR Yeomans et al
149
High restraint, high
disinhibition
400
b 500
Rated hunger
Rated hunger
a 500
300
200
High restraint, low
disinhibition
400
300
200
100
100
0
0
0
200
400
600
0
Low restraint, high
disinhibition
400
600
400
300
200
Low restraint, low
disinhibition
d 500
Rated hunger
Rated hunger
c 500
200
Intake (g)
Intake (g)
400
300
200
100
100
0
0
0
200
400
Intake (g)
600
0
200 400
Intake (g)
600
Figure 4 Best-fit quadratic functions relating rated hunger against both food
consumed for the Palatable (FF) and Bland ( . . . . . . . . . . . . ) lunches,
along with mean ratings (7s.e.m.) at the start and end of the meal,
immediately after tasting the food and at the points during the meal at which
25, 50 and 75% of lunch had been consumed (Palatable (K): Bland (J)).
Pleasantness ratings for the test lunches
As expected, overall pleasantness of the Palatable lunch
(rating ¼ 321.0710.2 on 500 pt scale) was significantly
greater (main effect of food, F(1,36) ¼ 11.42, Po0.005) than
that of the Bland lunch (273.5715.9 on 500 pt scale), and
pleasantness was much lower at the end of the meal than at
the start regardless of food consumed (main effect of time,
F(1,36) ¼ 58.88, Po0.001). There were no effects of restraint
or disinhibition classification on pleasantness (no main
effect or interaction involving R or D). These data suggest
that the differences in food intake and rated appetite
described earlier were not due to differences in the evaluation of the two lunches, but reflected different responses to
the hedonic distinction between foods.
Discussion
The present study found no evidence of any difference in
response to manipulated palatability for the HR–HD (‘unsuccessful dieter’) group relative to the LR–LD group.
However, responses to the palatability manipulation of the
HR–LD (‘successful dieter’) and LR–HD groups differed from
the HR–HD and LR–LD groups, and suggested firstly that the
HR–LD group was relatively insensitive to the effects of the
palatability manipulation, and in contrast that the LR–HD
group over-responded to palatability.
Two findings dissociated the eating behaviour of the HR–
LD group from the other groups. Firstly, HR–LD showed little
response to the palatability manipulation, eating the same in
both conditions and showing no evidence of enhanced
appetite once the Palatable food had been tasted. This lack of
enhanced appetite was seen despite no evidence that this
group found the food any less Palatable than the other
groups. Thus, these data implies that this group of women
have either learned to suppress their normal response to
palatable foods, or that their appetite control system is in
some way different from the other groups. Lack of response
to palatability by the HR–LD group here is in accord with a
failure to find disinhibited eating in response to a preload by
the same group,15 and by their failure to eat more in a multiitem meal than in a single-item meal.38 Whether this lack of
response to these various manipulations is the consequence
of a specific personality trait, a difference in their appetite
control system or some other mechanism requires substantiation, since this knowledge could lead to methods for
helping other groups who need to self-restrict their eating as
a weight-loss measure to do so successfully. The second
feature that distinguished the HR–LD group was their
tendency to end their meal with higher hunger and lower
fullness ratings than the other groups, irrespective of food
eaten. The obvious explanation is that this group has learned
to tolerate feelings of hunger, an idea that fits well both with
the original concept of restraint (eg 1) and the characterisation of this group as successful dieters.14 However, equivalent patterns of appetite were not evident in the same group
when eating a multi-item meal,38 and the present findings
therefore warrant replication.
Although we did not specifically predict the enhanced
response to palatability seen in the LR–HD group, this
finding shows consistency between the characterisation of
this group by the TFEQ-D scale as self-reporting a tendency
to overeat and their behaviour in the laboratory. Since
patients classified as suffering from binge-eating disorder
also score high on the TFEQ-D scale and low on the TFEQ-R
scale,39-41 the present findings are consistent with the idea
that LR–HD are hypersensitive to sensory stimulation of
appetite. Enhanced sensitivity to palatability coupled with
low restraint implies that if these individuals failed to
compensate for the enhanced intake of palatable food at
one meal by reducing subsequent energy intake, they would
be liable to weight gain and eventually obesity. Thus, unless
a restrained attitude is adopted in the future, women with
the LR–HD characteristics may be prone to dietary-induced
obesity. It is not possible to identify the mechanism underlying the heightened sensitivity to palatability in the LR–HD
group from the present study, but the study does demonstrate a method for future evaluation of this mechanism.
The lack of difference between LR–LD and HR–HD groups
could reflect the same appetite control processes in operaInternational Journal of Obesity
Palatability and restraint
MR Yeomans et al
150
tion in both groups. An alternative explanation, supported
in part by the analysis of intake data from the present study,
is that intake by the HR–HD group represents a combination
of a heightened response to palatability (as in the LR–HD
group) that has been moderated by attempts at restraint (as
in the HR–LD group). If the over-responsivity to palatability
that was seen in the LR–HD group reflected some form of
trait that was common to all who score high on the TFEQ-D
measure, this would imply that the dietary restraint
exhibited by the HR–HD group may be partially successful,
and again this warrants further investigation. One argument
against the idea that the HR–HD group is at least partially
successful dieters was the present finding that HR–HD and
HR–LD groups could be dissociated by the level of satiety
exhibited at the end of the meal, with lower satiety in the
HR–LD group. Thus, HR–LD appeared to terminate eating at
a lower level of satiety (ie cognitively restricting food intake,
true restraint), whereas HR–HD ended their meal as satiated
as LR participants, and did not appear to be restricting their
intake.
Although the present study found no differentiation
between HR–HD and LR–LD groups, this was in the context
of a laboratory-based study with a food that would not have
been classified as threatening to dieting. The present food is
not one that would be rated as forbidden,42 and it may be
that this both restricted the level of response of the HR–HD
group to the palatability manipulation, and allowed the HR–
LD group to suppress any response to palatability. It would
therefore be worthwhile repeating the present study with
foods that would be perceived as more threatening to
restraint. Indeed, mere exposure to forbidden foods may
act as a disinhibitor by itself,43 and since foods rated as
forbidden are also rated as highly palatable (unpublished
data), the distinction between HR–LD and HR–HD seen here
may be less evident if more threatening foods were used.
The present study relied on median-split of a large
database of women to determine high and low populations
of restrained and unrestrained women. The median-split
approach has been used widely in restraint research,16,17,42,44
but has the disadvantage of putting a relatively arbitrary
distinction between populations. Despite this potential
shortcoming, distinct differences were seen between the
four groups tested here; however, future studies might want
to use more stringent measures to evaluate these effects
further. The size of the sample in the current study, while
again in line with the general appetite and restraint
literatures, was still relatively small, with only 10 participants in each group. Since this was designed as a first test of
the palatability/restraint/disinhibition interaction, followup studies should look to broaden and expand the sample
tested.
In conclusion, the present study adds weight to the
argument that restraint is not a homogenous construct (eg
Westenhoefer14), and is the first study to report differentiation of responses to palatability as a function of classification
according to restraint and self-reported tendency to disinInternational Journal of Obesity
hibit in a nonclinical group. Future studies are needed using
more ‘threatening’ foods, and assessing the same relationships in clinically defined groups.
Acknowledgements
This research was partly funded by an undergraduate bursary
from the Nuffield Foundation to Heather Tovey.
References
1 Herman CP, Polivy J. A boundary model for the regulation of
eating. In: Stunkard AJ, Stellar E (eds.) Eating and its disorders.
Raven Press: New York; 1984. pp 141–156.
2 Ruderman. Dietary restraint: a theoretical and empirical review.
Psychol Bull 1986; 99: 247–262.
3 Herman CP, Polivy J. Restrained eating. In: Stunkard AJ (ed.)
Obesity. Saunders: Philadelphia; 1980. pp 208–225.
4 Van Strein T, Frijters JE, Bergers GPA, Defares PB. Dutch Eating
Behaviour Questionnaire for assessment of restrained emotional
and external eating behaviour. Int J Eating Disord 1986; 5:
295–315.
5 Stunkard AJ, Messick S. The three-factor eating questionnaire to
measure dietary restraint, disinhibition and hunger. J Psychosom
Res 1985; 29: 71–83.
6 Wardle J, Beales S. Restraint and food intake: an experimental
study of eating patterns in the laboratory and in normal life.
Behav Res Ther 1987; 25: 179–185.
7 Dritschel B, Cooper PJ, Charnock D. A problematic counterregulation experiment: implications for the link between dietary
restraint and overeating. Int J Eating Disord 1993; 13: 297–304.
8 Morgan PJ, Jeffrey DB. Restraint, weight suppression and selfreport reliability: how much do you really weigh? Addict Behav
1999; 24: 679–682.
9 Blanchard FA, Frost RO. Two factors of restraint: concern for
dieting and weight fluctuation. Behav Res Ther 1983; 21: 259–267.
10 Lowe MR. Dietary concern, weight fluctuation and weight status:
further explorations of the restraint scale. Behav Res Ther 1984;
22: 243–248.
11 Heatherton TF, Herman CP, Polivy J, King GA, McGree ST. The
(mis)measurement of restraint: an analysis of conceptual and
psychometric issues. J Abnorm Psychol 1988; 100: 78–83.
12 Lowe MR, Maycock B. Restraint, disinhibition, hunger and
negative affect eating. Addict Behav 1988; 13: 369–377.
13 Tuschl RJ. From dietary restraint to binge eating: some theoretical
considerations. Appetite 1990; 14: 105–109.
14 Westenhoefer J. Dietary restraint and disinhibition: is restraint a
homogenous construct. Appetite 1991; 16: 45–55.
15 Westenhoefer J, Broeckmann P, Munch A-K, Pudel V. Cognitive
control of eating behaviour and the disinhibition effect. Appetite
1994; 23: 27–41.
16 Herman CP, Mack D. Restrained and unrestrained eating. J
Personality 1975; 43: 647–660.
17 Spencer JA, Fremouw WJ. Binge eating as a function of restraint
and weight classification. J Abnorm Psychol 1979; 88: 262–267.
18 Wardle J, Steptoe A, Oliver G, Lipsey Z. Stress, dietary restraint
and food intake. J Psychosom Res 2000; 48: 195–202.
19 Ruderman AJ. Dysphoric mood and overeating: a taste of restraint
theory’s disinhibition hypothesis. J Abnorm Psychol 1985; 94:
78–85.
20 Polivy J, Herman CP. The effects of alcohol on eating behavior:
disinhibition or sedation? Addict Behav 1976; 1: 121–125.
21 Yeomans MR. Taste, palatability and the control of appetite. Proc
Nutr Soc 1998; 57: 609–615.
22 Maffeis C. Aetiology of overweight and obesity in children and
adolescents. Eur J Pediatr 2000; 159: S35–S44.
Palatability and restraint
MR Yeomans et al
151
23 McCrory MA, Suen VMM, Roberts SB. Biobehavioral influences
on energy intake and adult weight gain. J Nutr 2002; 132:
3830S–3834S.
24 Walker ARP, Walker BF, Adam F. Nutrition, diet, physical activity,
smoking and longevity: from primitive hunter–gatherer to
present passive consumerFhow far can we go? Nutrition 2003;
19: 169–173.
25 Bellisle F, Lucas F, Amrani R, Le Magnen J. Deprivation,
palatability and the micro-structure of meals in human subjects.
Appetite 1984; 5: 85–94.
26 Bobroff EM, Kissileff H. Effects of changes in palatability on food
intake and the cumulative food intake curve of man. Appetite
1986; 7: 85–96.
27 Yeomans MR. Palatability and the microstructure of eating in
humans: the appetiser effect. Appetite 1996; 27: 119–133.
28 Yeomans MR, Lee MD, Gray RW, French SJ. Effects of test-meal
palatability on compensatory eating following disguised fat and
carbohydrate preloads. Int J Obes Relat Metab Disord 2001; 25:
1215–1224.
29 Yeomans MR, Gray RW, Mitchell CJ, True S. Independent
effects of palatability and within-meal pauses on intake
and subjective appetite in human volunteers. Appetite 1997; 29:
61–76.
30 Tepper BJ. Dietary restraint and responsiveness to sensory-based
food cues as measured by cephalic phase salivation and sensory
specific satiety. Physiol Behav 1992; 52: 305–311.
31 Teff KL, Engelman K. Palatability and dietary restraint: effect on
cephalic phase insulin release in women. Physiol Behav 1996; 60:
567–573.
32 Rogers PJ, Hill AJ. Breakdown of dietary restraint following mere
exposure to food stimuli: interrelationships between restraint,
hunger, salivation, and food intake. Addict Behav 1989; 14:
387–397.
33 De Castro JM, Bellisle F, Dalix A-M, Pearcey SM. Palatability and
intake relationships in free-living humans: characterization and
34
35
36
37
38
39
40
41
42
43
44
independence of influence in North Americans. Physiol Behav
2000; 70: 343–350.
Yeomans MR. Rating changes over the course of meals: what do
they tell us about motivation to eat? Neurosci Biobehav Rev 2000;
24: 249–259.
Yeomans MR, Lartamo S, Procter EL, Lee MD, Gray RW. The
actual, but not labelled, fat content of a soup preload alters shortterm appetite in healthy men. Physiol Behav 2001; 73: 533–540.
Gray RW, French SJ, Robinson TM, Yeomans MR. Dissociation of
the effects of preload volume and energy content on subjective
appetite and food intake. Physiol Behav 2002; 76: 57–64.
Kissileff HR, Kilngsberg G, Van Italie TB. Universal eating monitor
for continuous recording of solid or liquid consumption in man.
Am J Physiol 1980; 238: R14–R22.
Haynes CL. The interactive effects of dietary restraint and
disinhibition on ingestive behaviour. University of Sussex:
Brighton; 2001. 411 pp.
Yanovski SZ, Leet M, Yanovski JA, Flood M, Gold PW, Kissileff HR,
Walsh BT. Food selection and intake of obese women with bingeeating disorder. Am J Clin Nutr 1992; 56: 975–980.
Peterson CB, Mitchell JE, Engbloom S, Nugent S, Mussell MP,
Crow SJ, Miller JP. Binge eating disorder with and without a
history of purging symptoms. Int J Eating Disord 1998; 24:
251–257.
Howard CE, Porzelius LK. The role of dieting in binge eating
disorder: etiology and treatment implications. Clin Psychol Rev
1999; 19: 35–44.
Knight LJ, Boland FJ. Restrained eating: an experimental disentanglement of the disinhibiting variables of perceived calories
and food type. J Abnorm Psychol 1989; 98: 412–420.
Stirling LJ, Yeomans MR. Effects of exposure to a forbidden food
on eating in restrained, unrestrained women. Int J Eating Dis, (in
press).
Steere J, Cooper PJ. The effects on eating of dietary restraint,
anxiety, hunger. Int J Eating Disord 1993; 13: 211–219.
International Journal of Obesity