Acute compensatory eating following exercise is associated with implicit hedonic wanting for
food
5
1
Finlayson G, 1Bryant E, 1Blundell JE & 2King NA.
1
BioPsychology Group, Institute of Psychological Sciences, University of Leeds, Leeds,
LS29JT, UK; 2Institute of Health and Biomedical Innovation, Human Movement Studies,
10
Queensland University of Technology, Brisbane. Australia
Correspondence to:
Graham Finlayson
15
Biopsychology group
Institute of Psychological Sciences
University of Leeds
Leeds LS2 9JT
United Kingdom
20
Tel: 44 113 343 7601
Fax: 44 113 343 6674
Authors’ email addresses:
g.s.finlayson@leeds.ac.uk
25
e.j.bryant@leeds.ac.uk
2
j.e.blundell@leeds.ac.uk
n.king@qut.edu.au
5
Running head: Acute compensatory eating following exercise
3
Abstract
The efficacy of exercise to promote weight loss could potentially be undermined by its
influence on explicit or implicit processes of liking and wanting for food which in turn alter
food preference. The present study was designed to examine hedonic and homeostatic
5
mechanisms in the acute effects of exercise on food intake. 24 healthy female subjects were
recruited to take part in two counterbalanced activity sessions; 50 minutes of high intensity
(70% max heart rate) exercise (Ex) or no exercise (NEx). Subjective appetite sensations,
explicit and implicit hedonic processes, food preference and energy intake (EI) were
measured immediately before and after each activity session and an ad libitum test meal. Two
10
groups of subjects were identified in which exercise exerted different effects on compensatory
EI and food preference. After exercise, compensators (C) increased their EI, rated the food to
be more palatable, and demonstrated increased implicit wanting. Compensators also showed a
preference for high-fat sweet food compared with non-compensators (NC), independent of the
exercise intervention. Exercise-induced changes in the hedonic response to food could play an
15
important role in weight control. The exercise-induced increase in implicit wanting for food
may help to explain the variable efficacy of exercise as a method of weight loss. Some
individuals could be resistant to the beneficial effects of exercise due to a predisposition to
compensate for exercise-induced energy expenditure as a result of changes in food
preferences.
20
4
Introduction
Physical activity is currently strongly promoted as a method of weight control.
However, in some individuals, a compensatory increase in energy intake (EI) might serve as a
barrier for the potential for exercise to promote a negative energy balance (1). Therefore, the
5
efficacy of exercise as a successful method of weight management will vary between
individuals (2). EI is commonly put forward as the compensatory mechanism responsible for
a lack of, or lower than expected, exercise-induced weight loss. Therefore, it is important to
identify the homeostatic and hedonic processes which could be mobilised by exercise to
influence food intake. For example, exercise may be linked to food intake through changes in
10
food preferences and macronutrient selection (3;4). This relationship seems to depend on the
characteristics of the population studied (5;6;7) and may reflect a biological drive for foods
with particular sensory or nutrient properties (8;9;10).
Exercise and food intake may also be linked by their rewarding potential. Evidence indicates
that vigorous exercise can stimulate brain substrates which are associated with reward and
15
dependence (11;12). More recently, ‘Sensitivity to Reward’ – thought to reflect an innate
characterological trait – has been identified (13;14) and is suggestive of a common neural
substrate for determining the impact of pleasurable activities including exercise and feeding
(15). It has been hypothesised that exercise may act as a buffer for reward driven eating (15),
however it is also possible that exercise has a sensitising action that enhances food reward.
20
Differences in the impact of exercise on food reward may explain some inter-individual
variability in compensatory eating after exercise.
Of course, deliberate inappropriate food choices and allowance of food ‘treats’ could also be
responsible for increases in EI (2). Nevertheless, the influence of hedonic processes on the
behavioural action of eating can be considered as two-fold: Strength of motivational response
25
to obtain available food, and the degree of subjective pleasure induced. Recently, the terms
5
liking and wanting have been used to distinguish such hedonic processes that are either
explicitly affective or implicitly motivational (16;17) and these may represent dual
components of food preference (18;19;20). Therefore, it can be hypothesised that hedonic
processes – via their influence on food preference – could be modulated by exercise-induced
5
energy expenditure to promote overconsumption in those prone to compensatory eating.
A key issue is whether changes in exercise-induced food preference and nutrient selection are
associated with hedonic or homeostatic mechanisms. Recent considerations of homeostatic
and non-homeostatic determinants of eating imply that such systems interact in the overall
expression of appetite (21;22;23). Since most of the evidence from acute exercise
10
interventions indicates that there is no automatic increase in hunger following exercise
(24;25;2), other non-homeostatic mechanisms need to be explored. Traditionally, methods
such as test meals providing a selection of foods varying in nutrient composition (26),
palatability visual analogue scales (27), and titration of the ratio of sweetness to fat in drinks
(28) can be used to assess the acute effects of exercise on food preference. However, these
15
methods are not designed – and therefore may not be sensitive enough – to detect more subtle
exercise-induced alterations in the hedonic processes that influence food preference.
Therefore, this study used a novel computer-based procedure to allow the separate
assessments of explicit and implicit operations of liking and wanting for an array of food
items (19) to determine the acute effects of exercise on food preference, and explore
20
individual variability in the compensatory response to exercise.
6
Method
Participants
Participants were recruited from the staff and student population of the University of Leeds.
5
Screening sessions involved completion of the Three Factor Eating Questionnaire (29),
acceptance of the study foods and level of habitual exercise (mean = 2.4 SD = 1.2
engagements/week) were verified by self report, and body composition was measured using
bioelectrical impedance analysis (Inbody, Biospace, California). 24 healthy females aged 1840 years (mean = 24.0, SD = 6.1 yr) and body mass index (BMI) between 20-25 kg/m2 (mean
10
= 22.3, SD = 2.9 kg/m2) were selected from the initial screening process to exclude those who
were taking medication, smokers, those currently following a weight loss routine, or reported
a history of eating or psychological disorders. Following criteria from previous studies
(Brunstrom et al. 2004; Tepper, 1992), all were classed as non-dietary restrained indicated by
a score of less than 11 for Restraint on the TFEQ (mean = 6.1, SD = 3.1). Two cases were
15
excluded from the analyses due to incomplete data, leaving a final sample of N=22.
Design
All the manipulations and measures were administered in the Human Appetite Research Unit
at University of Leeds. The study conformed to a within-subjects design consisting of two
20
counterbalanced conditions separated by approximately one week. The two conditions were
moderate-intensity exercise (Ex) which involved stationery cycling at approximately 70%
maximum heart rate for 50 minutes, and 50 minutes of no exercise (NEx). The NEx condition
involved being sedentary (e.g. reading, sitting quietly). The exercise sessions were conducted
under supervision. Physical activity levels prior to testing were standardised by asking
25
participants to refrain from exercise on the day before each condition. Prior to their
7
involvement in the study all participants were informed about the procedures and gave their
written consent. On the final test session, they received written and verbal debriefing and
were invited to ask questions. All procedures were followed in accordance with the standards
of the institutional ethics committee.
5
Measures
Subjective appetite sensations and hedonic evaluation of the test meal
Subjective appetite sensations were recorded throughout the test days using 100mm visual
analogue scales (VAS), anchored at each end with the statements “not at all” and “extremely”.
10
Appetite sensations were assessed by questions relating to hunger, thirst, fullness and desire
to eat. These were assessed immediately before and after the Ex and NEx sessions, and
following the test meal. VAS hedonic ratings of the test meal were recorded immediately
following consumption according to the pleasantness of the food.
15
Liking and wanting computer procedure
The computer procedure comprised two tasks designed to assess 1) explicit liking, 2) implicit
wanting, and 3) relative preference for the same visual food stimuli. This experimental
procedure has been described in more detail previously (19) and follows other laboratories
successful use of reaction time as an indicator of implicit processes (for example, the Implicit
20
Association Test; 30). The procedure uses a ‘forced choice’ reaction time measure of implicit
wanting in addition to explicit subjective measures of liking for visual food stimuli varying in
fat content and taste. Implicit wanting is operationalised as the reaction time of each pair trial
decision. Thus, the speed with which one category of stimuli is chosen relative to alternative
categories provides a quantifiable measure of implicit wanting for each food category in the
25
procedure. The visual food stimuli were selected from a database of photographs and sorted
8
according to their fat content and taste properties into one of four separate categories: high fat
savoury (HFSA); low fat savoury (LFSA); high fat sweet (HFSW); and low fat sweet
(LFSW). Each category was represented by five different foods; hence a total of 20 different
food stimuli were presented in the procedure (see table 1). Participants completed the task
5
immediately before Ex or NEx, immediately after Ex or NEx, and immediately after
consuming the test meal. Stimuli were presented using experiment generator software (EPrime v1.1.4) via 17” monitor and measured 150×100 mm2.
Table 1 about here
10
Explicit liking trials
The explicit task recorded subjective hedonic ratings for each food stimulus using VAS. The
trials consisted of the twenty food stimuli presented one at a time and rated according to a
100-mm VAS anchored at each end by the statements “not at all” and “extremely”. Subjects
15
were prompted with the statement “How pleasant would it be to taste some of this food
now?”. The VAS was presented on-screen beneath each food stimulus and subjects used the
mouse to move a centred cursor along the line to indicate their response. When a rating was
made, the procedure automatically cycled to the next stimulus trial. Mean ratings for each
food category (HFSA, LFSA, etc.) were automatically computed.
20
Implicit wanting trials and relative preference
Implicit wanting and relative preference were measured by a behavioural ‘forced choice’
methodology. In this task, a food stimulus from one of the four food categories was paired
with one stimulus from the remaining categories to form a series of 150 trials in which the
25
subjects were given the standardised instruction to select the food they “most want to eat
9
now”. In addition to measuring relative preference for the stimuli by recording the frequency
of selections made in each category (with a possible range of 0-75), reaction time (in ms) of
each choice was also captured. By covertly recording reaction time, subjects remained
unaware of implicit changes in their behaviour on the task, while remaining free to determine
5
the direction of their choices. Data from the forced choice task – including relative preference
(freq. of choice) and implicit wanting (reaction time in ms) – were recorded online for later
calculation of the means.
Test meal
10
The ad libitum test meal consisted of cheese salad sandwiches, ready-salted crisps and fruit
yoghurt. The average percentage energy provided by protein, fat and carbohydrate in the test
meal was 13%, 45% and 42% respectively. Energy intake was calculated by weighing the
food before and after consumption (to the nearest 0.1g) and with reference to the
manufacturers energy values and food tables (30).
15
Heart rate
Prior to the experimental sessions, subjects were fitted with a heart rate monitor (Polar
Electronic-4000, Finland). This was to ensure that exercise was performed at a fixed intensity
(~70% maximum heart rate) and to estimate net energy expenditure from the exercise bout.
20
Maximum heart rate was assessed on a screening day prior to test days, using an incremental
test on a cycling ergometer (Proergo, Tunturi Ltd, Japan). Participants were required to cycle
until volitional exhaustion, at which point maximum heart rate was inferred. Mean heart rate
from the 50-minute exercise bout was calculated to predict energy expenditure using the
Keytel-Hiiloskorpi equation for women (Keytel et al, 2005). Net energy expenditure was
25
then estimated by subtracting the energy cost of 50-minutes of no exercise using the Mifflin-
10
St Jeor equation for resting metabolic rate in women multiplied by the predicted energy cost
of sitting from the World Health Organisation technical report series appendices
(FAO/WHO/UNU, 1985).
5
Procedure
Participants arrived at 0900 and were required to stay in the research unit until the end of the
test session after lunch. On arrival at the unit they were provided with a fixed energy
breakfast (cereal with milk). Immediately before the exercise session, the liking and wanting
computer procedure and ratings of appetite were completed (T1). Participants were navigated
10
through the procedures at all times by written instructions presented on-screen. At 1100 they
began either the Ex or NEx activity. Immediately after the activity, participants completed the
liking and wanting procedure and subjective ratings of appetite (T2). They then showered
and changed before being provided with an ad libitum test meal at 1230. Participants ate
alone and were instructed to eat to comfortable fullness. Immediately after the test meal they
15
completed the liking and wanting procedure and a final set of appetite ratings (T3). Figure 1
shows a schematic diagram of the procedure. The testing procedure lasted approximately 2
hours. The experimental protocol was approved by the Institute of Psychological Sciences
Ethical Committee at University of Leeds. Participants gave their written consent and were
not paid for their participation.
20
Figure 1 about here
Treatment of data and statistical analyses
The main analyses for this report focused on the variability in compensatory eating measured
25
from energy intake at the test meal. Differences among participants classified on the basis of
11
relative energy intake (Ex-NEx) were used to guide further post hoc analyses and tested by
within-between subjects ANOVA.
Analysis of the liking and wanting measures focused on the explicit liking, implicit wanting
and relative food preference measures. The parametric data were analysed in separate 3-way
5
ANOVA with exercise activity (Ex or NEx), time (T1, T2, T3), and food category (HFSA,
LFSA, HFSW and LFSW) treated as repeated factors. Trends in the relative preference
measure were assessed descriptively.
From the initial sample of 24 women, two cases were excluded from the analyses due to
incomplete data, leaving a final sample of N=22.
10
12
Results
Energy intake and energy expenditure
5
As expected, differences in energy expenditure were highly significant between Ex and Nex
sessions. Mean heart rate and predicted energy expenditure were 94.2 ± 2.7 bpm and 189.3 ±
13.0 kcal for the Ex session and 81.4 ± 2.7 bpm and 47.0 kcal for the Nex session. No
differences were found between Ex and Nex for energy intake. Total energy intakes for the
Ex and NEx conditions were 1128.2 ± 72.8 and 1018.1 ± 73.0 kcal respectively.
10
Inter-individual variability in the acute effect of exercise on compensatory eating
Individual variability in relative energy intake (REI) was examined by subtracting energy
intake after NEx from energy intake after Ex. In addition, net differences in compensatory
eating were calculated by subtracting net energy expenditure (EE during Ex – EE during
15
NEx) from REI. These calculations provided measures of the impact of the exercise bout on
compensatory eating. Figure 2 illustrates this variability in compensatory eating and
demonstrates the range of responses including individuals who suppressed or made no
adjustment to their intake and some for whom exercise stimulated energy intake. As the mean
energy cost of exercise for the whole sample (n=22) was 149.0 kcal ±11.2, REI of >150kcal
20
indicated complete acute compensation.
Figure 2 about here
On the basis of this variation in relative energy intake, the sample was subdivided into two
25
groups of response. These groups corresponded to a) those who ate approximately the same
13
or less after exercise (non-compensators; N=11) and b) those who compensated or ate more
after exercise (compensators; N=11). The exercise-induced energy expenditure for the
compensators (146.3 kcal ±8.5) was similar to the non-compensators (151.2 kcal ±13.9).
5
Subject characteristics
The demographic and eating behaviour trait characteristics of compensators (C) and noncompensators (NC) were compared to try and identify factors that might explain the
differences in compensatory eating. C scored higher on BMI, % body fat and lower on self
reported frequency of habitual exercise compared with NC. Independent samples t-tests did
10
not reveal the differences to be statistically significant (highest t= -1.26, p=n.s.).
Subjective sensations of appetite and hedonic evaluation of the test meal
Examination of appetite sensations revealed no group differences in terms of hunger, fullness,
thirst or desire to eat. There was an overall main effect of time for hunger (F(2,40) = 46.35,
15
p<0.01) and desire to eat (F(2,40) = 54.13, p<0.01), with higher ratings after the intervention
compared to baseline (p<0.01) and lower ratings after consumption of the test meal (p<0.01).
As expected, a time*intervention interaction was observed for ratings of thirst (F(2,40) =
4.34, p<0.05) which increased following Ex compared to NEx.
Analysis of the hedonic evaluation of the test meal revealed a group*intervention interaction
20
in the NEx condition. Compensators rated the test meal as significantly less pleasant (F(1,20)
= 4.78, p<0.05) after NEx compared to NC. Figure 3 shows that the hedonic ratings after Ex
are similar in C and NC.
Figure 3 about here
25
14
Food preference computer procedure
Explicit liking
No group differences in explicit liking were found. There was a main effect of time (F(2,40)
= 19.53, p<0.01) with higher ratings after both Ex and NEx compared to baseline (p<0.01),
5
and lower ratings after the test meal compared to after Ex or NEx (p<0.01).
Implicit wanting
Analysis of implicit wanting responses revealed a significant three-way
group*intervention*time interaction (F(2,32) = 6.13, p<0.01). Further examination of the
10
means suggested that Ex exerted different effects on C compared with NC. Table 2 shows
mean values for implicit wanting (data for food categories pooled) for each time point.
Compensators’ reaction time decreased after exercise (T1-T2) relative to NC, while NC’s
reaction time increased after the test meal (T2-T3) relative to C. Post-hoc independent sample
t-tests revealed that C’s reaction time was slower for low fat stimuli at baseline (LFSA: t(20)
15
= -2.54, p<0.05), LFSW: t(20) = -3.61, p<0.01), while their reaction time after the test meal
was faster for high fat stimuli (HFSA: t(20) = 2.36, p<0.05), HFSW: t(20) = 2.16, p=0.06).
Table 2 about here
20
On examination of the mean frequency of choice for each food category, C showed a higher
preference for high-fat sweet foods compared to NC. This trend was relatively stable and did
not seem contingent on time or exercise condition.
15
Discussion
This study used a novel procedure to assess the role of food preference in the acute effect of
exercise on food intake and revealed some interesting and novel findings. Examination of
individual EI following exercise compared to no exercise revealed two diverging response
5
types in these participants; those who ate approximately the same or less after exercise
compared to no exercise (NC) and those who ate more after exercise compared to no exercise
(C). The identification of this individual difference in acute compensatory response revealed
some interesting further comparisons. Examination of the anthropometric characteristics of
both groups did not reveal any statistically significant differences; however C had on average
10
more body fat and a higher BMI. These are characteristics of individuals who have
experienced a positive energy balance and might be more susceptible to weight gain.
It is therefore surprising that no differences in subjective sensations of appetite were revealed
between the groups. This suggests that hunger levels were unlikely to have played a role in
the differences in food intake observed after Ex. The lack of an automatic exercise-induced
15
increase in hunger is in line with most of the recent literature that shows no relationship
between exercise and enhanced appetite (10;24;32;33). However, examination of homeostatic
mechanisms in isolation ignores the notion that other psychological and cognitive factors may
be influenced by volitional exercise to bring about changes in EI (34).
In the present study, some interesting group differences were apparent in the hedonic
20
evaluation of the test meal. Hedonic ratings of pleasantness differed between C and NC.
Compensators rated the meal as less pleasant after NEx while ratings were higher and did not
differ between groups after a bout of exercise. This suggests that while lower in the absence
of exercise, Ex exerted an effect on hedonic ratings of the meal in C but not in NC. Similar
exercise- induced increases in hedonic evaluation of sweetened preloads have been reported
25
in previous research (35;4), however intra-individual differences in the effect of exercise on
16
hedonic response to food has not previously been empirically demonstrated. These findings
may indicate a role for reward-based mechanisms in the tendency to overcompensate
following exercise.
This hypothesis was explored further by use of the novel food preference computer procedure
5
incorporating both explicit and implicit measures of liking and wanting. No significant group
differences were revealed in explicit liking for food stimuli presented in the task.
However, significant group differences were detected on the implicit wanting measure of the
task, which is less likely influenced by the cognitive interpretation of subjective feelings.
Indeed, the effect of exercise on C was pronounced. A single bout of high-intensity exercise
10
resulted in compensators responding faster compared to NC. Furthermore, after the test meal
(i.e. in a satiated state), the response time for NC was slower for savoury foods compared
with before the test meal (i.e. hungry state), while C maintained their pre-meal response rate.
These data reveal two distinct patterns of exercise-induced responses in C and NC. The
exercise-induced decrease in C’s response time, which was maintained in a fed state, may
15
partially explain the differences in test meal energy intake. It is also of interest to note the
unexpected differences at baseline between C and NC. Responses were slower overall in C
compared to the NEx treatment, and compared to NC. It is possible that anticipatory
responses to impending exercise could partly explain this group*condition interaction at
baseline. Implicit wanting for food may be influenced by the anticipation of exercise, and this
20
response may be different in C (who showed relatively less wanting/increased RT) and NC
(who showed a decreased RT at baseline under exercise). It is also interesting that acute
exercise appears to dissociate implicit and explicit hedonic response to food. We
speculatively suggest that metabolic signals associated with acute physical activity (for
example ghrelin (40; 41)) may selectively modulate neural pathways associated with wanting
25
for food in susceptible individuals.
17
Finally, further differences between the groups were expressed through the food choices in
the procedure (revealing relative preferences). It was noted that C had greater and sustained
preference for HFSW stimuli compared to NC. Therefore, C’s tendency to adjust EI after
exercise may be accompanied by other factors that could promote a positive energy balance.
5
Interestingly, selective preference for high-fat sweet foods has been demonstrated in subjects
with artificially-induced hypoglycaemia (36). Assuming that exercise caused a lowering of
blood glucose, one possibility is that differences in food preference may have resulted from
differences in glucostatic metabolism or status.
Compensators were characterised by enhanced implicit wanting for the food stimuli in the
10
task. Therefore, implicit processes rather than hunger or explicit liking for a range of food are
associated with compensatory feeding behaviour and may play a role in the failure of exercise
to facilitate weight loss in some individuals.
The lack of compensatory adjustment in EI experienced by the NC is in line with the general
consensus of research that homeostatic mechanisms of appetite are acutely insensitive to
15
perturbations in energy balance brought about by energy expenditure (e.g. 37;38). Recent
evidence from a long-term exercise intervention showed that despite large inter-individual
variability, some people don’t compensate and do experience the anticipated weight loss (39).
In the current study, compensators differed from NC in that they demonstrated some degree
of immediate compensatory response which in several of these subjects was enough to cancel
20
out the acute exercise-induced increase in energy expenditure. These individuals may
represent a wider population for whom exercise alone is not an appropriate strategy for weight
control.
Conclusion
18
Exercise-induced changes in the hedonic response to food could play an important role
in weight control. The enhanced implicit wanting for food after acute exercise may help to
explain the variable efficacy of exercise as a method of weight loss. Some individuals could
be resistant to the beneficial effects of exercise due to a predisposition to compensate for
5
exercise-induced energy expenditure as a result of changes in food preferences.
List of Abbreviations
BMI
10
15
C
Compensator
EI
Energy intake
EIe
Energy intake after Ex condition
EIn
Energy intake after NEx condition
Ex
Exercise
HARU
Human appetite research unit
HFSA
High fat savoury
HFSW
High fat sweet
LFSA
Low fat savoury
LFSW
Low fat sweet
NC
20
Body Mass Index
Non Compensator
NEx
No Exercise
VAS
Visual Analogue Scale
Author Contributions
EB conceived of the study and carried out the experimental work. GF participated in design of
25
the study, analysed the data and prepared the draft manuscript. JB participated in the design of
19
the study and helped revise the manuscript. NK participated in coordination and design of the
study and helped revise the manuscript. GF, JB and NK read and approved the final
manuscript.
5
20
References
01. King NA, Caudwell P, Hopkins M, Byrne NM, Colley R, Hills AP, Stubbs RJ, Blundell
5
JE: Automatic and volitional compensatory responses to exercise interventions: metabolic and
behavioural barriers to weight loss. Obesity, 2007, 15:1373-83.
02. Blundell JE, Stubbs RJ, Hughes DA, Whybrow S, King NA: Cross talk between physical
activity and appetite control: does physical activity stimulate appetite? Proc Nutr Soc, 2003,
10
62:651-61.
03. Leshem M, Abutbul A, Eilon R: Exercise increases the preference for salt in humans.
Appetite, 1999, 32:251-260.
15
04. Bellisle F: Food choice, appetite and physical activity. Public Health Nutrition, 1999,
2:357-361.
05. King NA, Snell L, Smith RD, Blundell JE: Effect of short-term exercise on appetite
responses in unrestrained females. Eur J Clin Nutr, 1996, 50:663-667.
20
06. Lluch A, King NA, Blundell JE: Exercise in dietary restrained women: no effect on
energy intake but change in hedonic ratings. Eur J Clin Nutr, 1998, 52:300-307.
07. Lluch A, King NA, Blundell JE: No energy compensation at the meal following exercise
25
in dietary restrained and unrestrained women. Br J Nutr, 2000, 84:219-225.
21
08. King NA, Blundell JE: High-fat foods overcome the energy expenditure induced by highintensity cycling or running. Eur J Clin Nutr, 1995, 49:114-23.
5
09. Tremblay A, Plourde G, Despres JP: Impact of dietary fat content and fat oxidation on
energy intake in humans. Am J Clin Nutr, 1998, 49:799-805.
10. Westerterp-plantenga MS, Verwegen CRT, Ijedema MJW, Wijckmans NEG, Saris WHM:
Acute effects of exercise or sauna on appetite in obese and nonobese men. Physiol Behav,
10
1997, 62:1345-1354.
11. Berg C, Sodersten P: Anorexia nervosa, self-starvation and the reward of stress. Nat Med,
1996, 2:21-22.
15
12. Lett BT, Grant VL, Gaborko LL: A small amount of wheel running facilitates eating in
non-deprived rats. Behavioral Neuroscience, 1996, 110:1492-1495.
13. Davis C, Strachan S, Berkson M: Sensitivity to reward: implications for overeating and
overweight. Appetite, 2004, 42:131-138.
20
14. Davis C, Fox J: Sensitivity to reward and body mass index (BMI): Evidence for a nonlinear relationship. Appetite, in Press.
15. Davis C, Woodside D: Sensitivity to rewarding effects of food and exercise in the eating
25
disorders. Compr Psychiatry, 2002, 43:189-194.
22
16. Berridge KC: Food reward: brain substrates of wanting and liking. Neurosci Biobehav
Rev, 1996, 20:1-25.
17. Berridge KC, Robinson TE: Parsing reward. Trends Neurosci, 2003, 26:507-513.
5
18. Mela DJ: Eating for pleasure or just wanting to eat? Reconsidering sensory hedonic
responses as a driver of obesity. Appetite, 2006, 47:10-17.
19. Finlayson GS, King NA, Blundell JE: The role of implicit wanting in relation to explicit
10
liking and wanting for food: Implications for appetite control. Appetite, 2008, 50, 120-127.
20. Finlayson GS, King NA, Blundell JE: Liking vs. wanting food: Importance for human
appetite control and weight regulation. Neurosci Biobehav Rev, 2007, 31, 987-1002.
15
21. Blundell JE, Finlayson GS: Is susceptibility to weight gain characterised by homeostatic
or hedonic risk factors for overconsumption? Physiol Behav, 2004, 82:21-25.
22. Berthoud HR: Neural control of appetite: cross-talk between homeostatic and nonhomeostatic systems. Appetite, 2004, 43:315-317.
20
23. Berthoud HR: Homeostatic and non-homeostatic pathways involved in the control of food
intake and energy balance. Obesity, 2006, Suppl 2:197-200.
24. King NA, Lluch A, Stubbs RJ, Blundell JE: High dose exercise does not increase hunger
25
or energy intake in free living males. Eur J Clin Nutr, 1997, 51:478-483.
23
25. Blundell JE, King NA: Effects of exercise on appetite control: loose coupling between
energy expenditure and energy intake. Int J Obes, 1999, 22:1-19.
5
26. Westerterp-Plentenga MS, Rolland V, Wilson SA, Westerterp KR: Satiety related to 24h
diet-induced thermogenesis during high protein/carbohydrate vs high fat diets measured in a
respiration chamber. Eur J Clin Nutr, 1999, 53:495-502.
27. King NA, Burley VJ, Blundell JE: Exercise-induced suppression of appetite: effects on
10
food intake and implications for energy balance. Eur J Clin Nutr, 1994, 48:715-726.
28. Horio T, Kawamura Y: Influence of physical exercise on human preferences for various
taste solutions. Chem Senses, 1998, 23:417-21.
15
29. Stunkard AJ, Messick S: The three-factor eating questionnaire to measure dietary
restraint, disinhibition and hunger. J Psychosom Res, 1985, 29:71-83.
30. Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. K. (1998). Measuring individual
differences in implicit cognition: The Implicit Association Test. Journal of Personality and
20
Social Psychology, 74, 1464–1480.
31. McCance RA, Widdowson EM: The composition of foods. The Royal Society of
Chemistry, Cambridge.
24
32. Blundell JE, King NA: Exercise, appetite control, and energy balance. Nutrition, 2000,
16:519-522.
33. Stubbs RJ, Hughes DA, Johnstone AM, Horgan GW, King NA, Blundell JE: A decrease
5
in physical activity affects appetite, energy, and nutrient balance in lean men feeding ad
libitum. Am J Clin Nutr, 2004, 79:62-69.
34. Hill JO, Melby C, Johnson SL, Peters JC: Physical activity and energy requirements. Am J
CLin Nutr, 62, 1995, Suppl 4:1059-1066.
10
35. King NA, Appleton K, Rogers PJ, Blundell JE: Effects of sweetness and energy drinks on
food intake following exercise. Physiol Behav, 1999, 66:375-379.
36. Dewan S, Gillett A, Mugarza JA, Dovey TM, Halford JCG, Wilding JPH: Effect of
15
insulin-induced hypoglycaemia on energy intake and food choice at a subsequent test meal.
Diabetes Metabolism Research Reviews, 2004, 20:405-410.
37. Stubbs RJ, Harbron CG, Murgatroyd P, Prentice A: Covert manipulation of dietary fat and
energy density: effect on substrate flux and food intake in men eating ad libitum. Am. J. Clin.
20
Nutr, 1995, 62:316-329.
38. Murgatroyd PR, Goldberg GR, Leahy FE, Gilsenan MB, Prentice AM: Effects of
inactivity and diet composition on human energy balance. Int J Obes, 1999, 23:1269-1281.
25
39. King NA, Caudwell P, Hopkins M, Stubbs RJ, Blundell JE. Individual variability
following 12 weeks of supervised exercise: identification and characterisation of
compensation for exercise-induced weight loss. Int J Obes, 2008, 32:177-184.
5
40. Mackelvie KJ, Meneilly GS, Elahi D, Wong AC, Barr SI, Chanoine JP. Regulation of
appetite in lean and obese adolescents after exercise: role of acylated and desacyl ghrelin. J
Clin Endocrinol Metab, 2007, 92:648-654.
41. Malik S, McGlone F, Bedrossian D, Dagher A. Ghrelin Modulates Brain Activity in Areas
10
that Control Appetitive Behavior. Cell Metabolism, 2007, 92:400-409.
26
Figure legends
Figure 1. A schematic diagram of the experimental protocol.
Legend: A=Appetite ratings, H=Hedonic ratings, L&W=Liking & Wanting.
5
Figure 2. Inter-individual variability in the acute effect of exercise on compensatory eating.
Cases are presented in rank order. Solid line indicates where energy intake was equal for Ex
and Nex conditions (no adjustment). Dashed line indicates where energy intake corrected for
net energy expenditure was equal for Ex and Nex conditions (perfect compensation).
10
Figure 3. Mean hedonic ratings (mm) of the test meal after Ex and NEx for C and NC groups.
Standard error shown as vertical lines. * p<0.05
Table 1. Photographic food stimuli used in the liking and wanting computer task (grouped by
15
food category). HFSA = high fat savoury, LFSA = low fat savoury, HFSW = high fat sweet, LFSW = low fat
sweet
Table 2. Mean (standard deviation) implicit wanting (choice reaction time in ms) averaged
20
25
across stimuli categories for Ex and NEx over time in C and NC groups.
27