Download Heightened sensitivity to punishment and reward in

Appetite 75 (2014) 97–102
Contents lists available at ScienceDirect
Appetite
journal homepage: www.elsevier.com/locate/appet
Research report
Heightened sensitivity to punishment and reward in anorexia nervosa q
Klaske A. Glashouwer a,b,⇑, Lotte Bloot a, Esther M. Veenstra a, Ingmar H.A. Franken c, Peter J. de Jong a
a
Department of Clinical Psychology, University of Groningen, Grote Kruisstraat 2/1, 9712 TS Groningen, The Netherlands
Center for Eating Disorders, Accare, Child and Adolescent Psychiatry, Smilde, The Netherlands
c
Institute of Psychology, Erasmus University Rotterdam, The Netherlands
b
a r t i c l e
i n f o
Article history:
Received 10 September 2013
Received in revised form 16 December 2013
Accepted 19 December 2013
Available online 31 December 2013
Keywords:
Reward
Punishment
Anorexia nervosa
Reinforcement sensitivity
a b s t r a c t
Objective: The aim of this study was to investigate reinforcement sensitivity in anorexia nervosa (AN). It
was tested whether self-reported punishment (PS) and reward sensitivity (RS) differed between adolescents with AN and healthy controls, and/or between AN-subtypes. In addition, the predictive validity of
PS and RS was examined for AN symptoms one year later. Method: In total, 165 female adolescents admitted for treatment of AN or eating disorder not otherwise specified resembling AN and 72 controls participated in the study. Participants completed measurements for eating disorder severity and the Sensitivity
to Punishment/Sensitivity to Reward Questionnaire (SPSRQ). Percentage of underweight and severity of
AN symptoms were measured again after one year in individuals with AN. Results: Individuals with AN
scored higher on PS and RS than controls. In addition, the AN purging type showed higher PS than the
AN restrictive type, whereas there were no differences in RS between AN-subtypes. Regression analyses
indicated that PS and RS were independently associated with the degree of eating disorder symptoms,
whereas only PS was related to percentage underweight. Yet, neither RS nor PS were related to percentage of underweight and AN symptoms after one year. Discussion: Although the present study clearly demonstrated that heightened punishment and reward sensitivity are both linked to AN, there was no
evidence that these characteristics are also involved in the course of AN symptoms.
Ó 2014 Elsevier Ltd. All rights reserved.
Introduction
Anorexia nervosa (AN) is a severe psychiatric disorder with the
highest mortality rate among all psychiatric disorders (Sullivan,
1995). Unfortunately, the underlying processes that cause and
maintain AN are still largely unknown. One of the puzzling features
of AN is that patients succeed in restricting their food intake, while
they actually are in a state of starvation. By nature, food has a high
reward value, even more for people that have been deprived of
food (e.g., Stroebe, Papies, & Aarts, 2008). So how do individuals
with AN overcome their biological drive to eat? One explanation
might be provided by the sensitivity of AN patients to environmental cues signaling reward or punishment. According to the Reinforcement Sensitivity Theory (e.g., Gray, 1987) individuals’
behaviors are guided by different brain systems. The behavioral
inhibition system reacts to aversive/punishing stimuli and is
q
Acknowledgements: We thank Sonja Fleminger for facilitating the present study
and for providing the required boundary conditions for testing the eating
disordered adolescents, Maike Scheperman, Annewieke Noot, Madelon van Hemel
– Ruiter and Birte Rahmsdorf and Nienke Boersma who helped collecting the data,
and those who kindly volunteered to participate in the study. We thank Gomarus
College in Groningen who provided us with controls. Declaration of interest: None.
⇑ Corresponding author.
E-mail address: [email protected] (K.A. Glashouwer).
0195-6663/$ - see front matter Ó 2014 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.appet.2013.12.019
thought to lead to avoidance behaviors. The behavioral activation
system is sensitive to rewarding stimuli and assumed to guide approach behaviors. Heightened punishment sensitivity might be a
risk factor for developing AN, because it could ‘help’ individuals
to abstain from eating. For individuals who are high in punishment
sensitivity and who base their self-esteem largely, or even exclusively, on their body shape and weight (Fairburn, 2008), the idea
of gaining weight can become very threatening. As a consequence,
they could start to avoid expected punishment (i.e. weight gain)
through avoidance behaviors such as restriction of food intake, or
excessive exercise, increasing the risk for developing AN.
Prior findings suggest that patients with AN indeed show
heightened punishment sensitivity supporting the hypothesis that
heightened punishment sensitivity is a risk factor for developing
AN (Beck, Smits, Claes, Vandereycken, & Bijttebier, 2009; Claes,
Nederkoorn, Vandereycken, Guerrieri, & Vertommen, 2006; Harrison, O’Brien, Lopez, & Treasure, 2010; Jappe et al., 2011). However,
up to now, studies mainly relied on adult samples. Yet, the impact/
relevance of reward and punishment sensitivity may be different
for younger, adolescent samples (e.g., adolescents generally seem
to be more sensitive to appetitive stimuli: Spear & Varlinskaya,
2010; Van Leijenhorst et al., 2010). In addition, and most critical
for the present context, the first onset of AN often takes place during early adolescence and the heightened punishment sensitivity
98
K.A. Glashouwer et al. / Appetite 75 (2014) 97–102
found in adults with AN might be the consequence of protracted
AN rather than a premorbid risk factor. Therefore, the first goal
of the present study is to examine whether punishment sensitivity
is already heightened in young adolescents with AN. This would
further support the idea that heightened punishment sensitivity
might indeed be a risk factor for developing AN.
In addition, an alternative explanation for the striking ability of
AN patients to resist (or overcome) their drive to eat could be that
they differ from healthy controls in their sensitivity to reward. Perhaps lowered reward sensitivity leads to a lower approach motivation to food cues, which increases the risk of developing AN.
Results in this respect are still mixed. Some studies showed that
individuals with AN of the restrictive type (AN-R) indeed displayed
lower levels of reward sensitivity than controls, while individuals
with AN of the purging/binging type (AN-P) showed higher levels
of reward sensitivity than individuals with AN-R and/or controls
(Beck et al., 2009; Claes et al., 2006; Harrison et al., 2010). The latter could help explain why individuals with AN-P sometimes show
episodes of binge eating. However, recently, Jappe and colleagues
(2011) showed the opposite, namely that both types of women
with AN displayed higher sensitivity to reward than healthy controls. They did not find differences in reward sensitivity between
AN-R and AN-P. Consequently, the second goal of the present study
is to examine reward sensitivity in a large group of AN patients,
since in prior studies especially subgroups of AN-P were usually
small. Because prior findings with respect to reward sensitivity
in AN were mixed, we do not have a clear hypothesis.
Although prior studies suggest that reward and punishment
sensitivity generally are independent constructs (see e.g., Franken
& Muris, 2006; O’Connor, Colder, & Hawk, 2004), it is still unknown
whether these constructs are also independently involved in AN, or
whether they should be seen as overlapping risk factors. Furthermore, we explore whether perhaps the relationship between reward sensitivity and the onset of AN is moderated by
punishment sensitivity, or vice versa (see e.g., Corr, 2002). Consequently, the third goal of this study is to test whether punishment
and reward sensitivity as well as their interaction are independently related to severity of eating disorder symptoms. Our
hypothesis is that both factors are independently linked to severity
of AN.
Finally, reward and punishment sensitivity might not only be
risk factors for the onset of AN, but perhaps they also play a role
in the persistence of AN over time. Therefore, the last goal is to
examine whether in individuals with AN, punishment sensitivity,
reward sensitivity, and/or their interaction predict AN symptoms
later in time. More clarity with respect to the potential role of punishment and reward sensitivity as risk factors for the development
and maintenance of AN-R and AN-P might be crucial for a better
understanding of AN and its underlying processes.
Method
Participants
The clinical group existed of 165 female adolescents who met
the criteria of AN or an eating disorder not otherwise specified
resembling AN1: 48 of the purging type (AN-P) and 117 of the
1
84 Participants fulfilled all criteria of AN; 20 participants fulfilled all criteria for
AN except criterion A ‘underweight’ which was between 0–15%; 25 participants
fulfilled all criteria for AN except criterion B ‘fear of gaining weight’; 2 participants
fulfilled all criteria for AN except criterion C ‘disturbance in the way in which one’s
body weight or shape is experienced, undue influence of body shape on selfevaluation, or denial of the seriousness of the current low body weight’; 11
participants fulfilled all criteria for AN except criterion D ‘amenorrhea’; 18 participants fulfilled all criteria for AN except criterion A and D; 5 participants fulfilled all
criteria for AN except criterion B and D.
restrictive type (AN-R). Participants with AN were recruited through
the Department of Eating Disorders of Accare, a facility for child and
adolescent psychiatry in the Netherlands. In addition, we included
72 female control participants who were recruited via a local high
school (Gomarus College in Groningen, the Netherlands). Control
participants did not differ significantly from AN participants on
age and educational level (see Table 1) and shared the same ethnic
background as the AN group (Caucasian). Control participants were
screened on eating disorder symptoms with the Eating Disorder
Examination-Questionnaire (EDE-Q (Fairburn & Bèglin, 1994)). The
measurements were part of routine outcome monitoring assessment
which was approved beforehand by the local authority of Accare,
child- and adolescent psychiatry. Participants of the control group
and their parents actively gave informed consent and the study protocol was approved by the Ethical Committee of Psychology of the
University of Groningen.
Procedure and measures
Clinical group. Within two weeks after registration participants
of the clinical group were diagnosed by health care professionals
of Accare using the Dutch child version of the Eating Disorder
Examination (ChEDE (Bryant-Waugh, Cooper, Taylor, & Lask,
1996; Decaluwé & Braet, 1999)). In addition, participants filled in
the Dutch version of the EDE-Q (Fairburn & Bèglin, 1994). The
EDE-Q is a self-report questionnaire which consists of 41 items
and reflects the format of the EDE interview including the four subscales and the global severity score. Furthermore, participants
completed the Dutch version of the Sensitivity to Punishment
and Sensitivity to Reward Questionnaire (SPSRQ (Franken & Muris,
2006; Torrubia, Ávila, Moltó, & Caseras, 2001)). The SPSRQ is a selfreport instrument and contains 48 items in a yes/no format. The
questionnaire provides scores for individual sensitivity to punishment (SP), and sensitivity to reward (SR) (range: 0–24).2 The SPSRQ
has already been widely used and is designed to index sensitivity to
specific cues that signal reward and punishment, in line with Gray’s
theory (Torrubia et al., 2001). The wording of some items was
slightly adapted to make them appropriate and understandable for
the adolescent age group. Psychometric evaluation showed that both
subscales of the SPSRQ present satisfactory internal consistency and
test–retest reliability as well as convergent and discriminant validity
(Torrubia et al., 2001). Furthermore, the validity of the Dutch version
of the SPSRQ has been shown to be comparable with the validity of
the original version in ED patients (Beck et al., 2009) and the SPSRQ
seems to be suitable for use among adolescents (e.g., Matton, Goossens, Braet, & Vervaet, 2013). Finally, weight and height data were
collected as well as demographics. Percentage underweight was calculated by dividing the current weight through the target weight
appropriate for the length and age of each participant multiplied
by one hundred. This number was deducted from 100. The target
weight was determined using growth curves (TNO).
Control group. The procedure for participants of the control
group was similar to that of the clinical group, with exception of
the ChEDE-interview, which was not administered in the control
group. Control participants were assessed once.
Study design
The clinical group was measured twice: at baseline the ChEDE,
the EDE-Q, the SPSRQ and weight and height data were collected.
Percentage of underweight and EDE-Q were measured again after
2
We also calculated the scales as proposed by O’Connor, Colder and Hawk (2004).
However, since the outcomes of the statistical analyses were generally similar to the
outcomes of the original calculations, we decided only to report the analyses with the
original calculations.
99
K.A. Glashouwer et al. / Appetite 75 (2014) 97–102
Table 1
Group characteristics for AN groups and healthy controls.
Age
Educationa
Underweight (%)
EDE-Q – Restraint
EDE-Q – Eating Concern
EDE-Q – Weight Concern
EDE-Q – Shape Concern
EDE-Q – Total Score
SPSRQ – Punishment
SPSRQ – Reward
(A) Controls
(n = 72)
M (SD)
(B) AN
(n = 165)
M (SD)
AN-R
(n = 117)
M (SD)
AN-P
(n = 48)
M (SD)
(C) AN T2
(n = 150)
M (SD)
t-Test
A vs. B
p
Paired t-test
B vs. C
p
14.84 (1.71)
34:66
3.19 (12.51)
0.82 (0.85)
0.46 (0.42)
0.94 (0.85)
0.96 (0.93)
0.80 (0.64)
8.35 (5.30)
8.05 (3.28)
15.02 (1.71)
43:57
20.26 (8.47)
3.34 (1.74)
2.91 (1.33)
3.75 (1.57)
4.09 (1.55)
3.52 (1.32)
14.44 (5.30)
9.70 (4.15)
14.96 (1.78)
43:57
21.66 (7.57)
2.96 (1.69)
2.74 (1.39)
3.47 (1.60)
3.85 (1.63)
3.26 (1.36)
13.57 (5.32)
9.55 (4.11)
15.19 (1.54)
43:57
16.85 (9.60)
4.27 (1.49)
3.30 (1.06)
4.45 (1.24)
4.69 (1.14)
4.17 (0.95)
16.54 (4.67)
10.09 (4.25)
–
–
5.46
2.09
1.71
2.28
2.82
2.22
–
–
.52
.47b
<.001
<.001
<.001
<.001
<.001
<.001
<.001
.001
–
–
.013
<.001
<.001
<.001
<.001
<.001
–
–
(10.16)
(2.03)
(1.53)
(1.65)
(1.70)
(1.40)
Note: SPSRQ = Sensitivity to Punishment and Sensitivity to Reward Questionnaire (Torrubia et al., 2001), EDE-Q = Eating Disorder Examination Questionnaire (Fairburn &
Bèglin, 1994), AN = Anorexia Nervosa.
a
% Lower:% higher education, for 28 participants of the AN-R group and 13 participants of the AN-P group educational level was missing.
b
Chi-square test was performed.
one year. In between baseline and follow-up patients received
treatment as usual, which meant that patients received inpatient
and/or outpatient treatment which included one or a combination
of the following therapies: individual treatment, group treatment,
family treatment, cognitive behavioral therapy, dietary advice or
medication. For control participants data was collected once for
the EDE-Q, the SPSRQ and weight and height.
Statistical analyses
ANOVA analyses and post hoc Hochbergs GT2 tests were used to
compare SPSRQ scores between the AN groups and the control
group. Pearson correlation analyses were conducted to explore
the relationships at baseline between the SPSRQ subscales and
the percentage underweight and EDE-Q. Linear regression analyses
were conducted to test whether punishment and reward sensitivity were independently related to eating disorder symptom severity and percentage underweight at baseline and to test predictive
validity over time.
Results
Missing data and drop-out
Originally, 76 control participants were included, but four control participants scored above the cut-off of 2.3 points on the global
severity score on the EDE-Q and were therefore excluded (cf.
Mond, Hay, Rodgers, Owen, & Beumont, 2004). This resulted in a final group of 72 controls. Of the 165 participants in the AN group at
baseline, 40 individuals had missing data on the SPSRQ and 18
individuals had missing data on the EDE-Q. Of the 72 controls at
baseline, 3 individuals had missing data on percentage underweight, 2 individuals had missing data on the SPSRQ, and 1 individual had missing data on the EDE-Q.
Unfortunately, after one year there was a substantial amount of
drop-out (41%) in the clinical group. There were various reasons for
dropping out: living too far away from the treatment center, doing
well and not wanting to be reminded, doing poorly and not feeling
able to participate, just not wanting to participate. Furthermore, 15
participants were not measured because they were referred to another treatment center or the follow-up measurement was not
conducted yet. Of the 165 participants with AN measured at baseline 88 participated in the 1-year follow-up (t2). Furthermore, 5
individuals had additional missing data for the percentage of
underweight at t2 and 3 for the EDE-Q at t2. To test whether patients who remained in the study differed from patients who
dropped out, we compared both groups at baseline by means of
independent sample t-tests. However, drop-outs and completers
did not differ on age, educational level, percentage underweight,
scores on the SPSRQ, type of AN, and scores on the ChEDE or
EDE-Q.
Missing data were estimated using multiple imputation which
was developed as one of the state-of-the-art and preferred methods for dealing with missing data (Schafer & Graham, 2002). Missing data was imputed 40 times with predictive mean matching
using IBM SPSS Statistics 20, based on all predictors that were included in the model as well as other variables outside the present
study (e.g., depressive symptoms, body image) to impute as accurately as possible. We report the pooled results.
Descriptives
In this study internal consistencies of both subscales of the
SPSRQ were good (PS: a = 0.90; RS: a = 0.77). Group characteristics
on all variables are reported in Table 1. Furthermore, Pearson’s correlations were calculated between the SPSRQ subscales, the EDE-Q
total score and percentage underweight at baseline over all participants (N = 237). As expected, the SP-subscale and the SR-subscale
did not correlate (r = .10, p = .15). Furthermore, reward sensitivity
was weakly correlated with the EDE-Q (r = .24 p < .001), whereas
punishment sensitivity showed a moderate correlation with the
EDE-Q (r = .49, p < .001). Percentage underweight correlated with
punishment sensitivity (r = .35, p < .001), but not with reward sensitivity (r = .08, p = .26). Percentage underweight and EDE-Q total
score were moderately correlated (r = .48 p < .001). Participants
with AN showed significantly lower symptom levels at t2 than at
baseline (Table 1).
Group differences on punishment and reward sensitivity
A three-group (AN-R, AN-P, controls) ANOVA on the SP-subscale
showed a significant main effect for group (F(2, 234) = 39.91,
p < .001, partial g2 = 0.26). As expected, both AN-groups showed
higher levels of punishment sensitivity than healthy controls
(AN-R vs. controls: mean difference = 5.23, p < .001, 95% CI
[3.35, 7.10]; AN-P vs. controls: mean difference = 8.19 p < .001,
95% CI [5.86, 10.52]). Furthermore, the AN-P group showed higher
punishment sensitivity than the AN-R group (mean difference = 2.96 p = .005, 95% CI [0.82, 5.10]).
A three-group (AN-R, AN-P, controls) ANOVA on the SR-subscale
showed a significant main effect for group (F(2, 234) = 4.93, p = .01,
partial g2 = 0.04). Both AN-groups showed higher levels of reward
100
K.A. Glashouwer et al. / Appetite 75 (2014) 97–102
sensitivity than healthy controls (AN-R vs. controls: mean difference = 1.50, p = .044, 95% CI [0.09, 2.91]; AN-P vs. controls: mean
difference = 2.04, p = .024, 95% CI [0.29, 3.80]). However, AN-R
and AN-P did not differ in their reward sensitivity (p = .75).
Predictive value of punishment and reward sensitivity at baseline
EDE-Q total scores. To test whether punishment and reward sensitivity were independently related to the EDE-Q scores at t1 we
conducted linear regression analysis with SP and SR as predictors
and EDE-Q total score at t1 as dependent variable. The model
(F(2, 234) = 45.8, p < .001, R2 = .28) was significant and both SP
and SR showed independent predictive validity for EDE-Q scores
(SP: b = .48; SR: b = .20). In a second step, the interaction between
SP and SR was added as predictor, but this did not improve the
model
significantly
(Fchange(1, 233) = 1.45,
p = .26,
R2change = .004).3
Percentage underweight. To test whether punishment and reward sensitivity were independently related to the percentage
underweight we conducted linear regression analysis with SP
and SR as predictors and the percentage underweight at t1 as
dependent variable. The model (F(2, 234) = 16.90, p < .001;
R2 = .13) was significant, but only SP showed independent predictive validity for percentage underweight (SP: b = .35; SR: b = .04).
In a second step, the interaction between SP and SR was added
as predictor, but this did not improve the model significantly
(Fchange(1, 233) = 0.40, p = .55, R2change = .001).
Predictive value of punishment sensitivity and reward sensitivity over
time
EDE-Q total scores. To test the predictive validity of SP and SR for
severity of eating disorder symptoms after one year we conducted
linear regression analysis with EDE-Q total score at t1 as predictor
and EDE-Q total score at t2 as dependent variable. The model was
shown to be marginally significant (F(1, 148) = 5.83, p = .05,
R2 = .04), indicating that EDE-Q scores at baseline were positively related to EDE-Q scores one year later (b = .19). SP and SR were included
as predictors in the second step, but this did not improve the model
significantly (Fchange(2, 146) = 0.50, p = .66, R2change = .006). In
the final step, the interaction between SP and SR was added as predictor. Again this did not significantly improve the regression model
(Fchange(1, 145) = 0.49, p = .58, R2change = .003).4
Percentage underweight. Linear regression analysis was used to
test the predictive validity of SP and SR for percentage underweight one year later. In the first step, we included percentage
underweight at t1 as predictor and percentage underweight at t2
as dependent variable. The model was not significant
(F(1, 148) = 1.62, p = .30, R2 = .01), indicating that percentage
underweight at baseline was not related to percentage underweight one year later. SP and SR were included as predictors in
the second step, but this did not improve the model significantly
(Fchange(2, 146) = 0.92, p = .50, R2change = .01). In the final step,
the interaction between SP and SR was added as predictor. Again
this did not significantly improve the regression model
(Fchange(1, 145) = 0.75, p = .50, R2change = .005).
Discussion
The main goal of the present study was to explore whether selfreported punishment and reward sensitivity might be considered
as risk factors for developing AN, and/or the persistence of AN
3
When we repeated the analysis with the total scale score of the ChEDE interview
at baseline as dependent variable, only SP showed independent predictive validity.
4
Similar results were obtained when using the total scale score of the ChEDE
interview.
symptoms. In line with prior studies in adults, outcomes showed
that adolescents with AN already display higher levels of punishment sensitivity than healthy controls (Harrison et al., 2010; Jappe
et al., 2011). However, in contrast to prior studies, individuals with
AN-P showed (even) higher punishment sensitivity than individuals with AN-R. The latter finding might be related to greater symptom severity in the AN-P group than in the AN-R group (M
difference EDE-Q total score = 0.92), since punishment sensitivity
and symptom severity were positively correlated (all participants:
r = .49). In addition, adolescents with AN showed higher levels of
reward sensitivity than healthy controls. Our findings are in line
with recent work among nonclinical adolescents which showed
that specifically high sensitivity to punishment in combination
with high sensitivity to reward was related to eating problems
(Matton et al., 2013). In addition, we found that individuals with
AN-P and AN-R did not significantly differ from each other with respect to reward sensitivity. This latter result is similar to the outcomes of Jappe and colleagues (2011), but differs from other
studies showing that individuals with AN-R displayed lower levels
of reward sensitivity than controls, while individuals with AN-P
showed higher levels of reward sensitivity than individuals with
AN-R and/or controls (Beck et al., 2009; Claes et al., 2006; Harrison
et al., 2010).
The inconsistencies with prior studies in reward sensitivity might
be due to different definitions and measurement instruments that
are being used to measure reward sensitivity. Studies that showed
lowered reward sensitivity in AN-R and heightened reward sensitivity in AN-P typically used the Behavioral Inhibition/Behavioral Activation System Scales (BIS/BAS) of Carver and White (1994) or the
Temperament and Character Inventory (TCI) of Cloninger, Svrakic,
and Przybeck (1993). Differences in reward sensitivity were only
demonstrated for the ‘fun seeking’ subscale of the BIS/BAS and the
‘novelty seeking’ subscale of the TCI which were shown to measure
impulsivity rather than reward sensitivity (Franken & Muris, 2006).
Although reward sensitivity and impulsivity both are being linked to
the behavioral activation system, the concepts are not interchangeable. Impulsivity typically refers to behaviors that are rash and spontaneous, while reward sensitivity reflects a heightened sensitivity to
cues that signal reward (cf. Dawe, Gullo, & Loxton, 2004). Perhaps
individuals with AN-P do not so much differ from AN-R in their sensitivity to reward, but in their impulsivity. The latter would explain
why individuals with AN-P show purging behaviors or engage in episodes of binge eating. Consequently, future studies should include
indices of both reward sensitivity and impulsivity to test whether indeed differences in impulsivity explain prior findings of higher reward sensitivity in AN-P than AN-R.
However, it still remains to be explained why individuals with
AN typically score similar to controls on the other subscales of
the BIS/BAS, ‘reward responsiveness’ and ‘drive’, while they score
higher than controls on the RS subscale of the SPSRQ in the present
study (and in Jappe et al. (2011)). The latter is especially puzzling,
taking into account that Franken and Muris (2006) showed that
these BIS/BAS subscales are loading higher on the factor that
seemed to reflect reward sensitivity than the RS subscale of the
SPSRQ. If individuals with AN would be actually more sensitive
to reward, one would expect to find this on both the BIS/BAS and
the SPSRQ. Taking a closer look at the separate items of the scales
might elucidate the apparent inconsistent findings. While the BIS/
BAS aims to measure how individuals generally respond to reinforcement, the RS subscale of the SPSRQ measures sensitivity to
overt rewarding stimuli, such as money, power, etc. In the present
study differences between the AN group and controls were mainly
driven by two items regarding appearance (items 6 and 14) and
five items regarding interpersonal rewards (items 10, 16, 22, 44,
46). When we excluded these items from the analyses, the difference between the AN group and the control group were no longer
K.A. Glashouwer et al. / Appetite 75 (2014) 97–102
significant. The BIS/BAS does not contain any items on appearance
or interpersonal reward. Thus, individuals with AN seem to be sensitive to reward within the domains of appearance and interpersonal feedback, instead of reward in general.
Taken together, the present pattern of findings suggests that
individuals with AN show heightened levels of punishment sensitivity and are more sensitive to rewards with respect to appearance and interpersonal feedback. In addition, punishment and
reward sensitivity were independently related to the severity of
eating disorder symptoms, but only punishment sensitivity was related to percentage underweight. These findings are in line with
the idea that heightened punishment and reward sensitivity might
be risk factors for developing AN, and that these temperamental
factors could ‘help’ individuals to abstain from eating. That is, because of high sensitivity to punishment, fears like ‘becoming fat’
might become so prominent that individuals restrict their food intake, thereby increasing the risk for AN. Furthermore, the focus on
appearance and obtaining approval of others might also set individuals at risk for developing anorectic symptoms, such as weight
and shape concerns. However, since this study has a cross-sectional design, it is impossible to determine whether punishment
and reward sensitivity indeed precede the onset of AN, or whether
they are merely correlates of AN.
In addition, as a first step to examine whether reward and punishment sensitivity might be risk factors for the persistence of AN
over time, we tested whether punishment and reward sensitivity
predicted the course of AN symptoms. Both punishment and reward sensitivity were not related to eating disorder symptoms
and percentage underweight one year later. So, although heightened punishment and reward sensitivity seem to be correlates of
AN, apparently these characteristics are not related to changes in
AN symptoms over time (due to treatment). Additionally, the influence of reward sensitivity on (the course of) AN symptoms was not
moderated by punishment sensitivity or vice versa, since we did
not find significant interaction effects. However, since AN patients
received treatment as usual (without experimental control), it cannot be ruled out that individuals with higher levels of punishment
and reward sensitivity received relatively intensive treatment,
thereby counteracting the potential inadvertent influence of reinforcement sensitivity on the course of symptoms.
Limitations
The present study suffered from a substantial amount of dropout (41%) which is unfortunately not exceptional in this field.
Although we used missing data imputation, the outcomes of the
longitudinal analyses should be interpreted with care. Additionally, it should be mentioned that the Reinforcement Sensitivity
Theory meanwhile was updated (Gray & McNaughton, 2000), but
that the SPSRQ is still based on the old model. Hereby, it seems that
different measurement instruments in this area measure different
aspects related to reinforcement sensitivity, which makes it a
highly complex matter to compare outcomes across different studies in this field. Finally, it should be mentioned that we did not include measures of (trait) depression/anxiety, or other potentially
relevant third variables in our design. Consequently, it is impossible to investigate whether reward and punishment sensitivity
might be proxy (or overlapping) risk factors (for more information
about different kind of risk factors that can be distinguished see:
Kraemer, Kazdin, Offord, & Kessler, 1997; Kraemer, Stice, Kazdin,
Offord, & Kupfer, 2001).
Future directions
To determine whether punishment and reward sensitivity indeed can be considered pre-morbid risk factors for developing
101
AN, a crucial next step would be to longitudinally study their relationship with onset of AN in a healthy sample of adolescents. Subsequently, to see whether it concerns ‘‘fixed’’ markers or
characteristics that are amenable to change, it would be important
to examine whether successful treatment leaves reward and punishment sensitivity unchanged. A fruitful way to unravel which
(causal) mechanisms underlie the link between punishment and
reward sensitivity and AN could be the use of behavioral outcome
measures. This might not only give more detail to the interpretation of the present findings, but also might show that behavioral
and subjective reward and punishment sensitivity could be (partly)
independent risk factors. Germane to this, research in the field of
addiction recently showed that substance use in adolescents was
related to enhanced attentional engagement with cues that predicted potential reward (van Hemel-Ruiter, de Jong, Oldehinkel,
& Ostafin, 2013). It seems valuable to apply similar behavioral
tasks to the context of eating disorders. In addition, it seems
important to link the behavioral tasks to functional brain activity,
to see whether there exist differences in activation of reward and
punishment circuits between AN individuals and healthy controls
(e.g., Beaver et al., 2006). Obviously, such processes eventually
should be longitudinally studied to determine whether they are
pre-morbid risk factors for AN and/or for persistence of AN symptoms. When sensitivity to reward and sensitivity to punishment do
appear to be important factors in the development and/or maintenance of AN, this could provide fresh theory-driven starting points
for improving present interventions. Perhaps treatment for AN
could be improved by more directly addressing these factors, such
as learning patients to focus more on safety information instead of
punishing information or using reward sensitivity to find alternative sources for functional rewards as a replacement of the dysfunctional role of the eating disorder (e.g., giving a sense of
control, feeling special; cf. Fairburn, 2008).
Conclusion
The present pattern of findings showed that heightened reward
and punishment sensitivity are correlates of AN, which is consistent with the view that they might be risk factors for developing
AN. Especially punishment sensitivity seems associated with percentage underweight, and heightened punishment sensitivity
may therefore be most important as a risk factor for AN. In line
with the alleged critical importance of overvalued ideas about
weight and shape in AN, specifically reward sensitivity with respect to appearance and interpersonal feedback was heightened
in AN. There were no differences in this respect between AN-P
and AN-R subgroups, thus enhanced reward sensitivity seems not
especially relevant in explaining the presence of binging and purging behaviors in AN-P. The present study found no evidence that
individuals’ reward and punishment sensitivity are risk factors
for the persistence of AN over time. Thus, although the present
study clearly demonstrated that heightened punishment and reward sensitivity are both linked to AN, these characteristics do
not seem to make individuals with AN more resistant to treatment.
References
Beaver, J. D., Lawrence, A. D., van Ditzhuijzen, J., Davis, M. H., Woods, A., & Calder, A.
J. (2006). Individual differences in reward drive predict neural responses to
images of food. The Journal of Neuroscience, 26, 5160–5166. http://dx.doi.org/
10.1523/JNEUROSCI.0350-06.2006.
Beck, I., Smits, D. J. M., Claes, L., Vandereycken, W., & Bijttebier, P. (2009).
Psychometric evaluation of the behavioral Inhibition/Behavioral activation
system scales and the sensitivity to punishment and sensitivity to reward
questionnaire in a sample of eating disordered patients. Personality and
Individual
Differences,
47,
407–412.
http://dx.doi.org/10.1016/
j.paid.2009.04.007.
102
K.A. Glashouwer et al. / Appetite 75 (2014) 97–102
Bryant-Waugh, R., Cooper, P. J., Taylor, C. L., & Lask, B. D. (1996). The use of the
eating disorder examination with children. A pilot study. International Journal of
Eating
Disorders,
19,
391–397.
10.1002/(SICI)1098108X(199605)19:4<391::AID-EAT6>3.0.CO;2-G.
Carver, C. S., & White, T. L. (1994). Behavioral inhibition, behavioral activation, and
affective responses to impending reward and punishment. The BIS/BAS scales.
Journal of Personality and Social Psychology, 67, 319–333. http://dx.doi.org/
10.1037/0022-3514.67.2.319.
Claes, L., Nederkoorn, C., Vandereycken, W., Guerrieri, R., & Vertommen, H. (2006).
Impulsiveness and lack of inhibitory control in eating disorders. Eating
Behaviors, 7, 196–203. http://dx.doi.org/10.1016/j.eatbeh.2006.05.001.
Cloninger, C. R., Svrakic, D. M., & Przybeck, T. R. (1993). A psychobiological model of
temperament and character. Archives of General Psychiatry, 50, 975–990. http://
dx.doi.org/10.1001/archpsyc.1993.01820240059008.
Corr, P. J. (2002). J. A. Gray’s reinforcement sensitivity theory. Tests of the joint
subsystems hypothesis of anxiety and impulsivity. Personality and Individual
Differences, 33, 511–532. http://dx.doi.org/10.1016/S0191-8869(01)00170-2.
Dawe, S., Gullo, M. J., & Loxton, N. J. (2004). Reward drive and rash impulsiveness as
dimensions of impulsivity. Implications for substance misuse. Addictive
Behaviors, 29, 1389–1405. http://dx.doi.org/10.1016/j.addbeh.2004.06.004.
Decaluwé, V., & Braet, C. (1999). Dutch translation of the child eating disorder
examination authored by C. G. Fairburn, Z. Cooper & R. Bryant-Waugh.
Unpublished manuscript.
Fairburn, C. G. (2008). Cognitive behavior therapy and eating disorders. New York, NY,
US: Guilford Press.
Fairburn, C. G., & Bèglin, S. J. (1994). Assessment of eating disorders. Interview or
self-report questionnaire? International Journal of Eating Disorders, 16, 363–370.
Franken, I. H. A., & Muris, P. (2006). Gray’s impulsivity dimension. A distinction
between reward sensitivity versus rash impulsiveness. Personality and
Individual
Differences,
40,
1337–1347.
http://dx.doi.org/10.1016/
j.paid.2005.11.016.
Gray, J. A. (1987). The psychology of fear and stress (2nd ed.). New York, NY US:
Cambridge University Press.
Gray, J. A., & McNaughton, N. (2000). The neuropsychology of anxiety. An enquiry in to
the functions of the septo-hippocampal system (2nd ed.). Oxford: Oxford
University Press.
Harrison, A., O’Brien, N., Lopez, C., & Treasure, J. (2010). Sensitivity to reward and
punishment in eating disorders. Psychiatry Research, 177, 1–11. http://
dx.doi.org/10.1016/j.psychres.2009.06.010.
Jappe, L. M., Frank, G. K. W., Shott, M. E., Rollin, M. D. H., Pryor, T., Hagman, J. O., et al.
(2011). Heightened sensitivity to reward and punishment in anorexia nervosa.
International Journal of Eating Disorders, 44, 317–324. http://dx.doi.org/10.1002/
eat.20815.
Kraemer, H. C., Kazdin, A. E., Offord, D. R., & Kessler, R. C. (1997). Coming to terms
with the terms of risk. Archives of General Psychiatry, 54, 337–343. http://
dx.doi.org/10.1001/archpsyc.1997.01830160065009.
Kraemer, H. C., Stice, E., Kazdin, A., Offord, D., & Kupfer, D. (2001). How do risk
factors work together? mediators, moderators, and independent, overlapping,
and proxy risk factors. The American Journal of Psychiatry, 158, 848–856. http://
dx.doi.org/10.1176/appi.ajp.158.6.848.
Matton, A., Goossens, L., Braet, C., & Vervaet, M. (2013). Punishment and reward
sensitivity. Are naturally occurring clusters in these traits related to eating and
weight problems in adolescents? European Eating Disorders Review, 21,
184–194. http://dx.doi.org/10.1002/erv.2226.
Mond, J. M., Hay, P. J., Rodgers, B., Owen, C., & Beumont, P. J. V. (2004). Validity of the
eating disorder examination questionnaire (EDE-Q) in screening for eating
disorders in community samples. Behaviour Research and Therapy, 42, 551–567.
http://dx.doi.org/10.1016/S0005-7967(03)00161-X.
O’Connor, R. M., Colder, C. R., & Hawk, L. W. J. (2004). Confirmatory factor analysis of
the sensitivity to punishment and sensitivity to reward questionnaire.
Personality and Individual Differences, 37, 985–1002. http://dx.doi.org/10.1016/
j.paid.2003.11.008.
Schafer, J. L., & Graham, J. W. (2002). Missing data. Our view of the state of the art.
Psychological
Methods,
7,
147–177.
http://dx.doi.org/10.1037/1082989X.7.2.147.
Spear, L. P., & Varlinskaya, E. I. (2010). Sensitivity to ethanol and other hedonic
stimuli in an animal model of adolescence. Implications for prevention science?
Developmental
Psychobiology, 52, 236–243. http://dx.doi.org/10.1002/
dev.20457.
Stroebe, W., Papies, E. K., & Aarts, H. (2008). From homeostatic to hedonic theories
of eating. Self-regulatory failure in food-rich environments. Applied Psychology.
An International Review, 57, 172–193. http://dx.doi.org/10.1111/j.14640597.2008.00360.x.
Sullivan, P. F. (1995). Mortality in anorexia nervosa. The American Journal of
Psychiatry, 152, 1073–1074.
TNO (2011). Groeidiagram 1–21 jaar. vijfde landelijke groeistudie. <www.tno.nl/
groei>. Retrieved 20.09.11.
Torrubia, R., Ávila, C., Moltó, J., & Caseras, X. (2001). The sensitivity to punishment
and sensitivity to reward questionnaire (SPSRQ) as a measure of gray’s anxiety
and impulsivity dimensions. Personality and Individual Differences, 31, 837–862.
http://dx.doi.org/10.1016/S0191-8869(00)00183-5.
van Hemel-Ruiter, M. E., de Jong, P. J., Oldehinkel, A. J., & Ostafin, B. D. (2013).
Reward-related attentional biases and adolescent substance use. The TRAILS
study. Psychology of Addictive Behaviors, 27, 142–150. http://dx.doi.org/10.1037/
a0028271.
Van Leijenhorst, L., Moor, B. G., Op, D. M., Rombouts, S. A. R. B., Westenberg, P. M., &
Crone, E. A. (2010). Adolescent risky decision-making. Neurocognitive
development of reward and control regions. NeuroImage, 51, 345–355. http://
dx.doi.org/10.1016/j.neuroimage.2010.02.038.