Subjective self-control and behavioural impulsivity coexist in

anorexia nervosa

G.K.L. Butler*, A.M.J. Montgomery

Department of Psychology, University of Greenwich, Eltham, London SE9 2UG, UK

Received 23 April 2004; received in revised form 5 November 2004; accepted 29 November 2004

Abstract

behavioural measures (a continuous performance task [CPT]; and a novel risk taking measure [Bets 16]) of

impulsivity were used along with the Beck Depression Inventory (BDI).

scores, but they also displayed impulsive behaviour on the CPT (more errors of commission with faster reaction

times).

relationship between impulsivity and disordered eating in AN is more complex than previously recognised and

supports the view that self-awareness in AN is low.

doi:10.1016/j.eatbeh.2004.11.002

* Corresponding author. Present address: School of Social Science, Kingston University, Penrhyn Road, Kingston Upon

Thames, Surrey, KT1 2EE, UK. Tel.: +44 20 8547 2000; fax: +44 20 8547 7388.

Eating Behaviors 6 (2005) 221–227

1. Introduction

and bulimia nervosa (BN) may be at opposing ends of an impulsive/control continuum: they reported

lower scores in an AN group for impulsivity and danger seeking and suggested that traits associated

with AN reflect accentuated self-control, caution, and conscientiousness. This view is supported by

and affective instability) which are opposite to those of AN (rigidity and constraint).

Empirical studies of impulsivity have used both self-report questionnaires and more objective

behavioural measures (e.g., reaction time tasks), as indices of different aspects of the multi-dimensional

investigations of impulsivity in disordered eating have mostly used self-report measures. Using the selfreport

than a BN group, but did not differ from controls on either measure. However, using a continuous

responded to more non-targets (errors of commission; indicating impulsivity) and missed more target

stimuli (errors of omission; indicating inattention) than controls, without any differences in response

latencies.

Overall, self-report measures of impulsivity indicate that those with restricting AN are more selfcontrolled

than those with BN. However, it is uncertain whether they also differ from controls, or

whether differences are limited to self-report measures, thereby indicating reduced self-awareness in AN

objective behavioural measures. The behavioural measures were a CPT, which assesses reaction time

and accuracy, and a risk-taking measure (Bets 16). The Bets 16 task does not depend on response speed,

thus avoiding potential difficulties of interpretation caused by psychomotor retardation associated with

Based on research using self-report measures, it was predicted that women with AN will report higher

without changes in reaction times.

2. Methods

2.1. Participants

S.D.=8.3), from a university in Southeast England, who were screened for abnormal eating attitudes and

2.2. Apparatus

The continuous performance task was presented on and data were recorded on a Phoenix NoteBIOS

4.0 multimedia notebook.

2.3. Measures

Is a 54-item questionnaire consisting of three scales, impulsiveness, venturesomeness, and empathy

2.3.2. Beck Depression Inventory

depression.

2.3.3. The Eating Attitudes Test

detecting previously undiagnosed cases of AN. A cut-off score of 20 successfully discriminated 84% of

2.3.4. The Bulimic Inventory Edinburgh

assess bulimic behaviours. Scores of 0–10 fall within the normal range. A score of 5 or greater on the

severity scale is considered clinically significant.

2.3.5. Bets 16

This risk-taking measure consists of 16 pairs of two-outcome hypothetical bets presented in pie chart

format, where paired bets have the same expected value (EV). For each pair of bets participants choose

between a guaranteed win (GW, the safe choice), and a long shot (LS, the risky choice). The GWoffers a

larger chance of winning a small amount of money against a less likely win of a moderate amount of

money. The LS offers a 10–30% probability of a proportionately larger win or a more likely win of

Bets 16 exhibits good internal consistency (.92), moderate convergent validity, and moderate test–retest

coefficient over 3 weeks (.77) (Montgomery, in preparation).

2.3.6. Continuous Performance Task

This is a DOS-based computer task which presents grey letters (approximately 1 in. in size) one at a

time onto a black background for 200 ms with an inter-stimulus interval of 1.5 s. The AX version was

used where, in target trials, participants respond to the letter X, but only if preceded by the letter A. On

other trials, participants should inhibit responses. Reaction times, errors of omission (failure to respond

on a target trial), and errors of commission (responding when the target letters were not presented—nontarget

trial) were recorded. Errors of omission reflect inattention and errors of commission reflect

2.4. Procedure

The AN group was tested individually in a single session (~1 h) at the eating disorder unit. The

control group was tested individually at their university over two sessions lasting 1–1.5 h overall. Their

tasks included the EAT-26 and BITE. The order of task completion was randomised for each participant.

2.5. Statistical analysis

Product Moment Correlation or Spearman’s Rho. Between group differences were assessed using onetailed

tests for self-report measures and the CPT, but two-tailed test were used for correlations and Bets

16 data.

3. Results

3.1. Group characteristics and between group differences

At testing, the time since admission for the clinical group ranged from 1 to 6 weeks and records

revealed that one was taking chlorpromazine and another metazepam. BDI scores were missing for four

AN group made CPT errors of commission, but overall the AN group made more CPT errors of

Re-analysis of the data omitting the two medicated patients did not alter the pattern of significant

effects or produce any further significant differences.

3.2. Correlational analysis

For the control group CPT hit reaction time correlated negatively with CPT errors of commission

4. Discussion

The present study investigated impulsivity in an inpatient group with anorexia nervosa and a control

group. Body mass index (BMI) scores were unavailable for the inpatient group. However, as ICD-10

all had lower BMI scores than those of the control group (average BMI 22.75).

The results from the self-report measures in the present study are in general agreement with earlier

comparisons with a control group rather than a BN group. Thus, the present AN group regarded

themselves as less likely than controls to engage in either risk-taking regardless of consequences

(impulsiveness) or sensation seeking where potential risks are considered (venturesomeness). This

contrasts with the findings from the behavioural measures. Although the AN group did not show an

altered preference for risky bets, they did make more CPT errors of commission, and these occurred with

errors of commission among an AN group, but they also found increased errors of omission and no

Table 1

16, and the CPT

Mean S.D. Mean S.D.

Impulsiveness 10.50 4.87 7.07* 3.49

Venturesomeness 8.50 3.74 5.33* 3.04

Empathy 14.19 2.74 15.40 2.61

Bets 16 5.56 5.73 3.73 4.54

Continuous performance test (CPT)

Omission errors 0.28 0.50 1.93 2.50

Commission errors 0.94 1.80 1.63* 3.50

Hit reaction time 432.60 115.60 400.30 65.70

BDI=Beck Depression Inventory.

CPT reaction times (RT) are in milliseconds.

inattention, as they could result from a failure to attend to the stimulus which preceded the X stimulus.

However, if the present increase in errors of commission resulted from inattention we should also expect

to see a similar increase in errors of omission, but we did not. It might be argued that the present

increased errors of commission were a consequence of the AN group having faster reaction times.

BDI scores, they were not exhibiting psychomotor retardation.

and was clearly sufficient to detect between groups differences on both self-report and behavioural

measures; however, a larger sample might also detect differences on the Bets 16 measure.

The results of the correlational analysis were in accord with previous studies reporting modest or

positively correlated with Bets 16 scores, as would be expected if both tasks measure risk taking where

the consequences of behaviour are considered. Two significant correlations in the AN group featured

CPT measures: CPT errors of omission correlated with Bets 16 scores, so those who made more risky

choices were also more inattentive; and CPT errors of commission reaction time correlated with

venturesomeness, perhaps surprisingly indicating that those who were more risk taking were slower to

respond to non-targets.

The reduced impulsiveness and venturesomeness reported by the AN group are in opposition to their

results on the CPT where their responding was fast and inaccurate or impulsive. This apparent conflict

may derive from the nature of self-report measures, which measure how participants view themselves,

whereas objective measures reflect behaviour. Indeed, this apparent lack of insight might contribute to

disordered eating is associated with reduced self-awareness. Alternatively, these apparently contradictory

Overall the present results indicate that alterations in impulsivity associated with AN are clearly more

complex than suggested by the results of self-report measures alone. Consequently it seems appropriate

to conclude that results from the self-report measure are at odds with the behavioural measures and

probably reflect either a lack of self-awareness, or a desire for self-control that is not evident in such

tests. If so, then it might be more appropriate to think in terms of profiles of impulsivity/self-control in

such groups.

Research into the time-course of AN has indicated that although a small proportion of patients

successfully manage to restrict their eating over protracted periods, for many the hypercontrol

of anorexia and then to BN. This suggests that levels of self-control and impulsivity can change over

time and this might have important implications for treatment strategies. Monitoring of changes to the

impulsivity profile over time might provide a basis for clinical interventions aimed at (a) addressing

self-control/impulsivity issues without necessarily having to directly confront eating behaviour, thus

possibly avoiding treatment resistance and (b) interrupting the transition from anorexia to bulimia and

the associated deterioration in prognosis.

Acknowledgements

GKLB was funded by a PhD bursary from the University of Greenwich, UK.

References

Casper, R. C., Hedeker, D., & McClough, J. F. (1992). Personality dimensions in eating disorders and their relevance for

Claes, L., Vandereycken, W., & Vertommen, H. (2002). Impulsive and compulsive traits in eating disordered patients compared

systems.

Eysenck, S. B. G., Pearson, P. R., Easting, G., & Allsopp, J. F. (1985). Age norms for impulsiveness, venturesomeness and

Garner, D. M., Olmsted, M. P., Bohr, Y., & Garfinkel, P. E. (1982). The eating attitudes test: Psychometric features and clinical

Gerbing, D. W., Ahadi, S. A., & Patton, J. H. (1987). Toward a conceptualisation of impulsivity: Components across the

Halperin, J. M., Wolf, L. E., Greenblatt, E. R., & Young, J. G. (1991). Subtype analysis of commission errors on the continuous

18–24.

Kagan, J., Rosman, B. L., Day, D., Albert, J., & Phillips, W. (1964). Information processing in the child: Significance of

Malle, B. F., & Neubauer, A. C. (1991). Impulsivity, reflection, and questionnaire response latencies: No evidence for a broad

Seed, J. A., Dixon, R. A., McCluskey, S. E., & Young, A. H. (2000). Basal activity of the hypothalamic–pituitary–adrenal axis