Preliminary communication
Iowa gambling task performance in euthymic bipolar I disorder: A meta-analysis and empirical study

https://doi.org/10.1016/j.jad.2012.11.027Get rights and content

Abstract

Background

The Iowa Gambling Task (IGT) has been recommended as an index of reward sensitivity, which is elevated in bipolar disorder. We conducted a meta-analysis of IGT performance in euthymic bipolar I disorder compared with control participants. Findings indicated that people with bipolar disorder make more risky choices than control participants, though the effect is small (g=0.35). It is not clear which of the many processes involved in IGT performance are involved in producing the observed group difference.

Methods

Fifty-five euthymic people with bipolar disorder and 39 control participants completed the IGT. The Expectancy Valence Model was used to examine differences in IGT. We also examined whether variation in IGT performance within the bipolar group was related to current mood, illness course, impulsivity, or demographics.

Results

Bipolar and control groups did not differ on the total number of risky choices, rate of learning, or any of the parameters of the Expectancy Valence Model. IGT performance in bipolar disorder was not related to any of the examined individual differences.

Limitations

It is possible that there are group differences that are too small to detect at our sample size or that are not amenable to study via the Expectancy Valence Model.

Conclusions

We were unable to identify group differences on the IGT or correlates of IGT performance within bipolar disorder. Though the IGT may serve as a useful model for decision-making, its structure may make it unsuitable for behavioral assessment of reward sensitivity independent of punishment sensitivity.

Introduction

Bipolar I disorder is a severe psychiatric condition defined by episodes of mania. The reward system has been a focus of psychological research on mania etiology (Johnson et al., 2012b). Over 20 years ago, Depue and Iacono (1989) argued that mania symptoms are consistent with an overactive pursuit of and response to reward and hypothesized that mania vulnerability involves elevated reward sensitivity—a tendency to show increased behavioral, affective, and cognitive responses to reward-relevant stimuli (Urosevic et al., 2008, Johnson et al., 2012b). Consistent with this, researchers have found that manic symptoms are more likely after life events involving goal attainment (Nusslock et al., 2007, Johnson et al., 2008). Self-reported sensitivity to reward, as measured by the Behavioral Activation System (BAS) scales (Carver and White, 1994), is elevated among people with bipolar disorder (Meyer et al., 2001) and is associated with a more severe course of mania (Meyer et al., 2001), the conversion of bipolar spectrum disorders to more severe forms of disorder (Alloy et al., 2012), and the onset of bipolar spectrum disorders (Alloy et al., 2008). Recent work has attempted to identify differences between those with and without bipolar disorder in responses to reward-relevant laboratory tasks (see Johnson et al., 2012b for review).

Recently, the National Institute of Mental Health (NIMH) RDoC workshop on Positive Valence Systems recommended the Iowa Gambling Task (IGT; Bechara et al., 1994) as a measure of approach motivation (NIMH, 2011). In the IGT, participants must choose cards from four decks. With each card selected, the participant gains money; for some cards, the participant gains money but also has to pay a penalty that may be larger than the amount gained. Two of the decks are “risky,” meaning that they provide larger payoffs but are more likely to lead to large losses, and two of the decks are “safe,” with smaller payoffs but smaller risks. In the long run, participants who play the risky decks will lose money, but participants who play the safe decks will gain money. Successful participants learn to choose cards from the safe decks over the course of the task. There are many ways to interpret IGT performance, but one interpretation is that those who select cards from risky decks are willing to hazard losses in the pursuit of larger rewards—a profile that may stem from greater reward sensitivity.

Several researchers have used the IGT with bipolar samples. Findings suggest that people experiencing acute mania and depression select more cards from risky decks than do control participants (Rubinsztein et al., 2006, Adida et al., 2011). However, reward models posit that people with bipolar disorder display elevated reward sensitivity even during remission. Some studies of IGT behavior among euthymic people with bipolar I disorder have found elevated risk taking among euthymic bipolar I participants (Adida et al., 2011, Malloy-Diniz et al., 2011), and others have found no differences in IGT behavior (Yechiam et al., 2008, Jogia et al., 2011, Martino et al., 2011). Given the mixed results, our first goal was to conduct a meta-analysis of relevant IGT findings.

Section snippets

Study 1

We identified published studies indexed in PubMed or PsycINFO with the key words (IGT or “Iowa Gambling”) and (mania, bipolar, or manic). Studies were excluded from analysis if they did not include a euthymic bipolar I group (Ibanez et al., 2012) and a control group (Christodoulou et al., 2006, Jollant et al., 2007).

Meta-analysis was performed in the metafor package for R (Viechtbauer, 2010) using a random-effects model with restricted maximum likelihood estimation. Fig. 1 shows the effect size

Study 1 discussion

Our meta-analysis raises two related questions. First, what is the source of the group difference? IGT performance rests on several cognitive processes, and analyses of overall IGT performance do not specify which processes underlie group differences. Second, the small effect size and relatively (though not significantly) heterogeneous pattern suggest that IGT performance varies among people with remitted bipolar disorder and that individual differences are important to consider. In study 2, we

Study 2

As noted, the IGT has been described as a measure of reward sensitivity (NIMH, 2011) and of real-life decision-making (Bechara et al., 1994), but IGT performance results from a complex of both “hot” and “cold” cognitive processes (Buelow and Suhr, 2009). Beyond reward sensitivity, potential explanations for poor performance include insensitivity to losses, difficulty learning or remembering the contingencies associated with each deck, random or inconsistent selections, or simple lack of

Participants

Participants were recruited through treatment centers, support groups, and community advertisements in the Miami and Palo Alto areas. Participants in the bipolar group (N=55) met criteria for bipolar I disorder as assessed by the Structured Clinical Interview for DSM-IV (SCID; First et al., 1997). Participants in the control group (N=39) did not meet current or lifetime criteria for any mood disorder, including bipolar I disorder, bipolar II disorder, cyclothymia, bipolar disorder NOS, major

Results

As shown in Table 1, bipolar and control groups did not differ on age, gender, years of education, or current manic or depressive symptoms. The bipolar group had higher rates of lifetime or current anxiety disorder and of lifetime substance-related disorder. The bipolar group reported a relatively severe illness history.

Discussion

Study 1 was a meta-analysis of five studies of IGT performance among euthymic persons with bipolar I disorder compared with control participants. Findings of the meta-analysis suggested an effect size of Hedges’ g=.35, which is relatively small compared with effect sizes associated with other cognitive tests in euthymic bipolar I disorder (Robinson et al., 2006, Bora et al., 2009). There was also marginal evidence of heterogeneity in the effect sizes across studies.

Study 2 examined IGT

Conflict of interest

All authors declare that they have no conflicts of interest.

Role of funding source

Data collection and manuscript preparation were supported by NIMH Grant RO1 076021 to Sheri Johnson. The NIMH was not otherwise involved in any phase of the research process, including study design, data collection, data analysis, manuscript preparation, or manuscript submission.

Acknowledgements

We thank Lori Eisner, Daniel Fulford, Ian Gotlib, Terence Ketter, Christopher Miller, Jennifer Nam, Bailey Smith, and Meggy Wang for their assistance in conducting this study. We thank Anthony Bishara for providing R code for fitting the Expectancy Valence Model.

References (51)

  • L.J. Robinson et al.

    A meta-analysis of cognitive deficits in euthymic patients with bipolar disorder

    Journal of Affective Disorders

    (2006)
  • G.S. Stern et al.

    Skill-set, success outcome, and mania as determinants of the illusion of control

    Journal of Research in Personality

    (1979)
  • J.A. Suhr et al.

    Affect and personality correlates of the Iowa gambling task

    Personality and Individual Differences

    (2007)
  • A.C. Swann et al.

    Impulsivity and phase of illness in bipolar disorder

    Journal of Affective Disorders

    (2003)
  • D.J. Upton et al.

    Propensity for risk taking and trait impulsivity in the Iowa Gambling Task

    Personality and Individual Differences

    (2011)
  • S. Urosevic et al.

    Dysregulation of the behavioral approach system (BAS) in bipolar spectrum disorders: review of theory and evidence

    Clinical Psychology Review

    (2008)
  • R. Wetzels et al.

    Bayesian parameter estimation in the Expectancy Valence model of the Iowa gambling task

    Journal of Mathematical Psychology

    (2010)
  • E. Yechiam et al.

    Decision making in bipolar disorder: a cognitive modeling approach

    Psychiatry Research

    (2008)
  • B. Abler et al.

    Human reward system activation is modulated by a single dose of olanzapine in healthy subjects in an event-related, double-blind, placebo-controlled fMRI study

    Psychopharmacology

    (2007)
  • W.Y. Ahn et al.

    Comparison of decision learning models using the generalization criterion method

    Cognitive Science

    (2008)
  • L.B. Alloy et al.

    Behavioral approach system and behavioral inhibition system sensitivities and bipolar spectrum disorders: prospective prediction of bipolar mood episodes

    Bipolar Disorders

    (2008)
  • L.B. Alloy et al.

    Progression along the bipolar spectrum: a longitudinal study of predictors of conversion from bipolar spectrum conditions to bipolar I and II disorders

    Journal of Abnormal Psychology

    (2012)
  • M.S. Bauer et al.

    Impact of an easy-access VA clinic-based program for patients with bipolar disorder

    Psychiatric Services

    (1997)
  • P. Bech

    The use of rating scales in affective disorders

    European Psychiatric Review

    (2008)
  • P. Bech et al.

    The Bech-Rafaelsen mania scale and the Hamilton depression scale

    Acta Psychiatrica Scandinavica

    (1979)
  • Cited by (26)

    • Decision making and its associations to neurocognitive functions, psychopathology, and the home environment in seven-year-old children at familial high risk of schizophrenia or bipolar disorder: The Danish High Risk and Resilience Study VIA 7

      2021, Journal of Affective Disorders
      Citation Excerpt :

      Traditionally, the ability of real-life decision making has been measured experimentally using the Iowa Gambling Task (IGT) (Bechara et al., 1994) with the notion of somatic markers such as skin conductance and emotions affecting the decision making process in the context of ambiguity (Damasio et al., 1991; Clark and Robbins, 2009). However, despite the presence of decision making deficits (measured with the IGT) in adult individuals with schizophrenia (Sevy et al., 2007) as well as bipolar disorder (Edge et al., 2013), the IGT has been criticized for relying heavily on several neurocognitive functions other than decision making, such as learning and working memory, which complicates the interpretation of the test results (Clark and Robbins, 2009). In an effort to disentangle the various components of the decision making process, the Cambridge Gambling Task (CGT) was designed to estimate decision making separate from risk proneness and inhibitory control functions in the context of uncertainty (Rogers et al., 1999; Clark and Robbins, 2009).

    • Characterizing decision-making and reward processing in bipolar disorder: A cluster analysis

      2018, European Neuropsychopharmacology
      Citation Excerpt :

      Overall, several studies in this field have consistently reported poorer outcomes on both decision-making abilities as well as response to reward in bipolar patients compared to healthy controls (Adida et al., 2008; Brambilla et al., 2013; Malloy-Diniz et al., 2009; Powers et al., 2013; Roiser et al., 2009a; Strakowski et al., 2009). More specifically, different studies confirm that bipolar patients underperform in the IGT independently of mood state, even during euthymia (Adida et al., 2011; Edge et al., 2013), and have shown that increased risk taking goes beyond manic states, being also elevated in patients presenting acute depression (Rubinsztein et al., 2006). Concerning sensitivity to punishment, some studies have reported that bipolar patients, in general terms, are more prone to avoid high-frequency penalties options (Adida et al., 2011; Adida et al., 2015; Powers et al., 2013) even during remission periods.

    • Emotion-related impulsivity and the mood disorders

      2018, Neurobiology of Abnormal Emotion and Motivated Behaviors: Integrating Animal and Human Research
    • Decision making measured by the Iowa Gambling Task in alcohol use disorder and gambling disorder: a systematic review and meta-analysis

      2017, Drug and Alcohol Dependence
      Citation Excerpt :

      The characteristics of impulsivity and decision-making are thoroughly addressed in several mental disorders. Literature up to date is inconsistent in the existence of significant differences in IGT performance between patients with schizophrenia or schizoaffective disorder and matched healthy control groups (Alves and Rozenthal, 2006; Pedersen et al., 2017), while findings indicate that people diagnosed with bipolar disorder tend to make more risky choices than the matched control groups (though the effect is relatively small (g = 0.35)) (Edge et al., 2013). However, the impairment in decision-making is also a characteristic feature of depressive disorders (Must et al., 2013), as it is in addictive disorders.

    • Structural characteristics of the brain reward circuit regions in patients with bipolar I disorder: A voxel-based morphometric study

      2017, Psychiatry Research - Neuroimaging
      Citation Excerpt :

      Recently, a VBM study has shown that individuals who have a smaller volume of gray matter in the DMPFC exhibit more risky behavior when they experience a negative emotional state (Muhlert and Lawrence, 2015). A meta-analysis of Iowa Gambling Task performance found that euthymic patients with BD-I make more risky choices for a reward than controls (Edge et al., 2013). Previous structural MRI studies have reported that patients with BD-I have decreased gray matter volume in the DMPFC (Ha et al., 2009; López-Larson et al., 2002; Lyoo et al., 2004).

    View all citing articles on Scopus
    View full text