Behavioral and Somatic Responses to Decision Making in Autism Spectrum Disorders: Evidence From the Iowa Gambling Test

Background:   There are ongoing questions regarding the relative contributions of cognition and emotion to decision-making in the autism spectrum disorders. In the context of previous work regarding possible disconnects between behavior and physiological response in autism--such as increased anxiety associated with increased risk-taking (South et al., 2011) or social anxiety atypically mediating emotional perception (Kleinhans et al., 2010)—we aimed to investigate autonomic arousal in ASD and matched controls during performance on the Iowa Gambling Task.

Objectives:   To characterize potential differences in the cognitive and somatic strategies used in ASD versus typical samples. Such characterization may provide insight into the developmental course of autism as well as improve the specificity of intervention techniques.

Methods:   Participants included 36 older children and adolescents diagnosed with an ASD and 30 typical controls. Groups were matched on both age (ages 11-16, mean =14.6) and IQ (mean=108.3). The Iowa Gambling Task is a behavioral task that models real-life decision-making in response to uncertainty, rewards, and punishments (Bechara et al., 1994). During each trial participants chose from one of four simulated card decks on a computer screen: each card comes with a reward, while some cards in each are also followed by a loss. Two “advantageous” decks come with small rewards but also small losses and lead to an overall profit, while the two “disadvantageous” decks have large rewards but larger losses, leading to an overall net loss. Disposable electrodes were used to collect SCR for the duration of the experiment.

Results:   We first analyzed the number of choices from the advantageous deck in each block of 20 cards. A 2 (diagnostic group) x 5 (trial block) repeated measures ANOVA showed an expected significant main effect for trial block, with an increase over time in the number of cards chosen from the advantageous decks. There was no main effect for diagnostic group; however, the ASD group learned more quickly and reliably to choose from the advantageous decks, leading to a significant group x trial block interaction. Consistent with the existing literature for this age range neither the ASD nor CON groups showed significant differences in SCR for anticipation of good vs. bad decks, or response to loss vs. no-loss trials. 

Conclusions: The ASD group performed the task very similar to healthy adults, while the worse performance by the CON group looked like studies of typical adolescents who make riskier choices on the IGT. We speculate that the ASD group showed a stronger reliance on cognitive rather than emotional components during decision-making. While sometimes advantageous, this dissociation of cognition and emotion may not helpful in some situations, especially during social interactions that are the hallmark difficulty of autism.