Autistic Adults Avoid Unpredictability in Decision-Making

Abstract

Decision-making under unpredictable conditions can cause discomfort in autistic persons due to their preference for predictability. Decision-making impairments might furthermore be associated with a dysregulation of sex and stress hormones. This prospective, cross-sectional study investigated decision-making in 32 autistic participants (AP, 14 female) and 31 non-autistic participants (NAP, 20 female) aged 18-64 years. The Iowa Gambling Task (IGT) and the Cambridge Risk Task (CRT) were used to assess decision-making under ambiguity and under risk with known outcome probabilities, respectively. Cortisol, estradiol, and testosterone serum levels were related to decision-making performance. Groups did not differ in overall IGT and CRT performance, but compared with NAP, AP preferred less profitable card decks with predictable outcomes while avoiding those with unpredictable outcomes. AP required more time to reach decisions compared to NAP. Additionally, AP without comorbid depression performed significantly worse than NAP in the IGT. Estradiol and cortisol concentrations were significant predictors of CRT scores in NAP, but not in AP. The study results imply that AP are 'risk-averse' in decision-making under ambiguity as they avoided choice options with unpredictable losses in comparison to NAP. Our findings highlight the intolerance for uncertainty, particularly in ambiguous situations. Thus, we recommend being as transparent and precise as possible when interacting with autistic individuals. Future research should explore decision-making in social situations among individuals with ASD, factoring in person-dependent variables such as depression.

Keywords: Autism spectrum; Cambridge risk task; Decision-making; Hormones; Iowa gambling task; Trail making test.

Fig. 1

The mean IGT net score of NAP and AP. Note. The IGT net score (advantageous decks – disadvantageous decks) (y-axis) for each block of the IGT consisting of 20 trials (x-axis) with (Figure A, N = 62) and without the influential case (Figure B, N = 61). Error bars indicate standard errors

Figs. 2

The number of selected card decks in the IGT of NAP and AP (a-d). Note. The number of selected card decks in the IGT for deck A (“bad deck”, infrequent gains), deck B (“bad deck”, frequent gains), deck C (“good deck”, infrequent gains) and deck D (“good deck”, frequent gains) of NAP (y-axis). The number of selected card decks is presented for each block of the IGT consisting of 20 trials (x-axis). Error bars indicate standard deviations

Fig. 3

Low risk choices (A) and deliberation time (B) of the CRT Note Note that the equal risk ratio is not meaningful for the number of low risk choices. Error bars indicate standard errors