Abstract
Restricted and repetitive patterns of behaviours, interests, or activities are a critical diagnostic criterion for autism spectrum disorder (ASD). Previous studies using gambling paradigms with ASD populations have identified that, unlike typically developed control participants, people with a diagnosis of ASD tend to maintain particular response patterns regardless of the magnitude of potential outcomes to uncertain gains or losses. Here we designed a gambling test that permitted calculation of the response consistency in gambling choices in situations that presented varying expected outcomes in terms of gains or losses. The task was administered to 33 adults with a diagnosis of ASDs and compared to a group of 47 typically-developed (TD) control participants who were matched for age and IQ. When presented with choices where participants could either make a risky gamble or a safe choice in terms of gains or losses (e.g., 20% chance of winning £5 vs. 100% chance of winning £1), the ASD participants did not differ from the TDs in their overall risk-taking behaviour. However, they were more consistent in their individual choices from trial to trial. Furthermore, the proportion of participants who either implemented an invariate response strategy (e.g., either always choosing the most risky or most "safe" option) was significantly higher in the ASD group compared with the controls. Additionally, while the ASD group were slower to make their responses in the win frame and the first half of the lose frame, by the end of the task their decision times were the same as the TD controls. These findings suggest that the ASD tendency towards repetitive behaviour may demonstrate itself even in high-level decision-making tasks, which needs to be understood if we are to be sure what such tasks are measuring.