Author: Robyn da Silva
da Silva, Robyn, 2025 Reinforcement sensitivity and learning: exploring processes underlying the relationship between autistic traits and social anxiety, Flinders University, College of Education, Psychology and Social Work
Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact copyright@flinders.edu.au with the details.
Autism is a neurodevelopmental disorder characterised by social difficulties and restricted, repetitive patterns of behaviour. Differences in sensitivity to reward and punishment, and cognitive reinforcement learning, i.e., decision-making based on outcomes of past experiences, have been associated with autism and autistic traits. Social anxiety, which frequently co-occurs with autism, has been linked to similar patterns of reinforcement sensitivity and learning. This thesis aimed to consolidate and improve current understanding of reinforcement learning differences in autism and investigate mechanisms underlying the co-occurrence of autistic traits and social anxiety in the general population, focusing on reinforcement sensitivity, reinforcement learning, and estimates of basal ganglia (BG) pathway functioning.
Chapter 2 reviewed research on autism, autistic traits, and cognitive reinforcement learning tasks. Meta-analyses of widely used tasks demonstrated that autistic participants made more errors, and perseverative errors, and made more lose shift and fewer win-stay choices on the Probabilistic Selection Task, linked to BG pathway functioning. Narrative synthesis revealed mixed findings regarding autistic traits. This review identified methodological inconsistencies across tasks, highlighting the need for standardisation and validation, and sufficiently sized samples.
Given the frequent co-occurrence of social anxiety and autism, Chapter 3 investigated the relationship between autistic traits, social anxiety, and sensitivity to reward and punishment. Autistic traits partially mediated the association between sensitivity to punishment and social anxiety and fully mediated the relationship between sensitivity to reward and social anxiety. However, the capture of sensitivity to reward and punishment was reliant on subjective self-report measures, necessitating further investigation leveraging objective cognitive reinforcement learning measures.
Sensitivity to reward and punishment have been associated with similar neurological regions as reinforcement learning. However, whether self-reported reinforcement sensitivity predicts learning from reinforcement has not been established. Chapter 4 explored the relationship between self-reported sensitivity to reward and punishment and cognitive reinforcement learning tasks predictive of BG pathway functioning. Principal components analysis was used to combine variables from cognitive tasks into separate estimates of BG direct and indirect pathway activity. Self-reported sensitivity to reward and punishment did not predict reinforcement learning or BG pathway activity, indicating that the self-report and cognitive measures may capture distinct constructs.
Autistic traits and social anxiety may be linked by similarities in reinforcement learning, but this has not been previously examined. Chapter 5 investigated whether BG estimates link autistic traits and social anxiety. Autistic traits, preference for routine, social difficulties, and social anxiety were examined in relation to BG pathway estimates and a pathway difference score (direct minus indirect). Autistic traits and preference for routine predicted the BG difference score, indicating an imbalance favouring direct pathway activity. However, the BG pathway difference score did not mediate the relationship between autistic traits and social anxiety, indicating other neurological pathways may drive this link.
Overall, this thesis enhanced understanding of the co-occurrence between autistic traits and social anxiety, identifying sensitivity to punishment as a potential clinical target. Additionally, the association between the BG pathway imbalance and autistic traits highlights the need for computational, neurological and genomic research to further develop and validate these findings.
Keywords: autism, autistic traits, reinforcement learning, social anxiety, reward and punishment, learning, cognition, basal ganglia
Subject: Psychology thesis
Thesis type: Doctor of Philosophy
Completed: 2025
School: College of Education, Psychology and Social Work
Supervisor: Sarah Cohen-Woods