Flexibility, uncertainty and anxiety in autism

Abstract

Autism is a heterogenous neurodevelopmental condition manifesting in complex interrelated features across multiple levels of explanation, from genes to systems-level brain function and all the way up to cognition, behaviour and co-occurring mental health difficulties. A prominent characteristic of autism is the tendency towards inflexible responses to change, which is reflected in the core symptom domain of restricted and repetitive behaviours (RRBs). These flexibility difficulties may also manifest in cognitive tasks involving rule shifts. As environmental conditions change, people experience heightened uncertainty and may struggle to predict future outcomes. Difficulties representing and responding to uncertainty are prevalent in autism as well as anxiety disorders, which co-occur in approximately half of autistic people. In this thesis, I take a top-down approach to investigate these key behavioural and cognitive constructs – flexibility, uncertainty, and anxiety – using advanced analytical methods to embrace the inherent heterogeneity in service of clinically relevant insights. In chapter 1, I introduce the constructs of flexibility, uncertainty and anxiety in autism, as well as data-driven and theory-driven computational psychiatry methods. In chapter 2, I perform a systematic review and meta-analysis of cognitive flexibility and demonstrate that autistic people have greater cognitive flexibility difficulties, yet that this profile is characterised by substantial heterogeneity. In chapter 3, I present a probabilistic reversal learning task with uncertainty captured by a 90:10 reward schedule and show that autistic and neurotypical adults did not significantly differ in task performance. Theory-driven computational modelling suggests that similar cognitive strategies were employed, with both groups showing equivalent learning in the face of uncertainty. In chapter 4, I use structural equation modelling to investigate the complex interrelationships of RRB subtypes, namely insistence on sameness (IS) and repetitive sensory motor behaviours (RSMB), prospective and inhibitory intolerance of uncertainty (IU), anxiety, and depression in autistic adults. The results suggest that there are significant positive associations between all these latent constructs, apart from a direct effect of IS on anxiety, and that both IU subtypes mediate the relationships between IS and RSMB with anxiety. In chapter 5, I take a data-driven computational approach to investigate the existence of RRB and IU subgroups, and potential differences in mental health before and during the COVID-19 pandemic, in a sample of autistic and neurotypical adults. A two-cluster solution was identified, with one cluster predominantly autistic and the other predominantly neurotypical, and the former characterised by significantly higher anxiety and depression, before and during lockdown. However, the data-driven approach was able to capture interindividual variability, with some autistic adults assigned to the predominantly neurotypical cluster, and vice versa, and crucially that their mental health was more closely aligned with the data-driven classification rather than diagnostic status. In chapter 6, I take a transdiagnostic data-driven and theory-driven computational approach, to investigate the existence of anxiety and depression subgroups in a sample of autistic and neurotypical adults with and without co-occurring anxiety. I also present a probabilistic reversal learning task with an 80:20 reward schedule and a card sorting task. Two transdiagnostic subgroups were identified, with this data-driven classification independent of formal diagnostic status. One cluster was characterised by significantly higher anxiety, depression, RRBs and IU, however the two clusters did not significantly differ in performance across the cognitive tasks, or in the learning mechanisms employed. In chapter 7, I discuss the relevance of these findings, integrated within the broader landscape of autism research and address current challenges and future research priorities.