The psychophysics of surprise, and a predictive coding approach to understanding intolerance of uncertainty

Abstract

This thesis investigates the mechanisms and computational underpinnings of how people respond to uncertainty, with a specific focus on intolerance of uncertainty (IU), a trait characterised by the perception of uncertain situations as threatening. Given that minimising uncertainty is a fundamental aspect of perception and behaviour, this research explores IU through the framework of Bayesian inference, whereby the brain generates predictions to interpret sensory input. Across seven experiments, behavioural responses and physiological measures (EEG and pupillometry) were used to investigate responses to uncertainty elicited through auditory stimuli. Key findings revealed that increasing the precision of sound sequences does not necessarily make otherwise equivalent sensory changes more surprising or easier to detect, and in some cases has the opposite effect. Additionally, under uncertain conditions with potential aversive outcomes, individuals with high IU anticipate upcoming stimuli to be more aversive, and also perceived aversive stimuli as more aversive than people with low IU. Unexpectedly, under particular conditions of uncertainty, people with high IU gave more nuanced and appropriate responses to the cues and stimuli presented during the experiments, and demonstrated more precise and accurate temporal expectations in their physiological responses than people with low IU. From these findings, a pattern of results emerged that suggest that people with high IU experience uncertainty with greater sensitivity, which might lead to both advantageous and deleterious consequences. I conclude by questioning whether intolerance of uncertainty could be reframed as ‘sensitivity to uncertainty’, which would reflect a more balanced view of how these individuals respond to uncertainty, thereby potentially reducing stigma, highlighting the adaptive potential, and recognising associated advantages.