The role of Attentional Binding in Hallucination Proneness

Background: Hallucinatory experiences are increasingly conceptualised along a continuum from non-clinical proneness to clinical presentation. Cognitive accounts suggest that under perceptual uncertainty and high attentional demand, perception relies more heavily on top down expectations, increasing the likelihood of misinterpretation. However, mechanistic evidence in healthy individuals with high hallucination proneness, particularly under tightly controlled visual tasks, remains limited.

Aim: This thesis investigates how hallucination proneness affects early visual processing, attentional binding, and object categorisation as a function of perceptual load, exposure duration, stimulus category (living versus non-living), colour, and occlusion, and whether observed differences reflect impaired perceptual sensitivity or response bias.

Methods: Four web-based experiments were conducted with healthy adults. Hallucination proneness was measured continuously using the Revised Launay–Slade Hallucination Scale. Experiments manipulated visual search difficulty, brief exposure durations (200–400 ms), category and colour in speeded living/non-living categorisation, and graded occlusion of Snodgrass and Vanderwart objects (25–75%). Analyses used linear and generalised mixedeffects models with participant random intercepts and Signal Detection Theory metrics (d′ for sensitivity, criterion c for bias).

Results: Higher hallucination proneness consistently reduced accuracy and sensitivity, with largest deficits under high load, brief exposures, living stimuli, and moderate occlusion (50-75%). Colour improved sensitivity for living items but sometimes slowed responses, indicating additional processing demands. Exposure duration had modest effects on sensitivity. Effects of hallucination proneness were driven primarily by reduced perceptual sensitivity rather than response bias.

Conclusion: Hallucination proneness reflects a quantitative vulnerability in early perceptual encoding and feature integration under uncertainty. Findings support predictive processing accounts, showing that when bottom-up precision is weak, top-down expectations dominate, impairing discrimination without promoting impulsive responding.