Dynamic Neural Encoding of Prior Expectations and Auditory Evidence Drive Perceptual Decision-Making

Abstract

Perceptual decision-making has been studied extensively as the integration of evidence over time. Yet how moment-to-moment fidelity of sensory encoding interacts with prior expectations in shaping our choices remains poorly understood--particularly in the auditory domain. Here we quantify how priors act at the stage of sensory encoding and how these fluctuations predict behaviour. In a human model-based EEG study, we augmented a canonical click-train evidence accumulation task with probabilistic cues to shape prior expectations and analysed all activity with trial-resolved linear encoding models. Probabilistic cues drove alpha-oscillatory EEG patterns and induced symmetric shifts in response bias but did not change temporal integration dynamics. Neural encoding strength in lateralised auditory cortex uniquely predicted single-trial choice and confidence beyond what was explained by evidence strength or cueing alone. These findings identify sensory encoding fidelity as a mechanistic, behaviourally relevant link between priors and perceptual choice, offering new understanding how neural noise and expectations jointly shape decisions under uncertainty.