Neural and Behavioural Correlates of Variance of Sensory Evidence

Abstract

Neurobiology of perceptual decisions has largely focused on the neural correlates of the mean strength of sensory evidence. Much less is known about the neural coding of sensory variability. Here, we analyzed the EEG signals obtained from participants who judged the mean orientation of a sequence of gratings with varying variance but constant mean to identify the neural signatures of sensory variance and their relation to individual differences in choice confidence. The neural responses in the stimulus-entrained (4 Hz) and alpha (9-11 Hz) bands tracked variability independently of mean. The frontal and centro-parietal regions demonstrated a quadratic relationship (i.e., strongest responses to intermediate levels of uncertainty) to the standard deviation of the sequence. The occipital response coded the visual stimulus variability linearly. These neural markers of variability were correlated with inter-individual differences in computational components of metacognition. Centro-parietal activity was most predictive of metacognitive sensitivity, aligning with its known role in evidence accumulation. These findings advance our understanding of how the brain dynamically encodes uncertainty and help better characterise the electrophysiological basis of individual differences in metacognitive evaluation.