研究セミナー Understanding Sensory Prediction Error: Nature and Neural Mechanisms

The brain operates as a predictive system, using internal models to anticipate sensations and the outcomes of our actions. Discrepancies between expected and actual events—prediction errors—are thought to update these models and direct attention toward unexpected stimuli. Despite their fundamental role in various neural computations, the circuit mechanisms underlying these signals remain poorly understood. In this talk, I will present our recent findings on the nature of sensory prediction error signals and the circuit mechanisms underlying their generation in the mouse primary visual cortex (V1). We find that violations of expectation do not simply encode residual differences between predicted and actual inputs, as proposed by classic predictive coding theories. Instead, they selectively enhance responses in layer 2/3 neurons tuned to the unexpected stimulus. This selective amplification arises from a cooperative interaction between inputs from the pulvinar, a higher-order thalamic nucleus, and cortical vasoactive intestinal peptide (VIP)-expressing interneurons, which disinhibit pyramidal neurons via somatostatin (SOM)-expressing interneurons. Together, these findings reveal that the brain prioritizes unexpected information by increasing the salience of unpredicted features through a synergistic interaction of cortical and subcortical circuits (Furutachi et al., 2024; Furutachi & Hofer, 2026). I will also discuss ongoing work on the functional roles of these error signals in updating internal models, and their implications for understanding autism spectrum disorder (ASD).