台大心理系

There is a Bayesian revolution sweeping cognitive neuroscience and perceptual psychology in the past decade. The predictive coding model of perception proposes that the brain is fundamentally a hypothesis generator, which actively supports perception by constantly attempting to match predictions and sensory inputs as much as possible (i.e., to minimise prediction error in the system). Electrophysiological data further contribute in numerous ways to the dissociation between the processing of predicted and non-predicted stimuli, which are associated with different levels of prediction error. However, the nature of this distinction is not fully understood. I will discuss recent experiments on prediction error processing in audition that aim to illuminate how the brain functions as proactive machine for perceptual activities.