Predictive coding frameworks of perception propose that neural networks form predictions of expected input and generate prediction errors when the external input does not match expectation. We therefore investigated the processing of unexpected sounds and silence in the auditory cortex using fMRI. Unexpected sounds, when compared to expected sounds, evoked greater activation in large areas of the left temporal and insular cortices. Additionally the left middle temporal gyrus exhibited greater activation to unexpected events in general, whether sounds or silence, when compared to the corresponding expected events. These findings support predictive coding models of perception, which suggest that regions of the temporal cortex function to integrate sensory information with predictive signals during auditory perception.
Keywords: Associative learning; Auditory perception; Expectation; Prediction error; Predictive coding.
© 2013 Elsevier Ltd. All rights reserved.