Neural signals for expectation and surprise in human visual cortex. (360G-Wellcome-092646_Z_10_Z)

Remarkably little is known about how (i) neural responses generated by the expectation of a percept, and (ii) the surprise responses that occur when that expectation is violated, contribute to visual perception. Here, we use a combination of computational modeling techniques and functional brain imaging to test the view that expectations are 'subtracted' from visual inputs, such that bottom-up, evoked responses correspond to 'prediction error' signals. Our approach to this question includes (i ) assessing the mean impact of expectation and suprise on fMRI signals in the extrastriate visual cortex, and asking whether they activate distinct patterns of voxels; (ii) testing whether surprise responses are sensitive to information about the statistical patterns of information in the environment (e.g. volatility, or entropy), to demonstrate that visual surprise reflects more than a release from neuronal fatigue; and (iii) measuring surprise responses to percepts rendered undetectable by mas king to test whether they depends on conscious attention or updating mechanisms.