Sequential prioritization of expected and unexpected information in visual processing

This electroencephalography (EEG) study provides evidence that expectations dynamically shape visually evoked brain responses through a biphasic temporal mechanism: an earlier enhancement of expected image components, followed by a later prioritization of unexpected components. Across six sessions each, six participants were presented with a deterministic sequence of four images with intermixed composite trials that contained both the expected and the unexpected image. To quantify expected and unexpected identity information over time, we applied Linear Discriminant Analysis. The observed temporal dissociation suggests that the brain initially integrates sensory input with prior expectations in a Bayesian manner before shifting to emphasize surprising information, thereby supporting both efficient recognition and adaptive model updating. Furthermore, expectations affected superordinate image processing (animal vs. non-animal), showing the same pattern as identity processing, with earlier enhancement of expected information followed by a prioritization of surprising information.