Neural transformation from image realism to perceived realness

Humans readily perceive whether a visual stimulus depicts something that could exist in the real world, yet the neural basis of perceived realness remains unclear. Because differences between real and unreal stimuli are ultimately expressed through visual features, realness might be expected to emerge along the ventral visual stream. Contrary to this prediction, we show that perceived-realness representations emerge in higher dorsal visual areas through a transformation of image-realism-related image-feature representations in early visual and early dorsal visual areas. fMRI revealed this representational shift toward intraparietal sulcus (IPS), with IPS patterns predicting perceived realness beyond image features and decision difficulty; source-localized MEG showed that realness coding in IPS emerged later than feature coding in early visual and dorsal visual areas. Together, these findings reveal a transformation from visual features to perceived realness along the dorsal visual stream. This representation may help guide subsequent actions and interactions with the real world.