Saccade onset, not fixation onset, best explains early responses across the human visual cortex during naturalistic vision

Visual processing has traditionally been investigated using static viewing paradigms, where participants are presented with streams of randomized stimuli. Observations from such experiments have been generalized to naturalistic vision, which is based on active sampling via eye movements. In studies of naturalistic vision, visual processing stages are thought to be initiated at the onset of fixations, equivalent to a stimulus onset. Here we test whether findings from static visual paradigms translate to active, naturalistic vision. Utilizing head-stabilized magnetoencephalography (MEG) and eye tracking data of 5 participants who freely explored thousands of natural images, we show that saccade onset, not fixation onset, explains most variance in latency and amplitude of the early sensory component M100. Source-projected MEG topographies of image and saccade onset were anticorrelated, demonstrating neural dynamics that share similar topographies but produce oppositely oriented fields. Our findings challenge the prevailing approach for studying natural vision and highlight the role of internally generated signals in the dynamics of sensory processing.