Oculomotor freezing reveals perceptual priority during free-viewing

Perceptual dynamics around saccadic eye movements are typically probed via local discrimination targets presented at specific locations around the time of a saccade. This approach requires the observer’s explicit judgement, usually via a key press at the end of a trial, which introduces an artificial perceptual task and necessarily interrupts the natural flow of eye movements. Building on the well-established finding that the mere appearance of a stimulus rapidly and automatically inhibits saccades when it reaches awareness (termed “oculomotor freezing”; White & Rolfs, 2016), we present an approach to infer perceptual priority across the visual field without disrupting active visual orienting by a secondary perceptual task. Using their gaze, observers rapidly scanned a field composed of 19 concentrically arranged tiles (r=~14°) to find a hidden search target. Certain tile colors were associated with higher or lower target likelihood. During their exploration, we repeatedly (~1Hz, jittered) flashed brief oculomotor probes (8.3ms) and selectively manipulated their contrast and distance relative to observers’ current gaze position. Flash presentation reliably induced oculomotor freezing during free gaze exploration, as reflected by a depression in saccadic frequencies ~100ms after flash presentation. Inhibition strength increased with flash contrast and decreased with flash distance from the current gaze position. Crucially, freezing was more pronounced for flashes presented at behaviorally relevant locations, such as the upcoming fixation location (saccade goal) or likely search target tiles. This indicates that implicit variations in saccade characteristics elicited by spatiotemporally specific probes directly reflect perceptual priority throughout the scene. By overcoming the need to ask participants for explicit perceptual judgements, our approach provides a versatile tool for testing continuous, uninterrupted perception during natural vision. This should prove particularly beneficial for testing populations that may struggle to learn or adhere to explicit task instructions, including children, patients, or non-human primates.