The time-course of visual masking effects on saccadic responses indicates that masking interferes with reentrant processing

Object substitution masking (OSM) occurs when a briefly presented target in a search array is surrounded by small dots that remain visible after the target disappears. Here, we tested the widespread assumption that OSM selectively impairs reentrant processing. If OSM interferes selectively with reentrant processing, then the first feedforward sweep should be left relatively intact. Using a standard OSM paradigm in combination with a saccadic choice task, giving access to an early phase of visual processing (the fastest saccades occurring only 100 ms after target onset), we compared the masking time-course of OSM, noise backward masking, as well as a simple target contrast decrease. Consistently with a reentrant account, a significantly stronger masking effect was observed for slow (larger than median RT; average median RT = 177 ms) relatively to fast saccades in the OSM condition. Interestingly, the same result was observed using backward masking. In a follow-up experiment, where we assessed observer’s visual awareness using single-trial visibility ratings, we demonstrated that these ultra-fast responses were actually linked to subsequent reported visibility. Taken together, these results suggest that OSM is indeed interfering specifically with reentrant processing during object recognition, which is consistent with traditional accounts of the OSM effect.