Abstract
In object substitution masking (OSM) a sparse, temporally-trailing four-dot mask impairs target identification, even though it has different contours from and does not spatially overlap with the target (Di Lollo, Enns, & Rensink, 2000; Enns & Di Lollo, 1997). OSM is thought to reflect “perceptual hypothesis testing” whereby iterative re-entrant processing loops are initiated from higher cortical areas to lower ones in an effort to confirm the identity of coarsely coded visual stimulation. Because the target is presented only briefly while the mask remains on the display, this hypothesis testing results in the mask being confirmed as the identity of the stimulus, thus excluding the target from consciousness. Here, we demonstrate a previously unknown characteristic of OSM: at prolonged (e.g., ∼ 600 ms) mask durations, observers show reduced masking relative to intermediate mask durations (e.g., ∼ 250 ms). In our experiments, observers identified the location of the gap (left versus right) in a Landolt C target, which was trailed by a four-dot mask for various durations (Supplementary Figure 1A). Target identification accuracy decreased up to mask durations of 240 ms, but then improved at longer durations (Supplementary Figure 1B). This recovery was obtained across a range of stimulus presentation conditions using both trained and naïve observers. Our findings demonstrate that although initially only one of two spatiotemporally adjacent stimuli presented to the visual system may gain access to consciousness, the “losing” stimulus is not irreversibly lost to awareness.