The Blanking Effect on Detecting Changes in Natural Scenes across Saccades

Every time we make a saccade, the image projected onto our retina drastically changes, yet our perceptual experience of the environment remains remarkably stable. Previous studies asking subjects to detect changes across saccades found that we are surprisingly poor at detecting trans-saccadic changes. However, performance markedly improved in the presence of a post-saccadic blank. This blanking effect suggests that the visual system assumes stability across saccades and accesses trans-saccadic memory only when there is extreme evidence (e.g., a blank) against stability. We investigated whether the stability mechanisms evidenced by the blanking effect are employed when attempting to detect trans-saccadic changes in natural scene images, and which signals play a role in triggering this effect. Subjects viewed two scene images presented successively while making a saccade, and performed a scene change detection task. In half of the trials, a 200 ms blank screen was displayed between the presentation of the two scenes. We systematically manipulated the degree of difference between scene images using a continuous scene wheel (Son et al., 2021), and the probability of “changed” responses as a function of scene difference was fitted to psychometric functions. In the saccade condition, we found a significantly heightened sensitivity to detect scene changes (a lower threshold) and a stronger bias to report “changed” (a higher guess rate) in the presence of a post-saccadic blank, indicating that trans-saccadic vision maintains a stable perception of complex natural scenes by assuming the visual world as stable in the absence of extreme evidence against stability. In contrast, an analogous blank had no such effect in a simulated-saccade condition where subjects remained fixated while the scene image shifted spatially to mimic a saccade-induced retinal shift, underscoring a critical role of extra-retinal signals in triggering this stability mechanism.