Purchase this article with an account.
Serena Castellotti, Martin Szinte, Anna Montagnini, Maria Michela Del Viva; Saccades’ trajectories deviate away from different kinds of salient visual features. Journal of Vision 2022;22(14):3874. https://doi.org/10.1167/jov.22.14.3874.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
When executing an eye movement to a specific location, the presence of a competing stimulus can influence the saccade trajectory to the target. Under appropriate spatiotemporal conditions, the overall effect of a distractor onset is to produce a deviation of the saccade trajectory away from its location. This is usually attributed to inhibitory processes in the oculomotor system. Additionally, it has been found that the more salient the distractors (e.g., bigger, brighter) the larger the deviation. Here, we investigate the saliency of visual features (spatial arrangements of black-and-white pixels), predicted by a constrained maximum-entropy model to be optimal or non-optimal information-carriers in fast vision, by using them as distracting stimuli in a saccadic task and measuring the resulting saccadic curvature. Observers were asked to perform a vertical saccade towards a placeholder randomly presented above or below the fixation (7°-eccentricity), while one brief task-irrelevant distractor could randomly appear on the right or on the left (25ms), half-way from the target, with variable delay with respect to target onset (from -150 to +50ms). We compared the saccadic curvature induced by model-predicted optimal and non-optimal features to that induced by high- and low-luminance square distractors. We found that the presentation of optimal features induces a strong deviation away, larger than that induced by non-optimal features. The magnitude of the deviation is maximal when the distractor-to-saccade onset asynchrony is large (DSOA from -300 to -200ms) and decreases as DSOA becomes less negative. Similar effects emerged by comparing the deviation produced by high-luminance vs low-luminance distractors. To conclude, the relative saliency of the features predicted by the constrained maximum-entropy model reflects on the magnitude of the interference with the correct programming of saccades to a target. Also, the saliency determined by the optimally-informative features was comparable to luminance-based saliency.
This PDF is available to Subscribers Only