Abstract
It is well known that the metrical properties of saccadic eye movements are strongly influenced by the extraction of low-level visual features (e.g. luminance). Higher-level, visual-cortical features (e.g. orientation) may also contribute, but it remains undetermined whether they intervene automatically, irrespective of their relevance for the task. Here, we investigated this issue in a saccade-target paradigm. We manipulated the uncertainty of target location in order to either force the extraction of orientation (maximal uncertainty) or make it unnecessary (minimal uncertainty). On each trial, the target was displayed with or without an additional, distractor stimulus on a textured background (composed of oriented-Gabor patches). The target was a vertically-elongated area, which differed from the background by its luminance, while the distractor differed from the background by its luminance, its orientation or both. In one set of blocks, the target, displayed at an eccentricity of 6.5 or 8.6°, was always 3.7° more eccentric than the distractor (blocked; minimal uncertainty). In another set, the target was 3.7° more or less eccentric than the distractor (mixed; maximal uncertainty), appearing at eccentricities of 6.5 and 8.6° or 2.8 and 4.9° respectively. Seven volunteers participated in the experiment. Results showed that irrespective of the uncertainty of target location, all types of distractors deviated the eyes away from the target, thus suggesting that both luminance and orientation were extracted even when they were not relevant for the task (i.e. the minimal-uncertainty condition). Still, the deviation induced by either luminance- or orientation-based distractors was much greater under maximal uncertainty, and often favored the execution of erroneous saccades towards the distractor. Thus, both retinal and visual-cortical features contribute to determine saccade metrics in a rather automatic, bottom-up manner, but their effect is reinforced and qualitatively different as a result of top-down processes.
Meeting abstract presented at VSS 2013