Purchase this article with an account.
Brent R. Beutter, Joseph Toscano, Leland S. Stone; Top-down and bottom-up influences on saccades in a visual search task. Journal of Vision 2005;5(8):449. doi: https://doi.org/10.1167/5.8.449.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We measured top-down and bottom-up influences on human saccadic eye movements by having observers perform a visual search for an oriented target in the presence of an orthogonally oriented distractor. Bottom-up, salience-based mechanisms predict the frequency of saccades both to the target (correct) and to the distractor (incorrect) will increase similarly as their contrasts increases, with correct saccades favored as the distractor contrast decreases. Top down, template-based mechanisms predict that the proportion of correct saccades will increase as target contrast increases and incorrect saccades will decrease as distractor contrast increases. Our experiment measured the relative contribution of these two mechanisms to saccadic targeting.
Two observers performed a 6 AFC (6 locations equally spaced at 6.0 deg eccentricity) search task to find the target (vertical Gabor) and disregard a distractor (horizontal Gabor). The Gabors were both odd-phase (SF: 3.3 c/d, σ: 0.23 deg) and added to white noise (RMS 26%). All combinations of 4 target contrasts (6% - 24%) and 5 distractor contrasts (0% - 24%) were used. The target and distractor locations were independently and randomly chosen. Observers searched the display and at the end of each trial indicated the target location with a mouse. We defined the 1st saccadic 6AFC decision as the element location nearest to the 1st saccade's endpoint.
Despite prior knowledge of the orientation of the target and distractor, observers were unable to fully use the orientation information. The mean (over distractor contrasts) slope of the proportion of saccades to the target vs. target contrast was 2.2. The mean (over target contrasts) slope of the proportion of saccades to the distractor vs. distractor contrast was 2.3. The positive distractor slope and its similarity to the target slope show that, in this task, bottom-up factors dominated with performance largely determined by the contrast difference between target and distractor.
This PDF is available to Subscribers Only