Abstract
Active visual tasks require (1) evaluating the content of displays based on information acquired during fixation pauses, and (2) choosing where to aim saccades. How do we apportion time and resources between these two operations (e.g., Hooge & Erkelens, 1999; Araujo et al., 2001)? We studied saccadic eye movements during a statistical estimation task in a display containing both target and distractor elements. Displays were made up of 25 circles (41′ diam), 12 targets, 13 distractors. Targets, drawn with thin (4′) outlines, contained a tilted line whose orientation was drawn from a distribution with mean 10 or 20 deg to the left or right of vertical (SD = 25 deg). Distractor circles had wider outlines (5′–6′) and contained vertical line segments. The task was to determine the mean tilt (left or right) of the parent distribution of the target line segments. In the absence of distractors, 4.5–6 target circles/2s trial were fixated, with fixation pause durations of 155–220 ms. With distractors present, fixation durations were about the same, the number of targets fixated/trial decreased, distractors were fixated frequently, and the accuracy of perceptual judgments declined. The ability to avoid distractors (without changing fixation pause duration) was somewhat better when the discriminability of the distractors from the targets increased. In separate “look only” conditions, fewer distractors were fixated, but only by increasing fixation pause duration up to 300 ms. These results show a reluctance to increase fixation duration for the purpose of improving target selection, even when such strategies would improve task performance. Preferences favor a high rate of saccadic production (Wu et al., Vision Res. 2010). This strategy, which leads to many useless fixations, is costly in terms of time and performance accuracy, but may reduce the overall cognitive load of the task.