Abstract
Eye movements are among the best measures of ongoing cognitive processing. The principles governing the cognitive control of fixation durations, however, are unclear. While the direct-control hypothesis assumes that fixation durations reflect processing demands at the fixation location, the indirect-control hypothesis assumes that fixation durations predominantly reflect current task demands and are unaffected by the processing load at the current fixation. We studied the adjustment of fixation durations to stepwise changes in processing difficulty. Participants searched for a target symbol in an array of stimulus elements. The target symbol was a closed ring, while distractors consisted of rings with a small gap. We manipulated processing demands by varying the similarity between target and distractor stimuli. Difficulty was controlled by gap size. In a first experiment, we prevented preview of the next symbol by gaze-contingent masking. During each fixation, the fixated symbol was shown while the remaining symbols were masked by Xs. Participants searched the array sequentially from left to right. There was no response to decreasing processing difficulty on the first symbol. For increasing processing difficulty, we observed an immediate but disproportionate prolongation of fixation durations on the first symbol. In both conditions, fixation durations on the second symbol matched the new processing needs. In a second experiment, we permitted preview of the entire stimulus. We replicated a delayed adjustment to easy symbols and an immediate prolongation on difficult symbols. Our results demonstrate that (i) direct control of fixation durations is limited to the situation of a sharp increase in foveal processing difficulty and that (ii) direct control can only prolong fixation durations. Moreover, there is an (iii) indirect control mechanism for the shortening of fixation durations, which unfolds over subsequent fixations following a change in processing difficulty.