Abstract
In a test of their DMS-model (Difficulty Moulds Search), Young and Hulleman (JEP:HPP in-press) showed that the Functional Visual Field (FVF) in visual search becomes progressively smaller when task difficulty increases from easy (/ amongst |), via medium (T amongst Ls), to hard (notched square amongst rotated notched squares). Here, we report eye movement data illustrating how task difficulty also shapes search strategy. First fixation durations in unrestricted search scaled with task difficulty. They were much longer for easy search (375-476 ms) than for medium (321-367 ms) and difficult search (263-300 ms). A gaze-contingent window experiment confirmed that this was due to FVF-size. Importantly, saccadic amplitude not only depended on task difficulty (cf. Rayner, 2009), but also on target presence and number of preceding saccades. In easy search, saccadic amplitudes for absent trials gradually increased (from 6 degrees for early to 10 degrees for later saccades), but for present trials they progressively decreased (from 6 down to 2 degrees). Though less pronounced, medium difficulty search showed the same pattern (rising from 5 to 8 degrees and falling from 5 to 4 degrees). For difficult search, saccadic amplitude (about 5 degrees) was essentially independent of both target presence and rank number. These results concur with the DMS-model and provide new insights into how task difficulty determines search strategy. In easy search, the large FVF allows most work to get done in a long duration first fixation, resulting in a rapid zooming in target-present trials. When no target is present, ever larger saccades are made to cover the entire display. In very difficult search the small FVF does not permit processing of large parts of the display in the opening fixation, resulting in shortened first fixation durations and subsequent item-by-item fixations without any idea whether the next item is the target.
Meeting abstract presented at VSS 2013