Defining the temporal threshold for ocular fixation in free-viewing visuocognitive tasks

Abstract

Patterns of ocular fixation in free-viewing tasks reflect aspects of visual attention and cognition. To quantify these patterns, fixations must be identified in raw eye movement data by applying explicit spatial and temporal thresholds. A temporal threshold of 200 ms is commonly used in clinical visuocognitive research, despite having been originally derived from a study of eye movements in reading. We systematically explored temporal fixation thresholds below 200 ms, using biologically relevant (human face) and abstract (complex geometric) stimuli. Short fixations (<200 ms) significantly altered spatiotemporal patterns of fixation for both the face and geometric stimuli, by increasing the number of fixations and the scanpath length. A threshold of 100 ms was found to discriminate fixations from other oculomotor activity effectively, and was consistent with current physiological and visuocognitive models. Statistical examination of fixation duration data for each subject suggested the median was a less biased and more robust measure of central tendency than the mean.