Abstract
Recent eye movement timing models provide evidence for a two-component model: an autonomous stochastic timer that is optimally fast, and a mechanism for direct control that inhibits and cancels saccades subject to cognitive demand (e.g. CRISP, SWIFT, ICAT). This study statistically modelled adult fixation durations from a novel gaze-contingent visual search task, which allowed the isolation of these two components, represented as a bimodal distribution. From the modelled parameters we measured the proportion, distribution, and synchronisation of fixation durations. Participants were tasked to control when their eyes moved to match an isochronous auditory beat (IOI), whilst orienting sequentially around twelve circles (an elliptical shape). The currently fixated circle was in colour (red or blue) and the search task was identifying infrequent colour changes (tapping a trackpad). Each trial required a sequence of 180 circles and randomly played one of the eight IOI (from 300 to 1000ms in 100ms intervals), repeated twice. Mixture-modelling of the distributions converged on a bi-modal fit for all participants at each IOI. The first distribution had the shortest fixation durations (mu = 252ms) with small deviation (sigma = 65ms), both were minimally variant across IOI levels representing the profile of the saccadic timer. The size of the 'direct control' second distribution (lambda) increased significantly with the target IOI to asymptote at 50%. These directly controlled fixation durations and deviations significantly increased as the IOI increased. The proportion of fixations under direct control significantly predicted synchronisation performance (R2 = .47). Interestingly adding the participants finger-tapping ability did not increase the variance accounted for by lambda alone, which may indicate a shared fine-motor control mechanism. These data successfully profiled the two-components of eye movement timing, the saccadic timer and fixations under direct control, and provided a novel quantification of the proportion of fixations that can be directly controlled during active vision.
Meeting abstract presented at VSS 2018