Dissociating temporal inhibition of return and saccadic momentum across multiple eye-movement tasks

Abstract

Saccade latencies are longer before eye movements to recently fixated locations, a phenomenon known as oculomotor inhibition of return (O-IOR). However, latencies are also influenced by saccadic momentum: latencies are shortest before forward saccades and increase as the angular deviation (difference in direction) of the current and previous saccades increases. Using a within-subjects and within-items design, the present study attempted to dissociate the temporal and spatial consequences of O-IOR and saccadic momentum across three tasks: scene search, scene memorization, and aesthetic preference. Spatial analyses revealed facilitation of return (rather than inhibition) in all tasks: return saccades were more frequent in the data compared to a shuffled baseline. Initial temporal analyses suggested that O-IOR was weaker in search (16 ms) than in memorization or preference (28 & 25 ms, respectively). Saccadic momentum appears even if previous and current saccades are not closely matched in amplitude, while O-IOR appears only when saccades are matched in amplitude so the current saccade lands within the zone of IOR (within 4° of the previous fixation location), making it possible to dissociate the two phenomena. Fixation durations increased linearly as a function of angular deviation outside the IOR zone, but curvilinearly within the zone of IOR, revealing an additional penalty for return saccades beyond that accounted for by saccadic momentum. Furthermore, saccadic momentum and O-IOR differentially affected the frequency distribution of fixation durations. Saccadic momentum shifted the mean of the distribution's Gaussian component, lengthening most fixations, while O-IOR only influenced the skewness of the distribution (the exponential component), lengthening a subset of fixations. Interestingly, when dissociated in this way, O-IOR was equivalent in magnitude across tasks while saccadic momentum was not present in search. These results suggest that O-IOR and saccadic momentum are independent phenomena, and that O-IOR is task-independent while saccadic momentum is not. Meeting abstract presented at VSS 2014