Different properties of successful and error saccades in marmosets

Abstract

Various oculomotor tasks have been used to study eye movements, cognitive control, attention, and neurological disorders. Typically, analysis focuses on successful trials, where the saccade lands very close to the intended target, in both humans or non-human primates (NHPs). Error trials, in which the saccade fails to land on the intended target, are often excluded from these analyses. In this study, we hypothesized that saccades contain information that can predict whether they will result in success or not. We collected data from common marmosets performing the gap saccade task and the oculomotor delayed response task. Successful saccades in both tasks were characterized by higher peak velocities, shorter durations, and shorter latencies compared to errant saccades, regardless of whether the amplitudes were matched or not. These results were further validated using a generalized linear model, with saccade velocity, duration, and latency as predictors. The model demonstrated high accuracy in distinguishing between behavioural outcomes. Our findings suggest that the likelihood of a saccadic eye movement leading to a successful outcome may be predetermined, potentially reflecting the interaction between cognitive processes and saccade programming.