Proceedings of The Physiological Society
University of Oxford (2011) Proc Physiol Soc 23, PC295
The Timing of Antisaccades
I. Noorani1,2, R. H. Carpenter2
1. School of Clinical Medicine, University of Cambridge, Cambridge, United Kingdom. 2. Department of Physiology, Development and Neuroscience, University of Cambridge, Cambridge, United Kingdom.
Antisaccades, saccades made deliberately in the opposite direction to a stimulus, are widely used as a research tool for investigating volitional control as well as neurological and psychiatric disorders [1-3]. Though commonly believed to result from competition in the brain  between processes representing normal pro-saccades and antisaccades, we lack both a precise quantitative model that could predict behaviour in detail, and quantitative data by which models could be evaluated. Response time, or latency, is increasingly used to provide information about neural decision processes: models can be tested rigorously by their ability to predict latency distributions. In our experiment, whose procedures had local ethical committee approval, we first asked 5 voluntarily-consenting subjects to make saccades to visual targets appearing unexpectedly 3 deg randomly to left or right of a central fixation target. From the distribution of these pro-saccadic reaction times, we estimated best fit values of the LATER parameters µ and σ of the basic decision mechanism. They then performed an anti-saccadic task, making saccades in the opposite direction to the targets presented. We tested the ability of a simple competition model to predict the observed latency distributions of both correct (anti-saccade) responses, and those incorrectly made to the target. The model involves a linear race to threshold between units representing antisaccades, pro-saccades, and ‘stop’ (suppression of pro-saccades); there is lateral inhibition between the antisaccade and pro-saccade units, and the starting level of the pro-saccade unit was equal to or lower than that for the antisaccade unit. The values of µ and σ for the antisaccade unit and pro-saccade unit were taken to be those from each subject’s control, pro-saccade data. The stop unit σ was constrained to be the same for all subjects; stop unit µ, starting level activity of the pro-saccade unit, and the level of lateral inhibition were fitted individually for each subject. A Monte Carlo method was used to simulate distributions, the optimum parameter values being estimated by iterative minimisation of the Kolmogorov-Smirnov statistic. Predicted correct and error response distributions did not differ significantly (p > 0.05) from observed for any of the five subjects. Thus a simple race model can predict the distributions of both correct and error responses in the antisaccade task. As well as advancing our understanding of decision mechanisms, by quantifying clinical antisaccade tests in terms of the underlying functional parameters this can aid the diagnosis and understanding of pathological conditions.
Where applicable, experiments conform with Society ethical requirements