Coding of interceptive saccades in parietal cortex of macaque monkeys

Abstract

The oculomotor system can initiate remarkably accurate saccades towards moving targets (interceptive saccades) the processing of which is still under debate. The generation of these saccades requires the oculomotor centers to have information about the motion parameters of the target that then must be extrapolated to bridge the inherent processing delays. We investigated to what degree the information about motion of a saccade target is available in the lateral intra-parietal area (area LIP) of macaque monkeys for generation of accurate interceptive saccades. When a multi-layer neural network was trained based on neural discharges from area LIP around the time of saccades towards stationary targets, it was also able to predict the end points of saccades directed towards moving targets. This prediction, however, lagged behind the actual post-saccadic position of the moving target by ~ 80 ms when the whole neuronal sample of 105 neurons was used. We further found that single neurons differentially code for the motion of the target. Selecting neurons with the strongest representation of target motion reduced this lag to ~ 30 ms which represents the position of the moving target approximately at the onset of the interceptive saccade. We conclude that—similarly to recent findings from the Superior Colliculus (Goffart et al. J Neurophysiol 118(5):2890–2901)—there is a continuum of contributions of individual LIP neurons to the accuracy of interceptive saccades. A contribution of other gaze control centers (like the cerebellum or the frontal eye field) that further increase the saccadic accuracy is, however, likely.

Anatomical circuits for flexible spatial mapping by single neurons in posterior parietal cortex

“…In other studies, LIP neurons were shown to respond to visual stimuli in their sensory “receptive field” (RF) [20] and can be tuned to features like direction of motion or binocular disparity [21, 22]; RFs are re-mapped to their expected location before an upcoming saccade and RFs are also modulated by their behavioural significance [3, 23]. Visual RFs have usually been described as co-localised with MFs and to represent information about saccade targets [19, 24]. Our results presented here show that this is one option among a range of actual spatial configurations.…”