The Contribution of Cortical Feature Processing to Oculomotor Target Selection

We effortlessly move our eyes to objects with specific features and avoid objects with other features. This feature-guided target selection behavior has been studied extensively in experimental psychology and systems neuroscience. By now, the visual and cognitive factors that mediate target selection and the neural signatures of target selection in oculomotor substrates are clear. For example, neural activation encoding targets and distractors gradually divergences over time therefore signalling stimulus identity, while visual/cognitive factors like bottom-up salience or top-down priority modulate the precise time of this divergence. But despite the extensive research on oculomotor target selection, little research has examined how neural activation in oculomotor substrates encoding potential eye movements vectors is reweighted to indicate stimulus identity. Meanwhile, a parallel branch of systems neuroscience has thoroughly examined the function and anatomy of the visual processing pipeline distributed throughout the neocortex of mammals. Heretofore, however, there has been little if any attempt to characterize the relationship between cortical visual feature processing and oculomotor vector encoding during feature-guided target selection.

This dissertation presents a series of behavioral experiments that provide several insights into this relationship. In these experiments, I measure the perturbation of target-directed saccades elicited by competitive remote distractors as a function of (1) the feature-space distance between targets and distractors and/or (2) distractor processing time. Given the close correspondence between saccade perturbation metrics and the underlying physiology of the oculomotor system, this methodology offers a non-invasive analog to examining the time course of oculomotor distractor activation during feature-guided target selection. In one set of experiments, I observed that distractor activation encodes the feature-space distance between targets and distractors in a manner consistent with attentional pruning of visual features observed in cortical feature representations during feature-based attentional deployment. In another set of experiments, I observed that the pattern of visual onset response latencies across distractor features mimics the pattern robustly observed between the cortical modules specialized for processing the respective features. These results indicate a close representational and temporal parity between feature encoding in oculomotor and (cortical) perceptual systems. I therefore propose a broad theory of oculomotor feature encoding whereby eye movement vectors in oculomotor substrates are dynamically and continuously reweighted by the feature-dependent network of cortical modules in the perceptual system necessary for representing the relevant feature set of the potential eye movement target.