Saccadic eye movements are guided by attention. Indeed, some saccade trajectory effects serve as an index the attentional strength of visual objects in the map of visual space used to plan a saccade. One approach to understanding saccade planning relies on simple tasks in sparse displays (containing a single target and distractor object) to develop neurophysiologically plausible models of saccade behavior. Under tightly controlled conditions, saccade trajectories can be well predicted by representing displays of objects with simple visual features and their relative salience.
But the world in which the saccade system typically operates is not sparse, and observer eye movements are guided by more than just salience. As such, another approach has been to examine saccadic behavior in complex scenes and complicated goals. Such scene context can drastically affect saccades in ways that are not well predicted by a context-free and expectation-free representation of visual salience.
This dissertation starts to bridge this gap between these literatures by focusing on object surfaces. Covert shifts of attention operate on representations informed not just by stimulus salience and location-based expectations, but also by the perceptual organization of object surfaces. Covert attention can be guided by surface context, such that targets and distractors are processed differently as a function of whether they are on the same or different surface. These effects are fragile, however, and have previously only been demonstrated in relatively engaging tasks and with strong perceptions of objecthood.
The present work tested the strength of the relationship between attention and saccades by testing whether surface context guides orienting eye movements. Observers made saccades to objects that could be organized with different surface structure. In four experiments (Chapters 2 and 3) I found no evidence that the saccade map encoded surface context. But in two experiments (Chapters 4 and 5) I demonstrate saccade trajectories are sensitive to surface context, independently of low or high task engagement. This demonstrates that object surface-based representations are not necessarily fragile and can affect the oculomotor map even for simple saccadic orienting for which the surface is task-irrelevant. This lends evidence to the theory that the nature of the representation of vision is one of object surfaces, and suggests that the strength of object encoding is stronger than has been previously demonstrated.