How Perception Adheres Color to Objects and Surfaces: Studies Using Visual Illusions and Transcranial Magnetic Stimulation

Citation

Wu, Daw-An (2006) How Perception Adheres Color to Objects and Surfaces: Studies Using Visual Illusions and Transcranial Magnetic Stimulation. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/QR9F-YR49. https://resolver.caltech.edu/CaltechETD:etd-09282005-121349

Abstract

In early visual cortex, visual input is encoded primarily as edges. Information then flows into more specialized regions of the brain, which process visual features such as color, motion, etc. This encoding scheme poses some problems in explaining the experience of seeing. If the cortex processes various visual features separately, how do we see unified objects? What mechanisms bind the features together? If the cortex encodes the visual scene in terms of its edges, then how do we see solid surfaces? What mechanisms fill-in the map of outlines?

This thesis investigates the problems of binding and filling-in using the techniques of visual illusion psychophysics and transcranial magnetic stimulation (TMS). We find that TMS can cause an instant replay effect, whereby recently presented visual stimuli are seen again. When TMS induces an instant replay shortly after the presentation of Cai’s asynchronous binding illusion, some subjects see an image of the actual visual stimulus, undistorted. It appears that TMS can selectively activate a hidden, accurate representation of the stimulus, revealing it without the distortion caused by other processes.

We also find a number of cases in which visual features are decomposed and/or misbound. TMS-induced instant replay can cause the color of one object to be bound to the position and orientation of another. It can also separately replay the color and orientation of a grating. In a non-TMS experiment, we create a stimulus that induces a steady-state misbinding of color and motion—-a vivid, long-lasting misbinding effect ideal for neurophysiological investigation. These experiments confirm the separate encoding of visual features and the existence of an active binding mechanism.

Finally, we study filling-in by manipulating an effect distilled from the artwork of Julian Stanczak, in which color is perceived to spread discretely among segregated patches of space. We find that color-filling is dependent on perceptual surfaces, such that overlaid surfaces can support separate filling processes. It appears possible that the neural mechanisms of binding and filling-in might be intimately related, both of them highly integrated with the process of surface segregation.

Thesis Files