Neural networks in space and time: A new cue to motion perception

Hyungjun Kim, Purdue University

Abstract

The purpose of the present study was to test the prediction that a moving object produces a trail of neural activity which could be a new spatial cue to motion perception. This cue was predicted through analysis of a neural network model of visual perception and was validated by a series of psychophysical experiments. In the first four experiments, multi-stable apparent motion displays were used to test a hypothesis that motion lines should bias motion percepts in an otherwise ambiguous motion display. In the second set of experiments, experimental verification of a prediction that movement direction is partly coded by the strength of its rebound trails was tested using an complementary orientational afterimage (CAI). With the introduction of stimulus properties such as luminance, size, and duration into the bars of a CAI inducing image, the perceived direction of CAI motion was shown to be as predicted by the model. The third set of experiments was designed to test another hypothesis that the trail contains information about the speed of the movement. Both luminance and size of inducing images controlled perceived speed in CAI, as predicted by the theory. All sets of psychophysical experiments suggested that the hypothesized trail codes information about the direction and speed of the movement. Moreover, trail signals were shown to imply motion parallel to their orientation. Thus, the experimental results cannot be understood by the properties of oriented cells in striate cortex, because their motion preference is orthogonal to their orientation preference. In the last section, a model was developed to incorporate new findings from the psychophysical experiments. A hypothetical connection from orientationally tuned cells to motion cells was embedded into the model to demonstrate how the trails can act as a cue to motion.

Degree

Ph.D.

Advisors

Francis, Purdue University.

Subject Area

Cognitive therapy|Psychology|Experiments|Psychobiology|Neurology

COinS