Psychophysics of Motion Perception

Summary

In this paper we make three major points about motion perception. As often happens, these same three points are echoed by other contributions to this Symposium. So we begin by using the ideas and words of others to anticipate what we ourselves wish to say. First, Berkley’s contribution (1981) reminds readers that motion perception is a complex skill, dependent on the combination of many abilities. Starting from a quite different perspective, Bonnet (1981) arrived independently at a similar position: what is called “motion perception” is actually a heterogeneous collection of diverse functions, not a single monolithic response. The papers by Berkley and Bonnet remind us how important it is to approach motion perception through a variety of complimentary paradigms, procedures and techniques. In fact, Bonnet has provided us with an excellent taxonomy of the paradigms used in the psychophysical study of motion perception. The reader of our paper, or any paper in the Symposium, will certainly profit from Bonnet’s thoughtful taxonomy and exhaustive bibliography.

Several symposiasts demonstrated how inhomogeneous motion responses are throughout the visual field. For example, Orban (1981) showed the properties of single neurons in the cat’s visual cortex vary with the spatial location of their receptive fields. Discussing applied aspects of motion perception, Leibowitz et al. (1981) emphasized differences between responses to moving targets mediated by central and peripheral vision. The second lesson then is that if we are to understand motion perception in its entirety, we must examine it at different points in the visual field. Unfortunately, we cannot assume that measurements made in one part of the visual field can be generalized to other parts.

A third lesson we learned from Pasternak et al. (1981) and Bonnet (1981). Although one can focus, as we shall in this paper, on responses to direction of motion, one cannot ignore another important variable, velocity. Along these lines Pasternak reported that the difference threshold for direction of motion depends upon the target speed used to measure it. In one of the main points of his contribution, Bónnet argued that motion perception undergoes qualitative changes as target speed varies.

These then are the three major points we wish to reinforce in our paper. The nature of motion perception itself requires that 1) motion be approached with a battery of paradigms, 2) responses in different parts of the visual field be compared, and (3) measurements be made at as large a range of target speeds as possible.

Now we can turn to the plan we shall follow in expressing these three ideas about motion perception. First, we shall describe the stimuli and general methods used in the experimental work. Then we shall outline the model that guides our work. Next will come a consideration of the model’s details, particularly as they bear upon the character of visual mechanisms enabling us to see moving targets.

The concluding section of the paper takes us further from the laboratory, to the consequences for motion perception of an observer’s inability to predict precisely what moving target he is looking for. This sort of effect, called stimulus uncertainty, has occupied a good deal of our time in the past three years and has taught us much about motion perception and about perception more generally. In the brief treatment of stimulus uncertainty here, we will consider three separate but complementary topics: performance losses associated with stimulus uncertainty, how those losses can be compensated for or mitigated, and the discrepancy between being able to see a target is moving and being able to identify the direction in which it moves.

We would like to acknowledge the research support provided by the National Science Foundation and the Air Force Office of Scientific Research.