John Perrone

Department of Psychology, Waikato University, Hamilton, New Zealand

The Visual Self-Motion Estimation Problem: Are We Heading in the Right Direction?

In the general area of visual motion perception, the topic of self-motion (or egomotion) estimation has always been prominent, possibly because locomotion is common to most species and the ability to navigate through the world without bumping into things is key to survival. Originally, the problem dealt with the question of how we can determine our heading direction from the pattern of instantaneous two-dimensional image motion (optic flow) generated on the retinae as we move through the world. However recently the emphasis appears to have shifted to examining other sources of heading information such as object shape and external target location. I will discuss the rise and fall of optic flow and indicate how the problem has been changed considerably from its original motion-based formulation. The biggest challenge to self-motion estimation models based on optic flow is the lack of physiological data supporting the idea that 2-D image velocity vectors are extracted and used by the visual system. I will review some of our work testing Middle Temporal (MT) neurons and Medial Superior Temporal (MST) neurons which favours the idea that heading estimation in primates is accomplished through the use of templates made up of collections of speed- and direction-tuned neurons.

This page was last updated on 1 Mar 2001.

Back.