Horace Barlow pointed out forty years ago that neural strategies for coding visual information should take advantage of the statistical regularities in natural images. The neural code should be "tuned" to the statistics of natural images.
We have been studying how regularities in the temporal changes in images that arise from natural movements (e.g. objects moving continuously in the world) determine optimal coding strategies for motion information. We use psychophysical studies to test hypotheses about what forms of motion information are coded in the brain and to test specific hypotheses about how neurons in certain areas of the brain (e.g. area MT) code motion information.