My research in systems neurobiology is aimed at developing an understanding of motion, depth, and form processing by the human visual cortex. We attack these problems from two perspectives. First, we gather quantitative psychophysical data measuring human visual function. We then integrate these data, which reflect the overall processing capabilities of the visual system, with primate single unit recording data obtained from other laboratories. We then use these data to develop detailed neural network models that accurately predict the human data and simultaneously account for the response characteristics of primate cortical neurons. We are applying this approach to visual pattern discrimination, stereopsis, and visual motion perception.

Our work on motion has generated the neural model shown schematically in A on this page. Briefly, the stimulus is first processed by orientation selective filters with the characteristics of simple cells in visual cortex. Following this filter stage, motion information is extracted by two parallel processing pathways. In the simpler one on the left in A, the directions of motion of pattern lines and edges (Fourier motions) are detected. The more complex pathway on the right in A involves a rectifying nonlinearity followed by further filtering at a different orientation before the extraction of motion information. This permits the model to compute the direction of motion of complex texture boundaries (non-Fourier motions). Both pathways are then combined at a final stage where a neural network uses feedback inhibition to compute the vector sum direction. This model permits accurate prediction of the perceived direction of patterns like that shown in B. In this pattern the two sets of intersecting parallel bars move in the directions shown by the two black vectors, yet the pattern is perceived to move in the direction of the white vector. This counter-intuitive result is a consequence of the final combination of the two parallel motion pathways in A. It is worth noting that there is anatomical evidence for the existence of these parallel motion pathways in monkey visual cortex: the simpler one corresponds to the V1ÞMT projection, while the more complex one corresponds to the V1ÞV2ÞMT projection.  (MT is known to be crucial for motion processing in monkeys.)

This example is illustrative of my approach to systems neurobiology by combining psychophysics with sophisticated neural modeling.  My future plans include extending this approach to depth perception and to the perception of motion in three dimensions.

 Wilson, H. R. (1997) A neural model of foveal light adaptation and afterimage formation. Visual Neuroscience, 14:403-423.

 Wilson, H. R., Wilkinson, F. & Asaad, W. (1997) Concentric orientation summation in human form vision. Vision Res, 37:2325-2330.

 Wilson, H. R. & Wilkinson, F. (1997) Evolving concepts of spatial channels in vision: from independence to nonlinear interactions. Perception, 26:939-960.

 Wilkinson, F., Wilson, H. R. & Ellemberg, D. (1997) Lateral Interactions in Peripherally-Viewed Texture Arrays. J. Opt. Soc. Am. A, 14:2057-2068.

 Wilson, H. R. (1998) Non-Fourier cortical processes in texture, form, and motion perception., ed. by P. S. Ulinski & E. G. Jones, Plenum, New York, Cerebral Cortex: Models of Cortical Circuitry , 14. In press.

 Wilkinson, F. and Wilson, H. R. (1998) Measurement of the texture coherence limit for bandpass arrays. J. Opt. Soc. Am. A. In press.

 Wilkinson, F., Wilson, H. R. & Habak, C. (1998) Detection and recognition of radial frequency patterns. Vision Res. In press.

 Wilson, H. R. & Kim, J. (1998) Dynamics of a divisive gain control in human vision. Vision Res. In press.

 Wilson, H. R. (1998) Spikes, Decisions & Actions: Dynamical Foundations of Neuroscience. Oxford University Press. In press.