Optic flow is an important visual cue that can be used to navigate through the environment. Evidence suggests that humans do use optic flow for navigation when it is available (Warren et al. 2001). In non-human primates, the dorsal section of the medial superior temporal area (MSTd) has been found to contain neurons that respond to patterns of optic flow, namely, large-field planar motion, complex motion, such as radial, rotational, and spiral motion, or complex motion combined with planar motion. Moreover, the middle temporal area (MT), which projects to area MST, contains neurons that are selective to small-field planar motion. As such, models of MSTd neurons have addressed how complex motion selectivity can be achieved by combining selectivity to small-field planar motion. In particular, attention has been paid to neural mechanisms that might provide the position invariant responses of MSTd neurons.

Among models of MSTd, most build complex motion selectivity by combining selectivity to several small-field planar motions, typically, planar motions found in the complex motion itself. In addition, despite the important role of inhibitory mechanisms in cortex, most do not discuss the role of inhibition in their models. Based on their electrophysiological studies of MSTd neurons in monkeys, Duffy and Wurtz (1991) suggested that complex motion selectivity might come from pairs of overlapping planar motion gradients combined with excitation and inhibition. We have modeled MSTd neurons on this premise. Our model uses a pair of subunits connected to a complex motion selective output unit, where one subunit provides excitation and one provides inhibition. Each subunit is selective to a particular planar motion and sums motion from part of the visual field. Using this architecture, we have modeled neurons selective to large-field planar, radial, rotational and spiral motion. More specifically, our model neurons exhibit gaussian tuning, as has been found with MSTd neurons, to the set of radial, rotational and spiral patterns. Further, by adding inhibition between model neurons selective to opposite complex motion patterns (e.g., expansion and contraction) we are able to provide position invariant responses. Finally, we have used our model in a complex motion discrimination task similar to that of Morrone et al. (1995) to illustrate that model neurons are capable of producing performance similar to humans on the same task.

Thus, our work provides a better understanding of how neurons in monkey area MSTd, and potentially in its human homologue, could process optic flow patterns using a simple inhibitory planar motion mechanism.

References:

Duffy, C. J. and Wurtz, R. H. (1991). Sensitivity of MST neurons to optic flow stimuli. II. Mechanisms of response selectivity revealed by small field stimuli. J. Neurophysiol. 65(6): 1346-1359.

Morrone, C., Burr, D. and Vaina, L. (1995). Two stages of visual processing for radial and circular motion. Nature. 376(6540): 507-509.

Warren, W. H. J., Kay, B. A., Zosh, W. D., Duchon, A. P. and Suhac, S. (2001). Optic flow is used to control human walking. Nature Neurosci. 4(2): 213-216