The complex patterns of visual motion formed across the retina during self-motion, often referred to as "optic flow", provide a rich source of information describing our dynamic relationship within the environment. Psychophysical studies indicate the existence of specialized detectors for radial, circular, and planar motion patterns whose visual motion properties are consistent with cells in the medial superior temporal area (MST) of non-human primates.

As our understanding of the psychophysics and physiology associated with motion pattern processing has progressed, so to has the development of neural models capable of encoding the requisite visual motion properties. Increasingly, networks of biologically inspired neural units have been developed to examine the emergence of visual motion properties within computationally constrained systems (e.g. Pitts et al. 1997, Wang 1996). Still others have examined how such systems could be used to encode perceptually relevant visual motion information (e.g. Lappe at al. 1996, Perrone and Stone 1998, Zemel and Sejnowski 1998). Through their development, such models have typically focused on how neural structures spanning visual areas can be used to integrate, encode, and extract perceptually useful visual motion information from optic flow. Only recently has attention begun to focus on the functional role of neural connections within cortical areas, particularly later in the visual motion pathway. Specifically, what types of lateral connections are present within visual motion responsive areas and what role do they play in the emergent computational properties of the visual system?

In the work presented here we combine computational modeling and psychophysics to investigate the structural and computational role of MST in performing a graded motion pattern (GMP) discrimination task. Within a population of biologically plausible MST-like neural units we attempt to identify laterally interconnected neural structures that are computationally sufficient to extract equivalent measures of perceptual performance on the GMP task. Given the anisotropic visual motion properties reported in MSTd, are there specific neural structures that are sufficient to encode the psychophysical task? And if so, how do these neural structures compensate the reported anisotropies to produce the observed perceptual performance? By systematically addressing each of these questions we illustrate how human psychophysical performance, together with the visual motion properties reported in non-human primates, can be used to probe the internal neural structures of the human visual motion pathway.

References:

Lappe M, Bremmer F, Pekel M, Thiele A, and Hoffmann K (1996). Optic flow processing in monkey STS: A theoretical and experimental approach. J. Neurosci. 16(19): 6265-6285.

Perrone JA and Stone, LS (1998). Emulating the visual receptive-field properties of MST neurons with a template model of heading estimation. J. Neurosci. 18(15): 5958-75.

Pitts RI, Sundareswaran V, and Vaina LM (1997). A model of position-invariant, optic flow pattern-selective cells. Computational Neuroscience: Trends in Research 1997. New York, Plenum Publishing Corporation: 171-176.

Wang R (1996). A network model for the optic flow computation of the MST neurons. Neural Netw. 9(3): 411-426.

Zemel RS and Sejnowski TJ (1998). A model for encoding multiple object motions and self-motion in area MST of primate visual cortex. J. Neurosci. 18(1): 531-547.