Motion Pattern Processing in MSTd: A Computational Model

Scott A Beardsley
Brain and Vision Research Laboratory
Department of Biomedical Engineering, 
Boston University, Boston, MA 

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 
suggest the existence of specialized detectors for the 
component motion patterns (radial, circular, planar) 
associated with self-motion, that are consistent with the 
visual motion properties of cells in several cortical areas 
including the dorsal division of the medial superior temporal 
area (MSTd). In the work presented here we use computational 
modeling in conjunction with psychophysical (graded motion 
pattern) task to investigate the functional role of horizontal 
connections within a population of MSTd-like units.
    
In the psychophysical task, we examined observers’ ability to 
discriminate global perturbations in motion pattern stimuli 
for perceptual correlates to the bias for expanding motions 
reported in MSTd. Across observers discrimination thresholds 
varied significantly with the type of motion pattern 
presented, following a cyclic trend with radial thresholds the 
lowest and circular thresholds among the highest. We propose 
that the perceptual data can be accounted for by an 
interconnected functional architecture within MSTd. Through 
the development of a biologically constrained computational 
model we demonstrate that robust psychophysical performance 
can only be achieved by interconnected neural populations that 
systematically inhibit non-responsive units. We show that 
while populations of independently responsive units are 
capable of extracting visual information relevant to the 
psychophysical task, such populations are not computationally 
sufficient to extract perceptual estimates consistent with 
human performance. The model’s performance indicates that the 
cyclic trend in psychophysical thresholds is mediated 
primarily through signal enhancement associated with the 
strength and spread of inhibitory connections within a neural 
population whose preferred motions are biased towards 
expansions. Taken together, the results suggest that robust 
processing of the motion patterns associated with self-motion 
and optic flow may be mediated, in part, by inhibitory neural 
structures in MSTd.