Abstract
Some primate motion-sensitive Middle Temporal (MT) neurons respond best to motion orthogonal to a contour’s orientation (component types) whereas another class (pattern type) respond maximally to the overall motion of a pattern and somehow disregard the motion of individual contours moving in different directions and speeds. It is possible to construct a model of the pattern-type MT neurons using simple integration of the activity generated in speed- and direction-tuned subunits and such a model can account for a wide range of MT neuron properties (Perrone & Krauzlis, JOV, 2008). However two sets of findings appear to challenge this simple model. Some MT neurons have been shown to change from pattern to component behavior over brief time intervals (Pack & Born, Nature, 2001) and when the amount of ‘transparency’ in a plaid stimulus changes (Stoner & Albright, Nature, 1992). These results suggest the existence of complex top-down feedback mechanisms that are considered to be beyond the scope of simple integration models. We tested our MT model using the same stimuli used in these two studies (presented as 256 x 256 pixel x 8 frame movies) and we were able to replicate the pattern-to-component effects. We discovered that the two types of V1 neurons feeding into our pattern units have slightly different time delays; this initially favors the component response thus mimicking the Pack & Born temporal effects. We also discovered that the Stoner & Albright plaid stimuli contain a contrast asymmetry that depends on the plaid direction and the intensity of the intersections. The ‘transparent’ intersection condition has higher contrast in the component directions compared to the plaid direction causing the model MT pattern units to act as component units instead. This behavior replicates the Stoner & Albright pattern-to-component effect. Both of these examples demonstrate that feed-forward mechanisms can produce apparent top-down effects in MT neurons.
Meeting abstract presented at VSS 2013