Abstract
Neurons in the Middle Temporal (MT/V5) area of the primate brain have been classified into Component (C) and Pattern (P) types (Movshon et al., 1983; Albright, J. Neurophysiol., 1984). The rationale for the existence of P types is well established and the common thinking is that they integrate multiple motion directions over a patch of the retina and respond to the overall pattern motion, not to the motion of individual components making up the pattern. The function/role of neurons that respond primarily to the components (C-type) is less well understood. While developing a velocity code based on the outputs from small sets of MT pattern neurons (Perrone & Krauzlis, VSS, 2007) we have discovered an important function that could be filled by MT component neurons. MT neurons are speed tuned only and so the determination of image velocity relies on the combination of outputs from a set of neurons tuned to a range of image speeds and consequently with different receptive field sizes. Component neurons can be used to prevent the velocity estimates being dominated/skewed by the outputs from MT pattern neurons with the largest receptive fields. In our velocity code we use C units to introduce a spatial derivative mechanism; we propose that each MT pattern neuron is inhibited by spatially adjacent MT component neurons tuned to the same speed. The amount of inhibition is made dependent upon the contrast of the stimulus by exploiting the fact that some MT neurons change their speed tuning at low contrast (Krekelberg et al., J. Neurosci., 2006). The resulting mechanism enables good spatial localisation of the moving image feature at high contrast but allows for greater pooling and spatial integration at low contrast, consistent with MT surround inhibitory effects (Pack et al., J. Neurophysiol., 2005).
Supported by the Marsden Fund Council from Government funding, administered by the Royal Society of New Zealand.