The long-term goal of this research program is a fundamental understanding of the neuronal computations performed by the cortex for the perception of visual motion. A fundamental assumption underlying this work is that such computations are performed by Hebbian """"""""neuronal assemblies"""""""" - cortical microcircuits composed of a few hundred to a few thousand nerve cells. We hypothesize that such cortical microcircuits are (1) spatially stable from trial-to trial and (2) that activity within different classes of microcircuits varies along one or more stimulus dimensions. The proposed research will test these hypotheses by subjecting high-speed movies of cortical activity to dynamic principal component analysis developed by the Co-Principal Investigator. In this analysis each movie frame is treated as a temporal sample of a spatial function. The method of principal components is used to find the best approximation to this function. If the approximation is well-behaved in the time domain, the fun ction is space-time separable and the empirical eigenfunctions identify and classify the spatial activity. In a preliminary series of experiments, we have applied dynamic principal component analysis to high-speed movies of cortical activity evoked in an isolated, but intact, turtle visual system. The cortex was stained with a voltage-sensitive dye (VSD) and the VSD signals recorded with a silicon photodiode-based imaging system. In Us in vitro preparation, cortical VSD signals have waveforms that closely mirror the complex waveforms of compound post-synaptic potentials generated within the apical and basal dendrites of spiny pyrarnidal cells (Senseman, 1996a). When the visual stimulus is a simple diffuse light flash, more than 95% of the evoked response is captured within the first 10 principal components. Phase plane plots of the empirical eigenfunctions; exhibit smooth, continuous trajectories indicating the response is separable. The proposed research extends these preliminary find ing to include cortical activity evoked by moving visual stimuli. Each visual stimulus will be a change in contrast that moves across the visual field (i.e. a moving edge). Magnitude of contrast, velocity of movement, orientation of the light- dark transition and """"""""sharpness"""""""" of the transition will be varied in a systematic fashion. If our stated hypotheses are correct, we expect new principal components will be found in cortical responses evoked by these moving stimuli. The discovery of principal components that are dependent on one or more stimulus dimensions in a reproducible fashion, will provide the first empirical evidence for the physical existence of Hebbian neuronal assemblies within the visual cortex of any vertebrate species.
Showing the most recent 10 out of 181 publications
