Neural mechanisms of motion integration and segmentation

Trustees Of Boston University, Boston

Investigators

Abstract

Neural Mechanisms of Motion Integration and Segmentation Abstract With National Science Foundation support, Drs. Mingolla and Pack will conduct a three-year study to investigate the mechanisms by which motion signals are integrated in the visual cortex of the alert primate. Coordinated theoretical and experimental methods will be used. Modeling work done in Mingolla's laboratory will include software implementation of competing models of motion integration and segmentation in order to derive the models respective parametric predictions of behavior of cells of the primate visual pathway. The experimental work will be performed by Christopher Pack in the laboratory of Richard Born at Harvard Medical School (HMS). Mingolla and Pack will work together with a graduate assistant at Boston University (BU) to ensure that model predictions are carefully derived and testable by electrophysiological techniques. Human psychophysical paradigms from the literature will be implemented and pilot tested at BU to ensure that they can be adapted to awake, behaving monkeys in a manner such that experimental results will discriminate meaningfully between the model predictions. The two categories of models to be investigated include those based on the intersection-of-constraints in velocity space and those that operate by a combination of feature tracking and vector averaging in networks of neurons involved in cooperative and competitive interactions. The experimental paradigms that have been identified as most promising are: (1) interaction of first-order and second-order motion signals, (2) multi-aperture stimuli, and (3) differential processing of intrinsic and extrinsic terminator signals. The neurophysiological experiments will be performed on alert rhesus macaque monkeys. All surgeries, animal care, training, and recording will be performed at Harvard Medical School. The bulk of the neural recordings will be done in cortical area MT, based on its central role in processing visual motion signals. We will also have the ability to record from areas V1, V2, and MST if necessary. All experiments will be carried out on two monkeys to ensure the generality of the findings. Mingolla and Pack have already collaborated successfully in both modeling and psychophysical projects. They have continued to meet on a regular basis since Pack joined Born's laboratory; this proposal is one outcome of their discussions. Funding would ensure that Mingolla's modeling work, partially funded by the Office of Naval Research, becomes ever more tightly coupled to the constraints of physiological data. At the same time, Pack's career development will be fostered by his taking primary responsibility for conducting theoretically important experiments in the area of human motion perception. Results of research will be reported at the Vision Sciences Society and Society for Neuroscience annual meetings. Software implementation of the NN model will be developed for distribution on the web. While the graduate assistant at BU will be the primary beneficiary of the enhanced training afforded by the collaboration, all of Mingolla's students at BU will profit from closer ties being fostered with HMS.