NEURAL CORRELATES OF MOTION PERCEPTION

本研究旨在建立前注意视觉运动的神经模型

This research aims to develop neural model of preattentive visual motion perception. A deeper understanding of design principles involved in biological vision and unification of the description of perceptual neural phenomena into a common theoretical language with a small number of dynamic laws and organizational principles are expected long-range outcomes of this research project. As a result of a variety of factors-such as the movements of the eyes and that of external objects, integration time, vergence, accommodation, and non-uniform sampling the retinal image is highly transient, blurred, and distorted. Yet, this problem received very little attention, for most of the models proposed in the literature axe built around the analysis of static (or steady-state) and uniformly in focus images. Thus, a fundamental problem in vision consists of the analysis of image transients to achieve dynamic yet sharp percepts despite these blurring effects. In this proposal, we will address this question by developing continuous-time form and motion channel models. We will carry out theoretical studies by using nonlinear analysis techniques and computer simulations to test and extend the predictions of specialized neural network architectures that we proposed to explain a wide range of data on motion and form perception in invertebrate and vertebrate species. The work will consider on the one hand motion perception without complex form perception in invertebrates and on the other hand motion perception with complex form perception in vertebrates. The study of the former will consist of the refinement of our earlier work by analyzing recent data and in particular, data on object tracking by motion cues. The study of the latter will consist of a motion channel and a form channel. For the motion channel we will extend, with suitable modifications, our invertebrate model to vertebrates. The fundamental properties of our model, namely nonlinear preprocessing, transient behavior, and habituation-sensitization will be studied by considering experimental paradigms probing these properties, i.e. Fourier & non-Fourier motion, apparent motion, and segregation by motion contrast respectively. For the form channel, we will test the predictions of the model by studying related experimental data: masking data to test the role of transient-sustained interactions in dynamic form perception and deblurring data to test the proposed temporal phases and their neural correlates. We will also carry out a bifurcation analysis to establish the properties of a more general class of extraretinal positive feedback interactions of the model. Modeling and testing will be based on extensive data ranging from single cell recordings to psychophysical experiments from a variety of species.