NEURAL VISUAL CODING--IMAGE TO OBJECT REPRESENTATION

点击翻译按钮获取中文摘要

The broad aim for the next five years is to find a physiological approach to a central question of perception: the conception and representation of object& Experiments are proposed that will test the hypothesis of temporary (episodic) object representations in vision which are thought to be the intermediate stage between the image processing in visual cortex and the processes of object recognition and memory. The experiments are aimed at distinguishing between several alternative theories on the nature of such representations. Central to the project is the "binding problem", i.e., the question of how the system associates the features, attributes, or parts that belong to one object when several objects are present in a scene, in which form and for how long it represents this information, how it retrieves information from previous moments in the case of moving or changing objects, and how it keeps track of object identity. Answers to these questions are important for understanding perception and consciousness, and the function of the cerebral cortex in general. Specifically, they will contribute to a better understanding of visual disorders of central origin. A new behavioral paradigm and new visual test displays will be used in combination with multi-electrode extracellular neuron recordings to study the formation and decay of object representations and their use in recognition and identification of objects. The emphasis will be on recordings from visual area V4 and the inferior temporal cortex Various methods of spike train analysis will be used to analyze the patterns of neuronal responses, with particular attention to temporal coherence within and between neurons. Current theories of image segmentation, feature binding, and object representation will be evaluated, and new models developed, on the basis of the results.