Abstract
Abstract: :
Purpose: The properties of a visual object (contrast, color, form, movement) are defined by their coherence in space. Within the information processing in the visual system this features has to be bound to allow the recognition of an object. The coherence in space is reflected in the close physical location of the processing elements in the nervous system. This allows us to study coding principles in the visual system by using multi–electrode recording. Here we present results which reveal how visual information is coded by the massive parallel and distributive processing in the vertebrate retina. Methods: The experiments employed superfused, flattened preparations of intact rabbit retinas. Visual stimuli with a variety of defined physical properties were used to probe the response characteristics of the random selected ganglion cell populations. Ganglion cell populations of several hundred units were recorded extracellularly using the a 3–dimensional sharp tipped multi–electrode array (UCEA). Results: Based on in vivo intrinsic deterministic signals in the population response the ganglion cell populations can be profiled and sub– or super–classes of single units are identified. The reconstruction of the applied stimuli from the recorded neuronal response patterns shows that by taking this information in account, common coding properties of ganglion cells in the retina can be identified. Conclusions: We are presenting the application and implementation of several techniques to identify, general common coding features of in vivo neuronal ensemble data.
Keywords: retinal connections, networks, circuitry • vision and action