Logical rules of visual brain: From anatomy through neurophysiology to cognition

Humans can easily recognize objects as complex as faces even if they have not seen them in such conditions or context before. It seems that perceptually we are insensitive to the exact properties of an object’s parts but the same parts in different configurations or contexts may introduce contradictory effects. From a computational point of view, we are often insensitive to changes of some symbols, but the same symbols may lead to different classification of the same object. In present work we are looking for the anatomical and neurophysiological basis of these perceptual effects. We describe interactions between parts and their configurations in different areas of the visual brain on the basis of a single cell electrophysiological activity in the thalamus, and cortical areas V1 and V4. Our model is based on feedforward (FF) and feedback (FB) interactions between these areas. In the retina and thalamus simple light spots are classified, V1 is the first area extracting edge orientation and V4 is the first area sensitive to simple shapes. The FF pathways combine properties extracted in each area into hypothetical objects. Area V1 presents concepts of orientations, while area V4 presents concepts of a simple shape. The FB pathways are responsible for comparison of the learned concepts with extracted object’s properties – they form predictions. In each area structure related predictions are tested against hypothesis. We formulate a theory in which different visual stimuli are described through their condition attributes: responses in LGN, V1, and V4 neurons are divided into several ranges and are treated as decision attributes. Applying rough set theory [Pawlak, Z. (1991). Rough sets – theoretical aspects of reasoning about data. Boston, London, Dordrecht: Kluwer Academic Publishers] we have divided our stimuli into area dependent equivalent classes. We propose that relationships between decision rules in each area are determined by two different logical rules: “driver logical rule” of FF and “modulator logical rule” of FB pathways. These interactions are proposed to be a neurophysiological basis of the object classification.