Cortical Granular Computing

Abstract

Cortical processing is a set of organized computations carried out by a large set of cortical cell assemblies (CA) widely distributed over the entire cortex. These CAs specialize in computing specific pieces of information, called here, information granules. The most elementary information granules are either defined sets of sensory data or control instructions to output devices as muscles, glands, etc. They are agglutinated into high order granules associated to specific meanings or taks of a cognitive assignment to be computed. Granule aggregation obeys different grammars (logic, visual, verbal, temporal, spatial, etc.) and may be graphically characterized and specified by their neural connectivity. Examples of high order granules are: what/who; where, when, how, why granules etc. Special granules, called here, affective granules are in charge of assessing granule syntactic coherence and semantic relevance, as well as the value (good/bad; rewarding/punishing) of the information carried about by the granule. Verbal and visual cortical computations involved in the decoding of visual and/or verbal narratives are formalized, here, in this context of Granular Computing. The major types of granules involved in these computations were described and the processes of their aggregation were discussed. In addition, the most probably cortical locations of these granules and the neural connectivity supporting their aggregation were proposed based on the actual neurosciences literature. The most important contribution of the paper is to clearly stress the complexity of the cognitive cortical computations by discussing the complexity of granule aggregation associated with the decoding of a visual narrative (The flying mosquito video) and a verbal & visual narrative (a illustrated verbal text about Life Extinction).