The chemical metaphor in neural computation

Abstract

Chemical reactions provide new insights concerning information processing in neural networks. Batch reactions are considered in first place and it is presented a model in which computational instructions will be conceptualised as being sent in parallel to multiple operational structures in sites where they are transformed in a way that generates a variety of computational viewpoints associated to a set of data. Anastomotic computations are considered as pre-processing operations implemented by convolutional weighing functions. Quantum valence theory is used to construct computational models for cognitive processes. The model of isolobal fragments is interpreted in a way that gives rise to classes of equivalence that represent concepts in extenso. The conscious quality of cognition is derived from the emergence of new macroscopic qualities from the relationships between valence bonds. Binding of representations is considered as dependent on dendro-dendritic relationships homeomorphic with orbital valence frontiers.