Neural Implementation of Lexical Meaning Using Topographic Representation of Input Topology

In a series of publications beginning in 1980, the philosopher John Searle identified intrinsic intentionality as a fundamental problem for then-current attempts to model human cognition and for construction of genuinely intelligent AI systems, and this remains problematic today. The present paper addresses one aspect of it: how to implement intentional lexical meaning in a physical neural system. It proposes preservation of the similarity structure of environmental input stimuli by topographic organization of neural activation as the fundamental mechanism, and that this can be mathematically modelled as a mapping from an input manifold to a graph that preserves the manifold topology. The motivation is technological - how to build an artificial language system that incorporates intrinsic meaning - though what is proposed may be found conceptually applicable to modelling of how linguistic meaning arises in the mind and how this relates to the structure and dynamics of the physical brain.