A cell assembly model of language

Abstract

Networks of formal neurons are introduced as paradigms for neuronal structures in the brain underlying language. Linguistic elements, e.g. syllables or words, are assumed to have their neuronal counterparts in Hebbian cell assemblies comprising parts which are located in different areas of the human cortex, in particular within the language regions. We investigate the effects of 'lesions' in artificial 'assemblies' in which only next neighbours are connected reciprocally (simulation A). In addition, we simulate the effects of lesions in a net consisting of several 'assemblies' which was taken as a model of the human language cortex (simulation B). The lesion studies lead to predictions on language disabilities in aphasics, i e. in individual with damage in the language-relevant cartex. The results provide an explanation of the fact that aphasic symptom complexes vary as a function of the damaged brain sites. The explanations cover the multimodal character of aphasias and the occurrences of the following symptoms: paraphasias, speech perception deficits, fluent versus non-fluent speech, and agrammatism. We conclude that systems of cell assemblies similar to our artificial networks are likely to play a crucial role in the linguistic machinery of the human brain.