Neurally and mathematically motivated architecture for language and thought

Area of Science:

• Cognitive Neuroscience
• Computational Linguistics
• Mathematical Psychology

Background:

• The neural basis of the interaction between human cognition and language remains largely unknown.
• Existing models struggle to reconcile psychological data with neural and computational constraints.
• The relationship between language acquisition and lifelong knowledge accumulation is poorly understood.

Purpose of the Study:

• To propose a novel computational architecture for the joint processing of language and thought.
• To explore the necessity of language for abstract thinking and its role in human rationality.
• To model the learning of word-object associations and explain cognitive differences in animals.

Main Methods:

• Development of a mathematical framework integrating neural, psychological, and computational principles.
• Analysis of hierarchical structures in language and thought.
• Modeling of language's emotionality and its influence on cognitive processes.

Main Results:

• A computable architecture for language-thought interaction is presented.
• The model supports the hypothesis that high-level abstract thought is dependent on language.
• It offers explanations for human irrationality, word-object learning, and the absence of human-like cognition in animals.

Conclusions:

• The proposed architecture provides a unified framework for understanding language-thought integration.
• Language's emotionality and grammatical structure significantly influence cognitive processes.
• The model opens avenues for future psychological and neuroimaging research.