Natural Language Processing and the Representation of Phenomenal Experience

This article is an interdisciplinary study of phenomenal judgments through the lens of linguistic correlations using computational linguistics and data mining. The research focus includes perceptual judgments describing interoceptive and olfactory states, considered in the context of the theory of embodied cognition and Charles Sanders Peirce's theory of perceptual judgments. The authors demonstrate that the linguistic expression of interoceptive and olfactory experiences reflects a deep connection between the body, culture, and language, and also reveals culture-specific strategies for conceptualizing sensory experiences. Particular attention is paid to the comparison of “natural” and “synthetic” olfactory judgments that was generated using large language models (LLMs). The developed methodology allows for the identification of parametric differences in lexical diversity, syntactic complexity, and stylistic richness of olfactory descriptions. The conducted analysis confirms that olfactory experience has high semantic instability and polymorphism, which complicates its formalization and automated processing. Nevertheless, the use of modern NLP methods opens up new opportunities for the parameterization of phenomenal judgments and an in-depth study of their structural and cognitive features. The work is of interest from philosophical, humanitarian, and engineering points of view, offering a methodological toolkit for studying the properties of embodied consciousness using methods of computer processing of a natural language.