Cartography of the Mental Topography of Free-Energy Minimization: A Computational Approach Based on Narrative Trajectory Analysis Using Large Language Models (LLMs)
Year: 2026
Type: preprint
Abstract: How does the brain protect its beliefs from contradictory facts? Neuroscience answers: through free-energy minimization (FEP, K. Friston), but it does not know the content of these defenses. Linguistics is able to analyze text topics but does not see the dynamics of cognitive tricks. We propose an interdisciplinary method that extracts from text corpora not static topics but dynamic transition graphs: from initial prior beliefs through a point of cognitive friction to final smoothing. Using LLMs in deep reasoning mode, we filter markers of cognitive effort, identify defense operators, and classify them by energy cost (external vs. internal attribution). A rigorous falsifiable hypothesis is advanced concerning a power-law rank distribution of operators, in which low-cost operators (externalization, escapism) monotonically dominate high-cost operators (reflection, restructuring of priors). The method includes a stratified sample across three genres, two-pass semantic compression guided by a taxonomy (Kahneman, A. Freud), and a second-derivative algorithm for identifying the defense core. The project translates qualitative hypotheses into measurable variables and creates the first empirical atlas of collective cognitive algorithms.
Language: English
Subfield: General Social Sciences
Field: Social Sciences
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7204074023
Cites:
Cited by:
Related to: