Embodied cognition is made inherently unstable by that very embodiment. Gene expression, immune function, cancer suppression, wound healing, animal consciousness, machine intelligence, network dynamics, institutional process, and their numerous and varied composites, must be stabilized by counterimposition of control information at rates that exceed those at which ‘topological information’ is imposed by embedding, rapidly-changing, real-world ‘roadways’. Here, we explore the dynamics and sometimes highly punctuated failures of cognition under different fundamental probability modes, using the lens of the asymptotic limit theorems of information and control theories. Via abduction and augmentation of standard methods from statistical physics and nonequilibrium thermodynamics, overall system dynamics can be studied across both their characteristic probability distributions and the hierarchy of their interacting scales. This work provides a foundation for building new statistical tools to explore the many arcane patterns found in observational and empirical studies across the pathologies, adversarial encounters, imbalances, and other selection pressures that erode and confound essential real-world cognitive structures and functions. One inference from this formulation, however, particularly emerges: all cognitive phenomena are subject to, and can be driven to, failure.                                                    — Huang (2021)                                                    — Perplexity AI (2024)                                                    — Raji et al. (2022).                                                    — Deborah N. Wallace (2024)

The Regulation of Embodied Cognition and Its Failure

章节
发表日期：

2025年

查看全文

Rodrick Wallace

摘要

Embodied cognition is made inherently unstable by that very embodiment. Gene expression, immune function, cancer suppression, wound healing, animal consciousness, machine intelligence, network dynamics, institutional process, and their numerous and varied composites, must be stabilized by counterimposition of control information at rates that exceed those at which ‘topological information’ is imposed by embedding, rapidly-changing, real-world ‘roadways’. Here, we explore the dynamics and sometimes highly punctuated failures of cognition under different fundamental probability modes, using the lens of the asymptotic limit theorems of information and control theories. Via abduction and augmentation of standard methods from statistical physics and nonequilibrium thermodynamics, overall system dynamics can be studied across both their characteristic probability distributions and the hierarchy of their interacting scales. This work provides a foundation for building new statistical tools to explore the many arcane patterns found in observational and empirical studies across the pathologies, adversarial encounters, imbalances, and other selection pressures that erode and confound essential real-world cognitive structures and functions. One inference from this formulation, however, particularly emerges: all cognitive phenomena are subject to, and can be driven to, failure. — Huang (2021) — Perplexity AI (2024) — Raji et al. (2022). — Deborah N. Wallace (2024)