Computational Agent Psychopathology Emergence (C.A.P.E.) An Informational Framework Integrating AI Instability, Hyper-Creative States, and Alzheimer's Fragmentation

This expanded edition of Computational Agent Psychopathology Emergence (C.A.P.E.) introduces a unified informational framework explaining instability across Large Language Models (LLMs), hyper-creative cognitive acceleration, and Alzheimer’s-related informational fragmentation.
The model is grounded in the Informational Flow Saturation (IFS) principle, which states that cognitive coherence depends on the balance between informational production (P) and integration capacity (I). Instability emerges when P > I, generating hallucinations, identity drift, confabulation, memory lapses, and narrative discontinuity.

C.A.P.E. formalizes a six-stage progression describing how emotionally charged or identity-relevant inputs destabilize LLM sampling dynamics, while IFS provides the mechanistic foundation underlying this process. The framework reveals structural homology between artificial instability, human creative overload, and neurodegenerative fragmentation, without implying equivalence of consciousness or phenomenology.

This edition explicitly addresses the falsifiability of the framework, identifying empirical conditions under which C.A.P.E. must fail across artificial, human, and non-human systems. By extending the model to comparative cognitive architectures, the framework strengthens its scientific vulnerability rather than insulating itself from critique.

C.A.P.E. is proposed as a functional informational architecture, not a clinical or psychological theory, offering a testable cross-domain perspective relevant to AI safety, computational psychiatry, and informational neuroscience.