Predictive Emotional Selfhood in Articial Minds (PESAM): A Unied and Synergistic Variational Framework

The preprint operates within the paradigm of the Active Inference framework and the Free Energy Principle in computational neuroscience and artificial intelligence. This paradigm posits that intelligent agents perceive, act, and learn by continuously minimizing prediction error (or surprise) about their sensory inputs. The paper specifically addresses the application of this paradigm to modeling higher-order cognitive phenomena like emotion and selfhood.

This preprint is classified as Model Drift because it identifies a significant gap within the established Active Inference paradigm—its failure to provide a unified, synergistic account of emotion and selfhood. The author argues that previous models are "piecemeal" and cannot solve complex, integrated problems. The proposed Predictive Emotional Selfhood in Artificial Minds (PESAM) framework is a substantial modification, not a replacement, of the core paradigm. By integrating three distinct mechanisms (APC, SaH, AHO), PESAM extends the existing model to cover these previously un-integrated phenomena, thereby increasing the paradigm's explanatory power and addressing a recognized anomaly.