The Predictive Emotional Selfhood Active-Inference Model (PESAM): A Computational Proof of Concept for Affective Self-Regulation under Uncertainty

This record contains version 1.2.0 of: “The Predictive Emotional Selfhood Active-Inference Model (PESAM):A Computational Proof of Concept for Affective Self-Regulation underUncertainty.” This version substantially revises the evidential scope and analysis ofthe earlier preprint. PESAM is presented as a computational proof of concept for diachronicself-coherence, treated as one restricted functional dimension ofemotional selfhood rather than as a complete or validated theory ofselfhood. The revised analysis: - treats avoidance-related cost, cumulative stress aggregate, and holding as the primary archived outcomes; - identifies the archived score as a deterministic, weight-dependent scalarization rather than an independent performance measure; - removes the zero floor when examining the archived scalarization; - reports episode-level aggregate cost per observed Avoid, paired bootstrap intervals, permutation tests, and descriptive cost–count regressions; - distinguishes architecture-relative counterfactual effects from claims of psychological necessity; - withdraws claims of competitive validation based on unmatched baseline comparisons; and - specifies direct empirical observables, competing models, and explicit disconfirmation criteria for prospective human and computational studies. The supporting version 1.2.0 dataset and reanalysis archive reproducesthe reported quantitative analyses from the preserved episode-leveloutputs. It does not contain the complete historical simulation-generatingimplementation, exact transition or observation matrices, condition-specificparameter values, or trial-wise latent-state trajectories. Supporting archive:https://doi.org/10.5281/zenodo.21715343 Version 1.1.1 remains available as the historical pre-revision version.Version 1.2.0 should be used for the current scientific interpretation.