Predictive Emotional Selfhood in Artificial Minds (PESAM): Simulation Data and Analysis Code

This archive contains Artifact B (data and analysis code) supporting thePredictive Emotional Selfhood in Artificial Minds (PESAM) framework. Version 1.1.0 (Adjusted v2, REALISTIC scoring) provides a corrected,fully audit-locked release of the simulation results previously reportedin v1.0.0, and should be treated as the authoritative dataset for allscientific use. Major updates in this release include: - Introduction of a realistic scoring protocol, in which policy decisions are belief-based, while internal stress dynamics are updated against environment-grounded contingencies, preventing belief-consistent bookkeeping. - Resolution of component ablation degeneracy (e.g., No_APC vs. No_SaH collapse) observed in earlier simulations, enabling meaningful lesion-based comparisons. - Full reproducibility lock via fixed random seeds, fixed configuration files, and SHA256 integrity logs generated from two independent FULLRUN executions. - Results packaged to exactly match all figures and tables reported in the camera-ready manuscript. In addition, this release includes a self-contained Audit Pack(PESAM_audit_pack_v1.1.0.tar.gz) comprising: - Unified analysis datasets (unified_synergy.csv, baselines.csv).- All manuscript figures and LaTeX tables corresponding to reported results.- Fixed random seeds and configuration files enabling paired statistical analyses.- A snapshot of the Python execution environment.- SHA256 checksums for all included artefacts.- The exact LaTeX manuscript source aligned to this dataset version. This version supersedes v1.0.0 and ensures transparent, deterministic,and independently auditable reproducibility of all reported results.