BAA_Computational_Pipeline_v1.0

Complete simulation code, EEG benchmark data, and figure generation routines for the Boundary-Attention Architecture (BAA), a computational model of psychiatric disorders based on the Structure–Interference–Resonance (SIR) framework.
BAA models the psyche as a five-channel dissipative dynamical system governed by stochastic differential equations with four core parameters: gain (G), plasticity (β), damping (γ), and noise (σ). Parametric variation of a single architecture generates six clinical phenotypes — Healthy, Mania, PTSD, Depression, OCD, and ADHD — each characterized by a distinct dynamical signature.
The pipeline reproduces all results reported in the manuscript, including: (1) clinical profile characterization with four complexity metrics; (2) bifurcation analysis revealing three symmetry-breaking mechanisms (entropy death, dominance lock-in, energy collapse); (3) EEG complexity benchmarking against the Bonn University dataset (Andrzejak et al., 2001); (4) hysteresis analysis confirming first-order phase transitions; (5) comorbidity predictions via parameter mixing; and (6) treatment simulations for PTSD, Depression, and OCD.
The repository is fully self-contained: the Bonn EEG dataset is included, and all figures (5 main + 4 supplementary) are generated automatically with no external dependencies beyond NumPy, SciPy, and Matplotlib.
Associated manuscript: Vasilchenko, K. (2026). The Boundary-Attention Architecture: A Computational Engine for Parametric Psychiatry. Submitted to European Journal of Neuroscience.