This paper formalizes the Synthesis of Self, an information-theoretic framework integrating the Free Energy Principle and functional hemispheric lateralization into a unified, clinically actionable computational neuroscience architecture. We model the brain as an asymmetric dual-processor system, quantifying the interactive dynamics between a discrete tokenization engine (the language-dominant Manager network) and a continuous, analog field-processor (the relational Architect network). By evaluating the transcallosal coupling coefficient , computed via a non-linear sigmoidal transfer mapping function of underlying hardware constraints and vertical gain states, the framework demonstrates that early psychological fragmentation is directly measurable through micro-temporal latency variance. To establish out-of-sample empirical validity, the model's fixed calibration constants were applied to a massive behavioral database independent human subjects operating under allostatic load), exposing a violent, non-linear bifurcation of motor execution variance into two highly stable phenotypic attractor regimes (Executive Fracture vs. Defensive Armor). This underlying network geometry successfully maps the biophysical pathomechanics of major clinical phenotypes—including Schizophrenia, Anorexia Nervosa, Bipolar Disorder, Obsessive-Compulsive Disorder, and the consolidated Borderline-Narcissistic mirror axis—translating legacy, descriptive syndromic classifications into an objective, coordinate-based quantization of network processing failures. Crucially, by anchoring these complex state-space configurations within high-velocity, passive behavioral telemetry, this framework delivers an economical, highly scalable treatment and diagnostic metric capable of tracking real-time clinical trajectories and therapeutic efficacy without relying on costly neuroimaging infrastructure.