Abstract
<title>Abstract</title> <p>Cognitive function is often modeled through large-scale neural dynamics, while physiological and metabolic processes are typically treated as constraints. Here, we introduce a proof-of-concept generative computational framework integrating neural oscillators, an autonomic-like oscillator, and a scalar energy-like variable within a unified dynamical system. The model implements bidirectional coupling between neural activity, physiological input, and the energy-like variable, enabling analysis of their joint influence on simulated cognition. Brain–body coordination is quantified using phase synchronization between global neural activity and the physiological signal, while neural dynamics are characterized using predictability-based measures of temporal structure. Across heterogeneous network topologies and parameter regimes, direct associations between brain–body coordination and synthetic outcomes remain consistent. Predictability-based neural measures show modest associations with energy-sensitive outcomes but minimal associations with a control outcome. Mediation analyses indicate that the energy-like variable partially accounts for relationships between dynamical measures and synthetic outcomes across configurations. These effects persist under perturbations and ablation of explicit energy feedback, suggesting robustness of the coupling architecture. Given the synthetic nature of outcomes, all statistical effects reflect properties of model-internal dynamics. Rather than providing empirical evidence for biological cognition, this framework establishes a hypothesis-generating formal model of how coupled nonlinear systems with resource-like constraints can generate structured dependencies among neural, physiological, and energy-like variables.</p>