Schoonebeek et al.'s modular state-space model decodes human perception-to-decision pipeline

The paper introduces a modular state-space model that formalizes human behavior as a perception-cognition-decision pipeline. Each stage—attentional selection, predictive inference, cognitive-state evolution, intention formation, and action selection—is represented by coupled mathematical mappings with clear neuro-cognitive interpretations. Unlike black-box predictors, this framework provides full access to latent internal dynamics while remaining mathematically analyzable. The authors prove sufficient conditions for boundedness, Lipschitz regularity, forward invariance, contraction under constant input, and input-to-state stability, ensuring reliable behavior in human-centered adaptive systems.

Numerical sensitivity analyses show the model produces interpretable changes in perceptual tracking, cognitive amplification, intention expression, and action decisiveness. The most compelling demonstration is a closed-loop rehabilitation scenario where a receding-horizon controller uses the model to adjust movement difficulty based on partial feedback. The model-based controller sustains simulated task participation and achieves lower realized cumulative cost compared to target-following and random baselines. This proof-of-concept highlights the potential for white-box dynamical models in estimation, validation, and control of human-machine interactions, particularly in healthcare and adaptive automation.