| name | modular-state-space-model-perception-cognition |
| description | A modular state-space model for human perception, cognition, and decision dynamics that links sensory inputs to behavior through latent internal states while maintaining interpretable connections to neuro-cognitive mechanisms. |
| metadata | {"arxiv_id":"2607.14078","authors":"Sven Schoonebeek, Carlo Cenedese, Anahita Jamshidnejad","published":"2026-07-17","categories":"eess.SY; cs.SY; q-bio.NC"} |
Modular State-Space Model of Human Perception, Cognition, and Decision Dynamics
arXiv: 2607.14078v2 | Published: 2026-07-17
Context
This paper addresses the need for behavioral models that are both psychologically interpretable and mathematically analyzable in human-centered adaptive systems. Many existing predictors operate as black-box input-output mappings or provide limited access to latent internal dynamics.
Core Methodology
The authors propose a modular state-space model where behavior is modeled as a perception-cognition-decision pipeline. The model consists of coupled mathematical mappings representing:
- Attentional selection
- Predictive inference
- Cognitive-state evolution
- Intention formation
- Action selection
The model links sensory inputs to observable behavior through latent internal states while retaining interpretable connections to neuro-cognitive mechanisms.
Key Contributions
- Provides a white-box dynamical structure for estimation, validation, and model-based control in human-centered settings.
- Establishes sufficient conditions for boundedness, Lipschitz regularity, forward invariance, contraction of perceptual inference under constant input, and input-to-state stability of cognitive state dynamics.
- Demonstrates a closed-loop rehabilitation case study where a receding-horizon controller uses the model to adapt movement difficulty from partial feedback.
- Shows that the model-based controller sustains simulated task participation and achieves lower realized cumulative cost than target-following and random baselines.
Implementation Steps
- Define the perception-cognition-decision pipeline as a series of coupled mathematical mappings.
- Model attentional selection, predictive inference, cognitive-state evolution, intention formation, and action selection as separate modules.
- Link sensory inputs to observable behavior through latent internal states.
- Ensure the model maintains interpretable connections to neuro-cognitive mechanisms.
- Establish mathematical properties (boundedness, Lipschitz regularity, etc.) for stability and robustness.
- Perform numerical sensitivity analysis to verify interpretable changes in perceptual tracking, cognitive amplification, intention expression, and action decisiveness.
- Apply the model to real-world scenarios such as rehabilitation, where a receding-horizon controller adapts task difficulty based on partial feedback.
Pitfalls
- Ensuring the model remains psychologically interpretable while being mathematically rigorous.
- Validating the model against empirical neuro-cognitive data.
- Computational complexity of estimating latent states in real-time applications.
- Balancing model complexity with interpretability.
Verification
- Compare model predictions with empirical behavioral and neurophysiological data.
- Conduct sensitivity analyses to ensure robustness of inferred parameters.
- Validate the model-based controller in simulated and real-world rehabilitation settings.
- Assess whether the model provides insights into neuro-cognitive mechanisms that black-box models obscure.
Activation
modular state-space model, perception-cognition-decision pipeline, human-centered modeling, computational neuroscience, brain network modeling, state-space modeling, 2607.14078