[Paper Review] To react or not to react? Intrinsic stochasticity of human control in virtual stick balancing

Arkady Zgonnikov, Ihor Lubashevsky+3 more|arXiv (Cornell University)|Feb 13, 2014

Human-Automation Interaction and Safety33 references3 citations

TL;DR

This paper proposes a noise-driven, intrinsically stochastic model of human control in virtual stick balancing, where control activation arises not from fixed thresholds but from a dynamic interplay between stabilizing intent and the desire to minimize effort. The model successfully reproduces experimental data, demonstrating that intrinsic stochasticity—rather than threshold-based intermittency—plays a key role in the fluctuations of human-controlled unstable systems.

ABSTRACT

Understanding how humans control unstable systems is central to many research problems, with applications ranging from quiet standing to aircraft landing. Increasingly much evidence appears in favor of event-driven control hypothesis: human operators only start actively controlling the system when the discrepancy between the current and desired system states becomes large enough. The event-driven models based on the concept of threshold can explain many features of the experimentally observed dynamics. However, much still remains unclear about the dynamics of human-controlled systems, which likely indicates that humans employ more intricate control mechanisms. The present paper argues that control activation in humans may be not threshold-driven, but instead intrinsically stochastic, noise-driven. Specifically, we suggest that control activation stems from stochastic interplay between the operator's need to keep the controlled system near the goal state on one hand and the tendency to postpone interrupting the system dynamics on the other hand. We propose a model capturing this interplay and show that it matches the experimental data on human balancing of virtual overdamped stick. Our results illuminate that the noise-driven activation mechanism plays a crucial role at least in the considered task, and, hypothetically, in a broad range of human-controlled processes.

Motivation & Objective

• To investigate the mechanisms underlying human control of unstable systems, particularly in the context of virtual stick balancing.
• To challenge the prevailing event-driven, threshold-based control hypothesis by proposing an alternative noise-driven activation mechanism.
• To develop and validate a stochastic model that captures the intrinsic variability in human control behavior.
• To demonstrate that noise-driven control better explains extreme fluctuations and system dynamics than traditional threshold models.

Proposed method

• The study uses a virtual overdamped stick-balancing task with 10 human participants to collect real-time control and state data.
• A stochastic control model is developed based on the interplay between the operator's need to stabilize the stick and the tendency to delay or avoid control actions.
• The model incorporates a loss function minimizing control effort and deviation from the upright position, solved via optimal control theory with Lagrange multipliers.
• The optimal control response is approximated as open-loop, with feedback only at the start of each control episode, based on Bellman’s principle of optimality.
• The model’s dynamics are derived from a system of differential equations with eigenvalues indicating oscillatory, damped behavior consistent with experimental observations.
• Theoretical solutions are validated by comparing simulated control patterns and stick angle fluctuations to empirical data from human subjects.

Experimental results

Research questions

• RQ1 What drives the intermittent activation of control in human balancing of unstable systems?
• RQ2 How does intrinsic stochasticity compare to threshold-based models in explaining human control dynamics?
• RQ3 Can a noise-driven control mechanism reproduce the extreme fluctuations observed in human-controlled virtual stick balancing?
• RQ4 What role does control effort minimization play in shaping the timing and amplitude of corrective actions?
• RQ5 To what extent can open-loop control approximations explain human behavior in unstable control tasks?

Key findings

• The proposed noise-driven control model successfully reproduces the statistical properties of human control behavior in virtual stick balancing, including the distribution of control intervals and stick angle fluctuations.
• The model's eigenvalues, derived from optimal control theory, indicate oscillatory, damped dynamics consistent with observed human responses.
• The control activation mechanism is not triggered by fixed thresholds but emerges from a stochastic interplay between stabilization needs and effort avoidance.
• The model's predictions align closely with experimental data, particularly in capturing the irregularity and variability of human control actions.
• The results suggest that intrinsic stochasticity, rather than deterministic thresholds, may underlie complex dynamics in a broad range of human-controlled processes.
• The study provides evidence that open-loop control approximations can effectively model human behavior when combined with stochastic activation mechanisms.