Nima Akbari,

Title:

Realistic Internal Dynamics Are Essential for Human-Like Control: An Optimal Feedback Control Perspective

Abstract:

Humans skillfully control objects with internal dynamics, yet the computational principles underlying such coordination remain unclear. We compare stochastic optimal feedback control (OFC) simulations of a cart–pendulum system with experimental data from a cup and ball like task. Using fixed physical parameters and noise levels, the OFC model reproduces smooth, double-peaked velocity profiles and realistic trial-to-trial variability. Varying movement duration shows that the controller preserves a stereotyped velocity structure while scaling the depth of the inter-peak braking phase. When the internal model lacks impedance or coupling, the resulting behavior is unstable and qualitatively unlike human performance. These results indicate that realistic internal dynamics and feedback structure are essential for capturing human-like manipulation within an OFC framework, linking control-theoretic predictions with experimental motor behavior.