Angular distribution of fractal temporal correlations supports adaptive responses to wobble board instability

Contemporary dynamical models of human postural control propose an intermittent controller regulating the postural centre of pressure (CoP) about a stable saddle-shaped manifold along anatomical anteroposterior (AP) and mediolateral (ML) axes, releasing CoP in an outwards spiral when inactive. Experimental manipulations can evoke this saddle-type topology in fractal temporal correlations along the AP axis and reducing correlations along the ML axis. However, true effects of task demands may often manifest within angular space between anatomical AP and ML axes-a space not typically modelled explicitly. We tested how instability and attentional load influence postural control across the full angular range of fractal variability along the two-dimensional (2D) support surface. Forty-eight healthy young adults performed a suprapostural Trail Making Test (TMT) while standing on a wobble board, inducing continuous perturbations along the ML axis. Stable, quiet standing exhibited classic saddle-like topology, with stronger fractal temporal correlations in CoP displacements along AP axes. The attentional demand of the TMT did not affect angular variation or strength of fractal temporal correlations across the 2Dsupport surface. However, maintaining upright balance on the wobble board reshaped and reoriented the angular distribution of fractal temporal correlations, accentuating saddle-like angular variation and rotating the strongest fractal temporal correlations predominantly along the ML axis. Stabilizing posture in the face of wobble board instability prompted the saddle-type angular distribution of fractal temporal correlations. These findings challenge the traditional dependence of postural control theories exclusively on external force-plate axes and underscore the significance of multifractality in defining control parameters that govern postural stability across the full angular range of the 2D support surface.