Multisensory fusion: simultaneous re-weighting of vision and touch for the control of human posture

Abstract

We examined the generally held belief that the postural control system is able to re-weight its available sensory inputs in order to optimize stance control in altered sensory environments. Our view is that previous accounts of sensory re-weighting provide only indirect evidence, which is subject to alternative explanations. The present results provide strong evidence for sensory re-weighting as the primary mechanism for changes observed in postural sway between conditions. Subjects were presented with small-amplitude, oscillatory visual and somatosensory stimuli at 0.20 and 0.28 Hz, respectively, in five conditions that manipulated the amplitudes of stimulus motion. Gain calculated in each trial with respect to each of the two stimuli was found to change systematically as stimulus motion amplitudes changed across condition. The observed pattern of gain rules out a constant-weight, linear account of posture and is consistent with the re-weighting hypothesis. Parameter fits of a third-order, linear stochastic model to postural sway trajectories in each condition showed that changes in gain across condition were primarily due to changes in coupling coefficients rather than changes in parameters that characterize the stability of the postural system. Visual gain was found to depend upon visual motion amplitude and touch gain was found to depend upon touch motion amplitude, indicating intra-modality dependencies. Visual gain also depended upon touch motion amplitude, indicating an inter-modality dependence. To our knowledge, simultaneous re-weighting of more than one sensory input has never been rigorously demonstrated. These techniques may be able to resolve the source of balance control deficits across populations with far more certainty than currently possible.