Sensory integration of a light touch reference in human standing balance

Abstract

In upright stance, light touch of a space-stationary touch reference reduces spontaneous sway. Moving the reference evokes sway responses which exhibit non-linear behavior that has been attributed to sensory reweighting. Reweighting refers to a change in the relative contribution of sensory cues signaling body sway in space and light touch cues signaling finger position with respect to the body. Here we test the hypothesis that the sensory fusion process involves a transformation of light touch signals into the same reference frame as other sensory inputs encoding body sway in space, or vice versa. Eight subjects lightly gripped a robotic manipulandum which moved in a circular arc around the ankle joint. A pseudo-randomized motion sequence with broad spectral characteristics was applied at three amplitudes. The stimulus was presented at two different heights and therefore different radial distances, which were matched in terms of angular motion. However, the higher stimulus evoked a significantly larger sway response, indicating that the response was not matched to stimulus angular motion. Instead, the body sway response was strongly related to the horizontal translation of the manipulandum. The results suggest that light touch is integrated as the horizontal distance between body COM and the finger. The data were well explained by a model with one feedback loop minimizing changes in horizontal COM-finger distance. The model further includes a second feedback loop estimating the horizontal finger motion and correcting the first loop when the touch reference is moving. The second loop includes the predicted transformation of sensory signals into the same reference frame and a non-linear threshold element that reproduces the non-linear sway responses, thus providing a mechanism that can explain reweighting.

Fig 1. Experimental setup.

A) Schema of the experimental setup with the two touch reference heights (not scaled proportionally). Arrows indicate touch reference motion. B) Definition of variables body-in-space (BS; body sway) and touch reference-in-space (TS; the stimulus), and body-to-touch reference (BT; light touch sensory input) as angle (_ang) and horizontal translation (_trans). C) Angular motion of the stimulus, where the stimulus is pairwise identical across touch reference heights. Shown are single cycles at three amplitudes; cycles were repeated 12 consecutive times within one experimental trial. D) Stimulus shown as horizontal motion, where the stimulus within one angular amplitude changes with touch reference height.

Fig 2. Angular body sway responses.

A) Angular body sway responses (BS_ang) to the stimulus applied at two heights (L, H) and three amplitudes (1, 2, 3), averaged across stimulus cycles and subjects, shown in the time domain. B) Root mean square (RMS) values and standard errors (SE) across subjects of the sway responses plotted against stimulus RMS. C) RMS and SE of the ratio between sway response RMS and stimulus RMS against stimulus RMS.

Fig 3. Horizontal translation of body sway responses.

A) Horizontal translation of body sway responses (BS_trans) to the stimulus applied at two heights (L, H) and three amplitudes (1, 2, 3), averaged across stimulus cycles and subjects, shown in the time domain. B) Root mean square (RMS) values and standard errors (SE) of sway responses plotted against stimulus RMS. C) RMS and SE of the ratio between sway response RMS and stimulus RMS against stimulus RMS.

Fig 4. Sensory integration model for light touch cues.

Triangles are gain factors, where values are given in Table 1 with the exception of the COM height h (0.96 m) and the touch reference height hTP (0.8 m or 1.2 m). Linearized inverted pendulum dynamics are implemented as a transfer function using the Laplace transform. ‘Set point’ provides the desired body orientation, which is adjusted by the light touch feedback. a-e indicate points of interest in the model. a) output of the inverted pendulum dynamics is the body angle in space. b) the sum calculates the physical horizontal distance between the finger and the body COM as it is sensed by light touch. c) the gain factor transforms the body angle as sensed by Sensors BS_ang to a horizontal distance ${\widehat{BT}}_{trans}$ signal using the small angle approximation. d) the sum provides a sensory reconstruction of the Stimulus. e) indicates the correction of the light touch feedback through the estimated touch motion. The model is drawn to represent the signal flow in the central nervous system as compared to an engineering view with stimulus as positive input on the left and body sway as output for systems identification on the right.

Fig 5. Frequency domain results.

Frequency response functions of horizontal COM and stimulus motion displayed as gain (top row) and phase (mid row) against frequency for all six experimental conditions. Gain here is the amplitude ration between body sway response and stimulus. Phase is the temporal relation. The bottom row shows the magnitude-squared coherence against frequency. Shown are group averages of experimental results (black) and simulation results (grey).

Fig 6. Random sway.

Random sway RMS +/- SE against stimulus amplitude; a measure of the sway component not correlated with the stimulus.