Interactive footstep sounds modulate the perceptual-motor aftereffect of treadmill walking

Abstract

In this study, we investigated the role of interactive auditory feedback in modulating the inadvertent forward drift experienced while attempting to walk in place with closed eyes following a few minutes of treadmill walking. Simulations of footstep sounds upon surface materials such as concrete and snow were provided by means of a system composed of headphones and shoes augmented with sensors. In a control condition, participants could hear their actual footstep sounds. Results showed an overall enhancement of the forward drift after treadmill walking independent of the sound perceived, while the strength of the aftereffect, measured as the proportional increase (posttest/pretest) in forward drift, was higher under the influence of snow compared to both concrete and actual sound. In addition, a higher knee angle flexion was found during the snow sound condition both before and after treadmill walking. Behavioral results confirmed those of a perceptual questionnaire, which showed that the snow sound was effective in producing strong pseudo-haptic illusions. Our results provide evidence that the walking in place aftereffect results from a recalibration of haptic, visuo-motor but also sound-motor control systems. Self-motion perception is multimodal.

Appendix

The developed footstep engine uses a sound synthesis technique known as physical modeling, where the physics of sound production mechanism is simulated.

Specifically, we adopted the impact model described in Avanzini and Rocchesso (2001) and a physically informed sonic model (PhiSM) (Cook 1997). These models were used to simulate walking on solid and aggregate surfaces, respectively. The two approaches are briefly recalled below.

The interaction between solid surfaces can be expressed by the force between two bodies (Hunt and Crossley 1975):

$$ f\left( {x,\dot{x}} \right) = - kx^{\alpha } - \lambda x^{\alpha } \dot{x} \quad {\text{if}}\ x\,>\,0, 0\;{\text{otherwise}} $$

where x represents the contact interpenetration, k accounts for the material stiffness, λ represents the force dissipation due to internal friction during the impact, and α is a coefficient which depends on the local geometry around the contact surface. The model described has been discretized using the numerical method proposed in Avanzini and Rocchesso (2001).

In order to simulate particle interactions typical of aggregate surfaces, we adopted a PhiSM model. In this model, the interaction between the foot and the floor can be represented using a simple Poisson distribution, where the probability of sound production is constant at each time step, giving rise to an exponential probability weighting time between events.

In the experiment described in this paper, the footstep sounds synthesis is driven interactively by the user wearing the shoes. From the real acoustical signal of a footstep sound, the ground reaction force (GRF) is estimated, i.e., the reaction force supplied by the ground at every step. Such GRF is used to drive the described physical models, as explained in detail in Turchet et al. (2010a). A description of the control algorithms based on the analysis of the values of the pressure sensors embedded in the shoes can be found in Turchet et al. (2010b). The sound synthesis engine and the relative control algorithms were implemented using the Max/MSP sound synthesis and multimedia real-time platform.