Self-motion perception is explored in several different ways at the Max Planck Institute for Biological Cybernetics. First, the perception of self-motion while moving through a space is investigated from a multi-sensory integration perspective looking at the contributions of visual, vestibular, and proprioceptive information. A second focus is on investigations of the illusion of self-motion - that is, how can we fool the user into believing that they are moving without having expensive technology to actually move them. Finally, self-motion perception during walking is studied with the specific aim of better simulating walking within an infinite plane through the development and testing of an omni-directional treadmill.

◘ Human locomotion and gait parameters
◘ Perception of self-motion in Virtual Reality
◘ Spatial updating during movement in two and three dimensions
◘ Perception and production of speed during self-motion
◘ Bayesian integration of visual and vestibular information
◘ Vection in a Large Screen Immersive Virtual Environment
◘ Vestibular Direction Detection Thresholds in the Horizontal Plane

Bayesian integration of visual and vestibular information

Self-motion through an environment involves a composite of signals, including visual and vestibular cues. It has been shown that visual-auditory cues and visual-haptic cues combine in a statistically optimal fashion [1] but very little is currently known about visual-vestibular cue integration.

Our goal - we investigated the relative weights of visual and vestibular cues during self-motion. Further, we tested the limits of this cue integration by creating spatial [2] and temporal offsets between the two cues, providing different acceleration profiles to the two cues [3], and by changing the immersive nature of the visual input (i.e. 2D vs. 3D) [4].

Participants performed a 2-interval forced choice task and were asked in which of two movement intervals did they moved more to the right (see Figure 1). From these responses we plotted psychometric functions from which we extracted the participants’ variance (reliability). We predicted the combined cue responses and the weights of the individual cues from the responses in the unimodal conditions [1]. To observe the weights assigned to each modality and to assess the limits of cue integration, we introduced either a spatial, temporal or motion profile discrepancy between the visual and vestibular cues in the standard visual-vestibular heading.

Our results show that visual-vestibular cue combination is robust, such that when there are conflicts between the visual and vestibular cues; participants exhibit a statistically optimal reduction of variance. Furthermore, we found that the unimodal cues did not predict the weights in the combined cue but that there is a prior which lends more weight to the vestibular cue. Finally, we found visual-vestibular cue integration breaks down when the visuals are not presented in stereo.

REFERENCE
◘ Ernst, M.O. and Banks, M.S.:Humans integrate visual and haptic information in a statistically optimal fashion., Nature 415, 429-433 (2002)

◘ Butler, J.S., Bülthoff, H. H. and Smith, S: Bayesian integration of visual and vestibular information., (Submitted)

◘ Butler, J.S., Campos, J.L. and Bülthoff, H.H.:: The robust nature of visual-vestibular combination for heading., ECVP, 1 (08 2008)

◘ Butler, J.S., Bülthoff, H. H. and Smith, S: : The role of stereo vision in visual and vestibular cue integration., (Submitted)

Figure 1: Procedure for the spatial
offset experiment. Participants were
translated along two heading intervals
and had to judge in which movement
they moved more to the right.
In this example, the first movement
is the comparison, the second movement
is the standard in which the vestibular
motion is straight ahead (0°) and the
visuals are offset by 6 degrees.

Figure 2: Apparatus. Participants
were seated on the MPI Stewart
motion platform facing a
projection screen which they
viewed through an aperture,
creating a field of view of 50°×50° .