Modeling Self-Motion Perception based on the Vestibular System

Understanding self-motion perception is important, e.g., for developing perception-based diagnostics for patients suffering from vestibular disorders, or for improving the realism of motion simulation. Here, self-motion perception was quantified with direction discrimination thresholds and reaction times for translational and rotational motions. Models based on the physiology of the vestibular system, which plays a central role in sensing inertial motions, are introduced that are able to describe psychophysical measurements. The Max Planck Institute CyberMotion Simulator was used to measure thresholds for 9 translational and 9 rotational motion stimuli with varying acceleration profiles. A forced-choice paradigm was used in which blindfolded participants had to judge the directions of motions with varying peak accelerations until the threshold acceleration was found that yielded a predefined performance level (e.g. 75 correct answers). A similar task was used with supra threshold accelerations in order to measure reaction times for 4 translational and 4 rotational motions. The results show that thresholds and reaction times depend on the actual shape and duration of an acceleration profile. The proposed models were fit to threshold measurements and are able to describe thresholds for arbitrary motion profiles. In accordance with previous research, the estimated model parameters indicate that velocity storage does not influence rotational thresholds. For translational motions, it was found that the sensitivity to jerk (the time derivative of acceleration) is higher than previously assumed. Furthermore, the models identified based on threshold measurements are able to predict differences between reaction times for varying motion profiles measured in another group of participants. This is an important finding, because it links reaction times and threshold measurements. Therefore, future research will be able to identify self-motion perception models based on reaction times. This is advantageous, since reaction time tasks are more convenient for participants and require fewer trials, allowing for faster testing