The goal of my PhD project is to gain a better understanding of self-motion perception in humans. I am part of the Cybernetics Approach to Perception and Action research group and we use the CyberMotion Simulator to investigate the vestibular system. This six degree of freedom motion simulator allows freely positioning and orientating people within a large workspace. Hence, it offers a good opportunity to investigate self-motion perception under a wide variety of conditions.

My research interests include:

• assessing self-motion perception with psychophysical experiments
• modeling self-motion perception based on the anatomy and physiology of the vestibular system
• multi-sensory integration, e.g. visual – vestibular integration
• self-motion perception in patients suffering from bilateral vestibular loss

Currently the Max Planck Institute for Biological Cybernetics is a partner of a European Union project with the title “Simulation of Upset Recovery in Aviation” (SUPRA). The project deals with the simulation of situations where pilots get spatially disoriented and therefore lose control over the airplane. Part of the project is to extend the state-of-the-art motion perception models for humans to be able to also deal with such situations. Our research directly adds to this part.

Explaining Vestibular Perception with Physiologically Motivated Models

Collaborators: Paolo Robuffo Giordano, Michael Barnett-Cowan

Introduction

The vestibular system enables us to perceive the direction of (passive) self-motion even in absence of visual stimuli, e.g., in complete darkness. Our understanding of the vestibular system has been significantly advanced through the seminal work of Fernandez and Goldberg [1]. Describing the physiology of the system in squirrel monkeys, they provided a transfer function model which accurately predicts change in neuronal activity in response to vestibular stimulation. In order to investigate the system's functionality in humans, non-invasive methods are necessary. Perceptual thresholds provide a suitable alternative measure.

Goals

Our main objective is to understand vestibular direction detection thresholds, their dependency on the inertial motion stimulus, and to describe this dependency with a physiologically plausible model. Previous work [2] has shown that the duration of a displacement influences its detection threshold (measured in terms of the peak acceleration of the motion profile). However it has not been investigated how the specific time course of the motion, e.g. sinusoidal or trapezoidal acceleration profile, influences the threshold.

Methods

The MPI CyberMotion simulator was used to present blindfolded participants with translational or rotational motions. Participants had to indicate the perceived direction of motion choosing from a predefined set of possible directions. The duration and the time course of the motion profiles were varied and the acceleration threshold for detecting the correct direction was measured.

The average change in firing rate of a vestibular neuron was modeled as a function of the motion stimulus. Introducing a threshold criterion for the firing rate allows for predicting the necessary peak acceleration of a motion in order for the firing rate to overcome that threshold.

Results

In terms of the dependency on the duration of the motion profile our results are in agreement with previous findings. In addition, we show that thresholds also depend on the specific time course of the acceleration: for smooth, triangular accelerations thresholds are significantly higher compared to jerky, trapezoidal accelerations [3,4].

Our model is able to accurately describe measurements for translational and rotational movement (Fig.1) and the best fitting parameters agree with data from electrophysiological recordings in primates.

Conclusion

We proposed a physiologically plausible model able to describe vestibular perception thresholds. This model could assist the diagnosis of patients with vestibular problems, because, as opposed to current diagnostics, it does not rely on measuring eye movements. Additionally, the model could be used to improve control algorithms for motion simulators in order to optimally move the simulator below the perceptual threshold level.

Future work will investigate the interaction of translational and rotational motions to understand how the independent signals are integrated in the brain.

References

1.      Fernandez C, Goldberg J (1976) Physiology of peripheral neurons innervating otolith organs of squirrel monkey 3. Response dynamics, Journal of Neurophysiology 39(5) 996–1008.

2.      Grabherr L, Nicoucar K, Mast FW, Merfeld DM (2008) Vestibular thresholds for yaw rotation about an earth-vertical axis as a function of frequency, Experimental Brain Research 186(4) 677-681.

3.      Soyka F, Robuffo Giordano P, Beykirch KA, Bülthoff HH (2011) Predicting direction detection thresholds for arbitrary translational acceleration profiles in the horizontal plane, Experimental Brain Research 209(1) 95-107.

4.         Soyka F, Robuffo Giordano P, Barnett-Cowan M, Bülthoff HH (submitted) Modeling direction detection thresholds for yaw rotations around an earth-vertical axis for arbitrary motion profiles.

Fig. 1: Measurements and best fitting model response for translational (left) and rotational (right) motions. The model accurately describes the differences in thresholds due to changes in the acceleration profile shape and duration.

Multisensory Integration of Visual and Vestibular Cues during Yaw Rotations

Collaborators: Michael Barnett-Cowan, Ksander De Winkle, Eric Groen

In collaboration with scientist from TNO, Soesterberg (Netherlands) we investigated if humans integrate visual and vestibular information in a statistically optimal fashion when discriminating rotational self-motion stimuli. Participants were consecutively rotated twice (2s sinusoidal acceleration) on a chair about an earth-vertical axis in vestibular-only, visual-only and visual-vestibular trials. The task was to report which rotation was perceived as faster and just-noticeable differences (JND) were estimated by fitting psychometric functions. Predictions for the visual-vestibular JNDs were calculated based on the unisensory JND measurements and optimal integration theory.

The visual-vestibular JND measurements are too high compared to the predictions and there is no JND reduction between visual-vestibular and visual-alone estimates. These findings may be explained by visual capture. Alternatively, the visual precision may not be equal between visual-vestibular and visual-alone conditions, since it has been shown that visual motion sensitivity is reduced during inertial self-motion. Therefore, measuring visual-alone JNDs with an underlying uncorrelated inertial motion might yield higher visual-alone JNDs compared to the stationary measurement. Theoretical calculations show that higher visual-alone JNDs would result in predictions consistent with the JND measurements for the visual-vestibular condition.

References

1.      Soyka F, de Winkel K, Barnett-Cowan M, Groen E, Bülthoff HH (2011) Integration of visual and vestibular information used to discriminate rotational self-motion, 12th International Multisensory Research Forum (IMRF 2011), Fukuoka, Japan, i-Perception, 2(8) 855.

2.      de Winkel K, Soyka F, Barnett-Cowan M, Groen E, Bülthoff HH (2011) Multisensory integration in the perception of self-motion about an Earth-vertical yaw axis, 34th European Conference on Visual Perception (ECVP 2011), Toulouse, France, Perception, 40(ECVP Abstract Supplement) 183.

Integration of Translational and Rotational Vestibular Cues for Direction Detection during Eccentric Rotations

Collaborators: Paolo Robuffo Giordano, Michael Barnett-Cowan

During eccentric yaw rotations around an Earth-vertical axis the semi-circular canals are stimulated (rotational acceleration) as well as the otoliths (tangential acceleration). Most likely the brain uses both sensory signals, the canal and the otolith signal, when faced with a rotation direction detection task. Keeping the rotational acceleration profile unchanged and increasing the radius of the eccentric rotation the tangential acceleration increases. Therefore, we hypothesized that thresholds would decrease with increasing radius of rotation.

Participants were tested in seven conditions: a head-centered rotation, a translation and five eccentric rotations with varying radii (R=0.1, 0.2, 0.3, 0.5, 0.8 m). The motion had 1s duration consisting of a single cycle sinusoidal acceleration and the task was to judge the direction of the rotation. The threshold was defined as the peak acceleration needed to detect the correct direction of motion in 75% of the trials.

The results show a significant decrease of thresholds with increasing radius. It can be seen that the detection process for eccentric rotations is not exclusively based on either the canal or the otolith signal, but that both signals are integrated. A model able to predict the thresholds of the eccentric rotations is proposed, which is solely based on the thresholds for the head-centered rotation and the translational motion. For small radii the detection processes is mainly based on the canal signal whereas for large radii it is dominated by the otolith signal. For intermediate radii the reduction in threshold due to the sensory combination is largest compared to using only one of the two sensors.

References

1.      Soyka F, Barnett-Cowan M, Robuffo Giordano P, Bülthoff HH (O2011) Integration of translational and rotational vestibular cues for direction detection during eccentric rotations, 12th Conference of Junior Neuroscientists of Tübingen (NeNa 2011).