1. The percept of visual verticality during combined roll-pitch tilt Maurice Dahmen Student medical biology December 2006-July 2007 Supervisors: Maaike de Vrijer & Jan van Gisbergen

2. Contents Introduction Vestibular system: Otoliths Research objective Subjective visual vertical Bayesian model Methods Results Experimental data + model fits Discussion Summary and conclusions

3. Vestibular system Introduction

4. Vestibular system:otoliths Introduction Pitched orientation Sensitivity for roll and pitch

5. Introduction What is the role of the vestibular system in spatial perception? The otoliths can measure head tilt with respect to gravity. Research objective

6. Subjective visual vertical Systematic errors during SVV adjustment task Subjects in roll tilt set a luminous line parallel to the earth vertical Under- and overestimation of tilt (A- and E-effect ) Magnitude of errors differs among subjects Introduction

7. Errors in SVV-task (example) E-effect A-effect Introduction

8. Bayesian interpretations of A-effect Sensory tilt signal is noisy A priori information: head is mostly near upright Brain combines sensory information and prior to obtain optimal tilt estimate Model De Vrijer et al., 2007 tilt (ρ)

9. Bayesian model Adapted from Carandini, 2006 tilt (ρ) Model

10. Fit Bayesian model Model Fits of the Bayesian model to SVV data of 8 subjects were very accurate

11. Further test of the Bayesian model Model predicts that a noisier tilt signal leads to a more biased SVV (larger A- effect) We used pitch tilt to modulate the noise in the roll tilt signal Model

12. How does pitch-tilt affect the pattern of systematic errors in SVV during roll tilt? Larger A-effect ?Normal A-effectSmaller A-effect ? Research question

13. Vestibular chair: pitch Methods -45°0°0°45°

14. Vestibular chair: roll Methods

15. Experimental setup 8 subjects (6 male, 2 female) In same pitch position during entire session Tilted to the various roll angles in complete darkness Roll-tilt varied from –90 to +90 at 15 degree intervals 20 seconds waiting time to extinguish canal signals SVH adjustment task Back to upright position Room lights on Methods

16. SVH adjustment task Subjects used a joystick to adjust the orientation of the line The line was polarized by a bright dot at one end. Subjects were instructed to set the line parallel to the virtual horizon with the dot pointing rightward A period of 12 seconds was available for each adjustment There were 10 adjustments during a run Methods

17. Estimation of orientation of utricle plane Reid’s plane: The plane passing through the inferior margin of the ocular orbits and the center of the external auditory canals. Angle between Reid’s plane and utricle plane: 25º (Blanks, Curthoys and Markham, 1975) Methods

18. Subject 1 Results Results as expected

19. Subject 2 Results Results as expected

20. Subject 3 Results Large E-effect, not expected

21. Pooled data Results

22. The Mittelstaedt-model Problem: E-effects cannot be explained by Bayesian model! Alternative: The Mittelstaedt-model Discussion

23. Introducing the Mittelstaedt-model Discussion

24. Fitting the M-model Parameters S: 0.42 M: 0.70 Discussion

25. Fitting the M-model Parameters S: 0.36 M: 0.24 Discussion

26. Fitting the M-model Discussion

27. Summary and conclusions The Bayesian model cannot fit our data (because of E-effect) A-effects become larger when subject is in backward pitch Mittelstaedt-model can fit our data Further explorations are essential to fully understand the model.

28. Questions?