1. Verticality perception during body rotation in roll
Rens Vingerhoets
Pieter Medendorp
Jan van Gisbergen
NICI & Department of Biophysics
Nici-Juniorendag, 3 mei 2007

2. Introduction

3. Perceptual updating Introduction
The brain has to combine retinal information and information on body orientation to maintain a stable percept of the world

4. Perceptual updating Introduction Information on body orientation from:
Visual panoramic cues
Somatosensory cues
Vestibular system:
Semicircular Canals
Otoliths

5. Introduction
Vestibular System

6. Semicircular canals Introduction Detect angular acceleration
Response to constant velocity decays

7. Otoliths fa g a Introduction Respond to gravito-inertial force (GIF)
Cannot discriminate between tilt and translation (ambiguity problem) fa g
F=G-A a

8. Otoliths Introduction Ambiguity problem:
Neural strategy for otolith disambiguation:
Canal-otolith interaction

9. Canal-otolith interaction model
Introduction
Canal-otolith interaction model b w
Subjective
vertical
GIF +
Otoliths
Internal
Model
Linear
acceleration
Angular
velocity
Angular
velocity
Canals
Head tilt leads to canal signal, acceleration does not

10. Canal-otolith interaction model
Introduction
Canal-otolith interaction model b
Bias mechanism w
Subjective
vertical
GIF + g
Otoliths
Internal
Model
Linear
acceleration
Angular
velocity
Angular
velocity
Canals
Bias mechanism based on static verticality estimates
Now test this model for dynamic verticality perception during roll rotation

11. Roll-rotation Introduction Sideward rotation about
an axis through the nose

12. g = Error in verticality percept
Introduction
Model predictions
g = Error in verticality percept g 90
180
270
360
SVV Error,  (deg)
Tilt angle (deg)
-180
-90
90L
90R
Static g
Preceding rotation (deg)
180
360
540
720
900
1080
-180
-360
-540
-720
-900
-1080
90L
90R
Dynamic

13. Questions Introduction
Are systematic errors under dynamic conditions similar to static errors?
Evidence for phase delay in the verticality percept as predicted by the internal model?

14. Methods

15. Methods
Vestibular Chair

16. Experiments Methods Subjective visual vertical (SVV)
Roll rotation at 30 deg/s, alternating CW and CCW
Measurements at 15 deg intervals
Scaling method using flashed lines
Static:
0 – 360 deg
10 measurements 30 s after rotation stop, every 2 s
Dynamic:
0 – 1080 deg = 3 complete cycles
measurements during rotation, every 2 s
Subjective body tilt (SBT)
Verbal estimation of body tilt at random times during rotation

17. Scaling method Methods Flash! 1 minute past the hour!
Error in SVV = real orientation (30o) – estimated orientation (6o) = 24o
Response error equals error in verticality percept (g)

18. Results

19. Static CW Results Systematic errors
90R
180
90L
180
Systematic errors
0 –150o & 240 – 360o verticality percept biased towards the head
Bias toward feet near 180o 90
Error in SVV (deg)
-90
-180 90
180
270
360
Tilt angle (deg)

20. Preceding rotation, Dr (deg)
Results
All subjects
Error in SVV, g (deg)
Preceding rotation, Dr (deg)

21. Comparison Static & Dynamic

22. Preceding rotation, Dr (deg)
Results
Static vs Dynamic
Error in SVV, g (deg)
Preceding rotation, Dr (deg)

23. Preceding rotation, Dr (deg)
Results
Static vs Dynamic
Error in SVV, g (deg)
Preceding rotation, Dr (deg)

24. Static vs Dynamic Results Dynamic data qualitatively similar but:
Larger errors in red zone
More inter-subject variability
Green zone is broader
Bistability

25. Dynamic: complete rotation

26. Dynamic: complete rotation
Results
Dynamic: complete rotation
Error in SVV, g (deg)
Analyse per proefpersoon is essentieel. Populatie gemiddelde is onzin.
Preceding rotation, Dr (deg)

27. Dynamic: complete rotation
Results
Dynamic: complete rotation
Error in SVV, g (deg)
Analyse per proefpersoon is essentieel. Populatie gemiddelde is onzin.
Preceding rotation, Dr (deg)

28. Dynamic: complete rotation
Results
Dynamic: complete rotation
Error pattern repeats in successive cycles
No phase delay

29. Dynamic Subjective Body Tilt

30. Results
Dynamic SBT
Error in SBT,  (deg)
Preceding rotation, Dr (deg)

31. Results
Dynamic SBT
Error in SBT,  (deg)
Preceding rotation, Dr (deg)

32. Dynamic SBT Results Smaller errors than in SVV No bistability
Errors in SVV are not caused by errors in SBT

33. Dissociation between SVV & SBT
Results
Dissociation between SVV & SBT
Errors in verticality estimates not caused by misjudgment of body tilt
SVV ideal subject
SVV real subject
Errors based on SBT

34. Model Fits

35. Model Fits Results SVV in red zone biased toward head  w>0
SVV in green zone biased toward feet  w<0
Dynamic errors larger than static wdyn > wstat

36. Preceding rotation, Dr (deg)
Results
Static Fits
Error in SVV, g (deg)
Preceding rotation, Dr (deg)

37. Preceding rotation, Dr (deg)
Results
Dynamic Fits
Error in SVV, g (deg)
Preceding rotation, Dr (deg)

38. Model Fits Results Model can describe SVV responses if:
W-values differ for static and dynamic
W-values differ in red and green zone

39. Analysis of phase shifts
Results
Analysis of phase shifts
Instead of focusing on the error g, we can also look at the amount of compensation for tilt. We refer to this angle as b.
= compensation for rotation (b = tilt - g)
Perfect task execution: b = body tilt angle g b

40. Analysis of phase shift
Introduction
Analysis of phase shift
Model prediction for b
Dynamic CW
360
Actual tilt
315 b
270
225
Tilt Angle (deg)
180
135 90
Phase shift:
= 0 when Dr ≈ 370 & Dr ≈ 735 45
180
360
540
720
900
1080
Preceding rotation, Dr (deg)

41. Analysis of phase shifts
Results
Analysis of phase shifts
1 Subject
 (deg)
50o
-720 o
-360 o o
360 o
720 o
50o
Model
 (deg)
50o
50o
All subjects
Preceding rotation, Dr (deg)
Preceding rotation, Dr (deg)

42. Analysis of phase shifts
Results
Analysis of phase shifts
No clear evidence for phase lag
If anything, it is a lead rather than lag

43. Conclusions

44. Conclusions Conclusions We have shown that:
Errors in verticality perception are not caused by misjudgments of body tilt
The egocentric bias is larger under dynamic conditions than under static conditions
Verticality judgments are biased toward the feet around 180o tilt. Both statically and dynamically
There is no evidence for a phase delay