Journal of Vestibular Research, Vol. 2, pp. 15-30, 1992 Printed in the U$A. All rights reserved.

0957-4271/92 $5.00 + .00 Copyright © 1992 Pergamon Press Ltd.

INFLUENCE OF VISUAL, VESTIBULAR, CERVICAL, AND SOMATOSENSORY TILT INFORMATION ON OCULAR ROTATION

AND PERCEPTION OF THE HORIZONTAL

Bernd de Graaf, * Harold Bekkering, * Corrie Erasmus, t and Willem Bles * t *TNO Institute for Perception, Soesterberg; tFree University Hospital, Amsterdam, The Netherlands

Reprint address: Bernd de Graaf, TNO Institute for Perception, P.O. Box 23, 3769 ZG Soesterberg, The Netherlands

o Abstract - By combining a tilting chair and a tilt­ing room we investigated the subjective horizontal (SH) and ocular counterrotation (OCR) as a func­tion of body tilt, trunk tilt, and tilt of a visual frame. Significant influences of (isolated or com­bined) vestibular and visual information were found, but no influence of neck proprioception. A second and similar experiment, however, now con­ducted with subjects devoid of labyrinthine func­tion, suggested a contribution of the neck as well as of somatosensory origin. This made a reinter­pretation of our data for nQ..rmal subjects possible.

D Keywords - multisensory tilt information; subjective horizontal; ocular counterrotation.

Introduction

A lot of research demonstrates the strength of a stable visual frame on the appearance of the subjective vertical (SV) or subjective horizontal (SH). Subjects, even when exposed to a severe body tilt, are able to set a line horizontal when adequate visual information is available. In the absence of such a frame, however, for instance in the dark, considerable illusions oc­cur. Early researchers (l ,2) noticed effects of head tilt on the subjective verticaL nowadays called the A (Aubert) and E (Muller) phe­nomena. The A phenomenon is described (3) as the apparent tilt of a truly vertical line in

lit is therefore not always correct to maintain the term OCR (ocular counterrotation). In cases where this could lead to a misunderstanding, we will use the term ocular rotation.

the direction opposite to the head tilt, so that the subjective vertical is tilted in the same di­rection as the head tilt. The E phenomenon is described as the apparent tilt of a true verti­cal in the same direction as the head tilt, and therefore the subjective vertical is tilted in th~ direction opposite that of the head. The A phenomenon is what one would expect if c subject underestimated the extent of the heac tilt, and the E phenomenon is what on( would expect from an overestimation of th(

-head tilt. Muller found that most people ex perience the E phenomenon with small angle: of head tilt and the A phenomenon witl larger angles. Some people experience only th, A phenomenon, however. Witkin and Ascl (4) observed that of their subjects 770/0 pro duced an E phenomenon and 23 % an A phe nomenon while exposed to a 28° body tilt an~ that with a body tilt of 90° almost all (94% of the subjects experienced theA phenomenon

When the head (or body) is tilted to on side, the eyes rotate in their orbits about th visual axis in the opposite direction. This IE

sponse is known as ocular counterrotatio (OCR). The static (residual) OCR of health persons is about 10% of body tilt, with maximum of 6° at 60 c to 75(; body tilt (5,6 If our perceptual system failed to take th counterrotation into account, the resulting e ror would be in the same direction as the phenomenon. It is therefore tempting to COl

elude that the OCR is responsible for the phenomenon, but there is evidence that it not that simple. Fischer (7) measured OC

RECEIVED 15 November 1990; REVISED MANUSCRIPT RECEIVED 10 July 1991; ACCEPTED l7 July 1991

15

16

and the apparent tilt of a vertical line as a function of body tilt, but found no direct or even proportional relationship. Miller and colleagues (6) replicated these findings with normal subjects, and they also found E phe­nomena in people who showed no OCR be­cause of vestibular malfunction.

The physiological processes responsible for the OCR and the setting of the SV or SH in darkness are not quite clear. Ivliller and col­leagues (8) showed that the function relating (he magnirucie 0f lnt 1:, anG A LO

the degree of body tilt resembles a sine func­tion. This made Schone and Udo de Haes (9) suggest that the illusions are related to the shearing force acting on the macular surface of the utricule, because this is also a sine func­tion of body tilt. Evidence inconsistent with this utricular force theory comes from exper­iments with people with loss of vestibular function. As stated above, they also show E and A phenomena. But, it may be that substi­tution of other sensory information to com­pensate for the loss of vestibular function, like neck proprioception and somaesthetic cues, is responsible for these observations (10).

Collewijn and colleagues (11) used their precise scleral coil technique to measure OCR during the dynamic phase of head roll. They found an OCR from 40070 to more than 70070 of head motion. This high amount of com­pensation was probably due to activation of the semicircular canals. In situations of con­tinuous rotation at constant velocity (and therefore without canal stimulation), Dia­mond and Markham (12,13) showed a con­stant OCR of about 10070 of head rotation. With static tilt of the head lasting several hours it was found (14) that OCR is not re­duced, which provides evidence that static OCR is driven by nonadapting utricular receptors.

Other factors contributing to OCR, such as visual information and neck propriocep­tion are controversial. Goodenough and col­leagues (15) found torsional movements of the eyes of about 1 0 when upright subjects in­spected a tilted visual framework. Collewijn and colleagues (11) also reported a small but significant influence of visual cues on (dy­namic and static) OCR. Howard and Temple-

B. de Graaf et al

ton (16), however, found no evidence of eye torsion when subjects looked at a tilted line that subtended 100 at the eye. With respect to the neck, Krejcovi and colleagues (17) found that the amplitude of OCR produced by a tilt of the head alone is the same as that from the whole body, suggesting that there is no signif­icant contribution of neck proprioception to OCR. But others (18-21) suggested that an additional input from neck proprioception is likely.

AS Wl[h respeCI ;:0 possible cues of somato­sensory origin, experiments with subjects without labyrinthine function showed that pressure between body and tilting device could generate some OCR, but this would be very small in amplitude (maximally 10; [22]).

In the present experiment on body tilt we investigated the isolated or combined influ­ence of a) information from a visual frame, b) vestibular information, and c) information from neck proprioceptors on eye rotation and subjective horizontal. A tilting chair, which made independent adjustments of head and body tilt possible, enabled us to investigate not only the vestibular but also the possible cervical influence on OCR and SH in the dark. In the light, with this chair placed in a tilting room we could investigate the possibil­ity of a visually induced eye rotation as well. We could also measure the possible contribu­tion of the visual frame beyond the vestibular component both on OCR and SH by expos­ing the subject to one particular body tilt (with respect to gravity) but with different combinations of frame (room) tilt. The range of tilts was up to 25 0

, to the left or right. It was examined with small steps of 50 to pro­vide insight into the course of the variables.

Methods

Apparatus

For this experiment we used a motor-driven tilting room (Tonnies, Freiburg), which could be tilted laterally from the base. The device (2.5 m x 2.5 m x 2 m) was completely closed, except for a door in the front side. The walls,

OCR and Subjective Horizontal in Normal and LD

painted in flamingo red, were fitted in a black frame. The room tilt was indicated by a dig­ital voltmeter (10° to the left up to 10° to the right, accuracy ± 0.1°). Two ceiling lights illuminated the room dimly (35 to 65 lux at the back wall). In this room we installed a tilt chair, by which a subject could be tilted about the x-axis through the neck, up to 15° to left or right. This could be done in the following ways: a) with the whole body tilted (head and trunk in line) or b) with only the trunk tilted (head upright). Foamrubber cushions were placed 'between the subject and the chair to diminish somaesthetic inputs. Figure 1 illus­trates the experimental apparatus. The sub­ject could indicate the subjective horizontal by means of a LED-illuminated, potentiom­eter driven, rotatable test-bar (length 50 cm) at 125 cm distance at the back wall of the room (22.6° of visual angle; axis of rotation at eye level). A removable fluorescent lamp

17

(length 50 cm) could be placed 10 cm above this test-bar in order to create an afterimage. We had two dependent variables: (1) the de­viation of the subjective horizontal (SH) with respect to the objective horizontal and (2) the amount of eye rotation in the orbit.

With regard to variable 1, subjects were in­structed to set the test-bar in agreement with what they thought was horizontal. We mea­sured the deviation (in degrees) of this adjust­ment from the objective horizontal. With regard to variable 2, subjects were told to put the test-bar parallel to their afterimage (which was created before, with the subject in the normal upright position). The difference be­tween the actual amount of head (or body) tilt and the deviation of the test-bar (from the objective horizontal) quantified the eye rota­tion. The 12 paid observers, 7 male and 5 fe­male, were between 19 and 32 years old. They were naive with respect to the apparatus, the

Figure 1. A photo of the experimental apparatus. The room and the chair could both be tilted laterally. The chair could be tilted independently, but a tilt of the room also causes a tilt of the chair (it is however possi­ble to compensate this by adjusting the chair). With the help of a lever on the chair, it is also possible to tilt the trunk of a subject independent of the head or vice versa. The head position is controlled by means of an adjustable headrest on the lever.

18

experimental paradigm, and the purpose of the experiment.

Procedure

The experimenter showed the subject the tilting room and indicated how the chair and room could tilt independently. Then the sub­ject was seated, facing the back wall of the room, and given the instructions. In the SH measurement conditions lilt \\! as asked. to level the bar with respect to the true hori­zontal, independent of the tilts of the room and chair, so "that a little ball placed on the test-bar will not fall off." In the OCR mea­surement conditions, the subject was asked to adjust the bar parallel to the afterimage. The circumstances were identical for the SH and OCR conditions, except that in the OCR con­ditions, just before the chair and room were tilted, an afterimage was created by having the subject fixate the fluorescent lamp for 40 s. Then, for both the SH and OCR condi­tions, the subject was asked to close the eyes. The experimenter then first adj listed in ap­proximately 10 s the desired (randomized) an­gle of the chair. Thereupon the room was tilted (0°, 5°, or 10° to the left or right, in randomized order). After each of the tilts of the room, we switched the test-bar by turns about 20° to the left or to the right, and then asked the subject to open the eyes and adjust the test-bar. After such a sequence of room tilts (with one particular chair tilt), which lasted about 5 minutes, the subject got a few minutes rest. Then the chair was adjusted to a new position and the room tilt sequence was repeated. The order ofSH and OCR sequences of room tilt (with one particular chair tilt) was balanced. A nested variable was the light sit­uation in the room, because some measure­ments were taken in the dark. Figure 2 shows the experimental design. This design was run twice, once for SH measurements and once for OCR measurements.

In addition, we measured the influence of cervical stimulation on the SH and OCR in darkness by tilting the trunk of the subject while leaving the head upright. For that pur­pose the chair was tilted in random order up

B. de Graaf et al

til t of room (deg)

cnOcff r:--

-10 -5 5 10 lights

-15 a a

C f a a

-25 -20 -15 -15 -10 -5 off

C b -10 -20 -15 -10 -5 0 on

CJ) C b Qi -5 ~

-15 -10 -5 0 5 on ~

a 0 ..c u

d d bOT i d d

-10 -5 o 0 5 10 on

'+-0

5 b C

-5 0 5 10 15 on _. ..;::; I

b i C

10 0 5 10 15 20 on

15 a a

~51~5 a a

5 10 20 25 off

tilt of trunk (deg)

-15 -10 - 5 0 5 10

Figure 2. Experimental Design: The combination of tilting chair and tilting room enabled us to investigate the isolated or combined influence of visual informa­tion, vestibular information, and proprioception from the neck, on ocular rotation and subjective horizontal. The design offers the input variables: Upper inset of the figure. Each value inside the matrix represents the vestibular input, being the sum of the tilt of the chair and the tilt of the room (+ = clockwise tilt, - = counter­clockwise tilt). With the lights on, the room could act as a visual frame of reference. a = vestibular informa­tion only. b = invariant vestibular, but different visual information. c = invariant visual frame information, with respect to the gravitational vertical. Vestibular in­formation varies. d = vestibular information varies, but the visual angle between the subject and the room is always 0°. N.B. Three conditions were measured both with lights on and with lights off, in order to obtain ad­ditional information. Lower inset of the figure. To in­vestigate a possible cervical influence on OCR and SH, we tilted the trunk of the subjects, by tilting the chair, while the head remained upright. These mea­surements were performed in the dark.

The design was run twice, once for SH and once for OCR measurements, but in different (randomized) order.

to maximally 15 ° to the left or right, and the room remained in horizontal position (see also Figure 2).

Results

None of the subjects had difficulties with the adjustments of the test bar in OCR and

OCR and Subjective Horizontal in Normal and LD

SH conditions. Therefore, the following anal­yses could be performed on the data gathered from 12 subjects.

OCR

The data obtained in the dark (see Fig­ure 3) show a strong vestibular influence on OCR (ANOVA: F= 64.5; dj = 10,110; P < 0.001, 810J0 variance explained). The ampli­tude of OCR increased with larger body tilts, with a mean maximum of 6° with tilts of 25° to right or left. Post hoc Newman-Keuls analysis (23) revealed significant differences (P < 0.01) in OCR between 0° tilt of the body and 5°, 10°, and 15° to the right or to the left, respectively. The latter tilts on their turn caused smaller OCRs than body tilts of 20° and 25°.

There also exists a small but significant vi­sual influence on OCR (Figure 4a). In the light, different tilts of the room, while the body was not tilted, revealed a difference in ocular rotation in the same direction as the room tilt (ANOVA: F = 7.9; dj = 4,44; P < 0.001, 26070 variance explained)!. The mean maximum ocular rotation was about 1 ° with

12

10

8

6

4

C, 2 QJ

E 0

'-u

-2 0

-4

-6

-8

-10

19

a room tilt of 10° to the left or to the right. We could also examine the influence of visual information on ocular rotation on other occa­sions. For example, in the sequence of tilts where the subject was visually locked with the tilting room (situation d, see experimental de­sign, Figure 2), we found a significantly smaller OCR than with the same amounts of body tilt in the dark (situation a, see also Figure 2) (ANOVA: F = 11.5; dj = 1,11; P < 0.01, 10070 variance explained, see Figure 4b). Ob­viously, a (tilted) visual frame could have a restraining influence on the OCR of a tilted subject. Also, in another situation in which we measured the same condition twice (in the light and in the dark), when the body was tilted for 15° to the left, but with an upright (0° tilted) room, in the light, we found a sig­nificantly larger OCR than in the dark (t test: t = 1.82, df = 22, P < 0.05). This was, how­ever, not the case in a comparable situation in which the body was tilted 15° to the right (too much deviation in the response). On the ba­sis of these results, we conclude that there ex­ists a small visual influence on ocular rotation which could add to or subtract from a much larger vestibular component.

We found no significant influence of neck

'~' . . : . . .

body pos it ion

Figure 3. Ocular counterrotation (OCR) as a function of body position in the dark (stituation a, Figure 2). A posi­tive value stands for a tilt (x-axis) and rotation (y-axis) clockwise, a negative value represents a tilt (x-axis) and rotation (y-axis) counterclockwise. The data show a strong vestibular influence on OCR. The line drawn is a best fit 3rd degree polynome which levels at ±6°.

20

c a .,..,

.oJ cu .oJ a

12

10

8

6

4

2

o

B. de Graaf et al

~;:: . "

: :.

c.. -2 c.. cu .-; -4 :J U

a -6

en OJ ~

c. u 0

-8~ 1 -lot I

~-~~~5---~~~0---~~~5~--1~'O~ __ ~'5~~6--~~~~110~~11=5--~2~'O--~2~'5~

room pas it ion (deg)

12

10

8

6

4

2

0

-2

-10

(a)

position

(b)

Figure 4. (a) Ocular rotation as a function of room position in the light (situation b, Figure 2). The data show a small but significant visually induced ocular rotation in the same direction as the room tilt. (b) OCR as a func­tion of body position, as measured in the light (open circles; situation d, see Figure 2) and in the dark (filled circles; situtation a, see Figure 2). In the situtation in which subject and room are visually locked, that is, tilted '(0 the same amount, the visual frame had a restraining influence on OCR.

proprioception on OCR (Figure 5). Different angles of body tilt, but with the head upright with respect to gravity, did not produce dif­ferences in the OCR of subjects, The values of the OCR unsystematically balanced around 0, with larger standard deviations at the ex­tremes of tilt.

Subjective Horizontal

Figure 6 reproduces the mean adjustments of the subjective horizontal. A comparison with the input variables given in Figure 2 shows that, in the light, the room acts as a frame of reference and dominates the adjustment of

OCR and Subjective Horizontal in Normal and LD 21

12

10

8

6

4

Oi 2 CIl

.s I 0 :.----

c... -u

-2 0

-4

-6

-8

-10

-25 -20 -15 -10 -5 0 trunk pos it ion (deg)

Figure 5. OCR as a function of trunk position. No significant influence of neck proprioception on OCR was found. The values of OCR balanced around 0, with larger standard deviations at the extremes of tilt of the trunk with respect to the head.

the horizontal. The response was always de­termined by the room, despite differences in vestibular information (ANOVA: F = 48; dj = 4,44; P < 0.001, 690/0 variance ex-

tilt of room (deg)

on 0 off r---

-10 -5 5 10 lights

-15 3.1 1.4 -0611.7 00 1.4 off

-10 -7·8 - 3.1 0.7 4.4 8.2 on

CJ)

-5 Q) -7.4 - 3.1 0.6 4.3 7.8 on '"0

!....

0 0 1:; u

-7.3 -3.7 03fo: 3.9 8.3 on

'0 5

I I ~ ~ I ~ c r.I"",

I

-b.D - ju 1.1 L.U 1.1 '-";;

10 -7.7 - 4.6 -10 2.9 6.3 on

15 I-tO - 0.6 0+- - 05 -1.5 off

Figure 6. Mean SH output data presented like the de­sign of Figure 2. The measurements taken with the lights on clearly show the dominance of a visual frame of reference on the adjustments of the subjective horizontal.

plained. See Figures 2 and 6, any column). Also, in situations with invariant vestibular input (like in situation b, see Figure 2) re­sponses were significantly different.

The mean SH adjustments in the dark, merely determined by vestibular information (situation a in Figure 2), did not significantly differ from each other or from O. See Figure 7.

We found no influence of neck proprio­ception on the adjustment of SH (Figure 8). Different angles of trunk tilt, but with the head upright with respect to gravity, did not produce differences in the SH of subjects. The values of the SH unsystematically bal­anced around O.

Discussion

We found that ocular counterrotation is mainly determined by vestibular information. With angles of tilt up to 25° we found a (static) OCR with a progressive course that will pre­sumably saturate by 20° to 25 ° of body tilt, at a maximum of 6°. The range of the gain OCR with a progressive course that will pre­sumably saturate by 20° to 25° of body tilt, at a maximum of 6°. The range of the gain

22 B. de Graaf et al

12

10

Cl 8 Q)

E 6 ...... III 4 +J C 0 2 N ..... c... 0 0 .c. (l) -2 > ..... +J u -4 (l)

0 -6

~ -8~

-10

I I I I I I I I I I

-25 -20 -15 -10 -5 0 5 10 15 20 25 body DOS it ion (deg)

Figure 7. Adjustments of SH as a function of body position. A positive value stands for a tilt (x-axis) and test bar adjustment (y-axis) clockwise, a negative value represents a tilt (x-axis) and adjustment (y-axis) counter­clockwise. The mean data obtained in the dark and, therefore, merely determined by vestibular information (situation a, see Figure 2) did not differ significantly from each other or from O. The line drawn is a best fit 3rd degree polynomial.

was 0.23 to 0.48, which is in correspondence with OCR data taken from the literature by Vogel and colleagues (24) and conformable to their logarithmic description of the relation between OCR and tilt.

12

10

Cl 8 Q)

E 6 ...... III 4 +J C 0 2 N .....

We found a small but significant visual influence on OCR, which could add to or subtract from the vestibular component, de­pending on the position of the subject with respect to the visual frame. These results are

c... 0 ~ 0 ~ .c. (l) -2 > ..... +J u

-4 (l)

..0 -6 ::J (IJ

-8

-10

-25 -20 -15 -10 -5 0 trunkposition

5 10 (deg)

15 20 25

Figure 8. SH as a function of trunk position. No significant influence of neck proprioception on SH was found. The values of SH did not significantly differ from each other or from O.

OCR and Subjective Horizontal in Normal and LD

in quantitative agreement with earlier results (11,15).

Our data did not support the suggestion that there is a cervically induced contribution to OCR.

Although the subjects were told that the room could and would be tilted, they still ad­justed their subjective horizontal according to the room. The fact that people, for example, tilted 10° to the right adjusted their SH ac­cording to a room that is even 5 ° more tilted to the right (15° to the right) says a lot about the dominance of the room as a visual frame of reference. After our measurements, the subjects explained that they were sometimes aware of a discrepancy between visual and vestibular information, but nevertheless expe­rienced the visual information to be more valid.

In the dark, however, the mean of our data as shown in Figure 7 suggests that peo­ple make accurate judgements of horizontal. But individual analysis shows that this is cer­tainly not true. Of the 12 subjects, 5 always exhibited an E phenomenon (overestimated their body tilt, see Figure 9a), and 3 subjects always exhibited a specific A phenomenon (underestimated the tilt, see Figure 9b) (The remaining 4 subjects were not so clearly spe­cific; 3 of them tended to an E phenomenon, and 1 tended more to an A phenomenon.) Now it is clear that the SH of the individuals deviates considerably with larger tilts, but in a very specific way. These findings replicate the observations of, for example, Witkin and Asch (4). A close comparison of OCR and SH data obtained in the dark in identical con­ditions suggests a stringent relation between OCR and the Muller (E) phenomenon. A model which deals with this matter is pre­sented in the Appendix.

We found no signs of a systematic influ­ence of activation of the neck on the adjust­ment of SH. Subjects who had their trunks tilted sidewards, but with their heads upright with respect to gravity, did not experience a SH different from normal, with both head and trunk upright.

Although foam rubber cushions were placed between the subject and the chair to diminish somaesthetic inputs, it is not un-

23

thinkable that pressure between body and chair could have had influence on OCR and SH as well. To investigate this latter possible source of information and to isolate the other sources of interest even more, a second exper­iment was performed, similar in design, but now with people without vestibular function.

EXPERIMENT 2

Four subjects with bilateral loss of vestibular function were examined on the occurrence of OCR and SH with part of the experimental setup as used for the healthy subjects. This was done to provide insight into which infor­mation is still available for these subjects and how they use it to compensate for the absent vestibular otolith information. Because the experiment is reported elsewhere in detail (25), only the essentials relevant for our present purposes are mentioned here.

Methods

Apparatus

For this experiment, again, combinations of room and chair adjustments were used. This was done with the whole body tilted (head and trunk in line, condition WHOLE­BODY), with only the head tilted (trunk up­right, condition HEAD), or with only the trunk tilted (head upright, condition TRUNK). By combining, in darkness, the different chair operation modes with the proper room tilt in condition WHOLEBODY, a body tilt from 25° counterclockwise up to 25° clockwise could be achieved. In conditions HEAD and TRUNK, the maximally possible tilt of, re­spectively, the head or trunk was from 15° counterclockwise up to 15 ° clockwise. An ad­ditional condition was included in which the room was illuminated and the tilt of the sur­round relative to the upright sitting subject was possible from 10° counterclockwise up to 10° clockwise (condition VISUAL).

In subjects without labyrinthine function, this means stimulation of the somatosen­sory system in condition WHOLEBODY, of the neck in condition HEAD, of the neck mi­nus the somatosensory system in condition TRUNK, and of the eyes in condition VISUAL.

24

..... n:J

.;..J

C o N ·rl

L o

£:.

12

10

8

6

4

2

o

Cll -2 > ·rl

.;..J -4-u UJ

B. de Graaf et al

~:; . . . ,

-§ -Of' j IJJ -8

-10

~-~~~5---~~-0---~i~5~-~10~--~'5~~6--~~~~1~~---1~~~~2~IO~~215~ body pos it ion (deg)

(a)

12

10

C, 8 Cll

~ 61- -..... ro 41-

.;..J C 0 2f-N ·rl

L 01-0 £:.

Cll -2f-> .....

.;..J -41-u Cll .,.., -6 ..0 ::J U)

-8

-10

J I I I I I I

-25 -20 -15 -10 -5 0 5 10 15 20 25 body pas it ion (deg)

(b)

Figure 9. Reinterpretation of the data presented in Figure 7. (a) SH data obtained from the subgroup of sub­jects (5) who always showed an E phenomenon, and (b) SH data obtained from the subgroup of subjects (3) who explicitly showed an A phenomenon. It is clear that the SH of individuals belonging to those subgroups deviates considerably with larger body tilts, but in a very specific way. The lines drawn are the best fit 3rd degree polynomial.

Results

OCR

All 4 subjects showed an OCR in condition WHOLEBODY. The mean OCR at 25° tilt

was up to about 4° (see Figure 10). This OCR (gain 0.24) is not different from the OCR found in the healthy subjects in the same con­dition (situation a: gain 0.27). In the patients, however, this OCR could be invoked only by somatosensory stimulation.

OCR and Subjective Horizontal in Normal and LD

Tilt of the head alone (condition HEAD) also induced a consistent OCR (gain 0.33) (Figure 10) which, in the patients, should be due to stimulation of the neck. The mean am­plitude of OCR was not significantly different in conditions WHOLEBODY and HEAD.

Tilt of the trunk only (condition TRUNK) induced an OCR with a small but significant gain of 0.17 (Figure 10). This was in contrast to what was observed in healthy subjects, where no OCR was found at all. It is assumed that this OCR reflects the net effect of the so­matosensory and counteracting cervical OCR: The OCR found in condition TRUNK should equal the OCR found in condition HEAD minus the OCR found in condition WHOLE­BODY (A clockwise tilt of the trunk causes a counterclockwise stimulation of the neck with, as result, a clockwise OCR. But a clockwise tilt of the trunk also causes clockwise soma­tosensory stimulation, resulting in a counter­clockwise OCR).

The visually induced ocular rotation (con­dition VISUAL, see Figure l1a) showed a gain of 0.30, which is significantly larger (ANOVA: F= 5.4; df = 1,14; P< 0.05) than in normal subjects (gain 0.11).

25

Subjective Horizontal

In contrast to what was observed in the healthy subjects, the data of the subjects without vestibular function were very noisy (except perhaps for subject A). This is re­flected in the poor reproducibility of data at 0° tilt angle, where the three conditions WHOLEBODY, HEAD, and TRUNK in principle do not differ (see Figure 12).

The SH estimates during whole body tilt were different for each subject (Figure 12). Subject A set the bar almost completely per­pendicular to his body, so apparently almost no tilt was perceived: He underestimated the tilt (Aubert (A) phenomenon). Similar behav­iour was seen in subject D, although to a lesser extent. However, subject B overesti­mated the amount of tilt (Muller (E) phenom­enon): she mentioned during the experiment that with the larger tilt angles to the right she was afraid of falling down. Subject C was highly variable in her estimates: for example, at 20° tilt to the right, the trunk served as ref­erence frame (SH perpendicular to the trunk), but at 20° to the left, the amount of tilt was 100070 overestimated.

10~~--~--~~--~--~~---,--,---,--.--,

8

6

4

0; 2 Q.)

::£ 0 U o -2

.. LL

-8

......... ,

,

whole boch'

head

trunk

" '" -!

~ -6 f

-10L .. __ ~ __ L--~ __ ~ __ ~ __ ~~ __ ~ __ ~ __ ~~ __ ~ - I I I I I I I I I I I

j -25 -20 -15 -10 -5 0 5 10 15 20 25

tilt (deg)

Figure 10. Mean OCR of 4 subjects devoid of labyrinthine function. Shown are the OCR found due to whole body tilt ( ..... ), to head tilt (----), and to trunk tilt (-. -' -). Clockwise rotation is shown as positive values, counterclockwise as negative.

26 B, de Graaf et al

10

8

6 C,

4 Q.l

~ c 2 0 ';:;

0 co

2 -2

co :J -4 <..l 0

-6

8~ -10

I I I I

j -10 -5 0 5 10

room tilt (deg)

(a)

10

8

C, 6 Q.l

~ 4 ro

C 2 0 N

'§ 0 £. Q.l -2 ,~ <..l -4 Q.l

'is -6 :J

en

-8

-10

-10 -5 0 5 room tilt (deg)

(b)

Figure 11. Ocular rotation (a) and subjective horizontal (b) as a function of room tilt in the light (VISUAL). A clockwise deviation of the SH from the true horizontal (respectively clockwise ocular rotation in the orbits) is shown as positive, a counterclockwise deviation (respectively counterclockwise ocular rotation) as negative. The lines drawn are the best fit 3rd degree polynomial.

Tilt of the head alone was estimated rather well by subjects A, C, and D, but was over­estimated by subject B.

A strong effect was found during trunk tilt: Apparently in conditions HEAD and TRUNK the subjects took their trunk as ref­erence and perceived their head as tilted. This

resulted in a deviation of the SH in the same direction as the trunk tilt in condition TRUNK (the "ideotropic tendency," [26]). As for the SH in the illuminated tilted room: Subjects experienced the tilted room always as vertical and adjusted their horizontal according to this frame of reference (Figure 11 b).

OCR and Subjective Horizontal in Normal and LD

en OJ

-0

I (j)

en OJ ::s

I (j)

en OJ

::s I (j)

en (l)

E I If)

25 20 15 10 5 0

-5 -10 -15 -20

25 20 15 10 5 0

-5 -10 -15 -20

25 20 15 10 5 0

-5

15 10 5 0

-5 -10

-- whole body - - - - - head --- trunk

I

1\ 1\

/ \

\

subject A

subject B

\ \

::j\SUbject C

I \ ,. . .,/

\ 1\ \ /' \ If' \ I , \

... \ I/,\ --....~~_~/I . ,.: _____ -

........ / subject 0

-15~~~-L~L-~-L __ L-~~ __ L-~ -25 -20 -15 -10 -5 0 5 10 15 20 25

tilt (deg)

Figure 12. Individual SH settings of the 4 labyrinthine­defective subjects obtained during body tilt (--), during head tilt (----) and during trunk tilt (. _. -. ). A clockwise deviation of the SH from the true horizon is shown as positive, a counterclockwise deviation as negative.

Discussion

OCR

The data indicate that in patients devoid of vestibular function an OCR that is not signif­icantly smaller than in healthy subjects may

27

still be observed during body tilt (WHOLE­BODY). The amplitude of OCR found in our 4 subjects is larger than the OCR in the lab­yrinthine-defective subjects reported by Gray­biel (22). This may be due to the intersubject variability. Since this OCR cannot be of ves­tibular origin, it must be due to somatosen­sory cues from the contact between body and chair or from a "Bauch" organ, as was men­tioned by Mittelstaedt (27).

Head tilt (HEAD) induces a strong OCR as well, which should be of cervical origin in these patients. This confirms an earlier obser­vation of Vogel and colleagues (24), who found in a labyrinthine-defective subject an OCR of 5° during head tilt. Individual differ­ences are present, but on the average the cer­vically and the somatosensory-induced OCR are equally strong. In principle, this should have been reflected in the data of the situation where only the trunk was tilted (TRUNK): The two counteracting forces should have cancelled each other there. This was, how­ever, not the case. A slight but significant OCR was found in that condition. We think that this remaining OCR was due to a skewed distribution in our small group of subjects: Of the 4 patients, 3 had a cervically caused OCR that was significantly larger than that caused by somatosensory stimulation.

Apparently, for static tilt up to 15°, the ocular counterrotation in subjects without labyrinthine function is of the same order of magnitude as that in healthy controls. The vi­sually induced ocular rotation is even better than in the controls, which, all together, helps to stabilize the image on the retina.

In view of the present data, a reinterpreta­tion of our data for healthy subjects is possi­ble: We first assumed that the contribution of the somatosensory system was negligible (com­pare reference 22) and, since in TRUNK no OCR was observed, we concluded that there was no cervical contribution to the OCR in healthy subjects. This was also reported by Krejcov<i and colleagues (17), who reached the same conclusion because there was no dif­ference between the OCR obtained by tilting the whole body or only the head. However, the data obtained by the second experiment

28

indicate that there is a clear somatosensory contribution to the OCR in patients, and it may be incorrect to assume that this contribu­tion is negligible when the otolithic system is working properly. In fact, the data obtained from normals and labyrinthine-defective sub­jects do not allow us to differentiate between the assumption 1) that somatosensory and cervical information contribute to OCR only in patients and not in normals, and the as·· sumption 2) that the otoliths do not contrib­ute to the OCR at alL neither in patients nor in normals. For horizontal eye movements it was found that normals do show somatosen­sory and cervical nystagmus, but with about half the gain of the labyrinthine-defective pa­tients (10,28); therefore, a similar difference in gain may be conceivable for OCR as well. This means that, in normals during body tilt, at least for the investigated range, the OCR is for 50070 due to somatosensory stimulation and for 50% to the otoliths. (If normals have about half of the gain (50%) of the labyrin­thine-defective patients with somatosensory stimulation, then only the remaining 50% could be due to the otoliths hecause no dif­ference in OCR amplitude was found in WHOLEBODY between normals ["situation a"] and patients.) Since the cervi cally induced OCR roughly matches the counteracting so­matosensory induced OCR, it is understand­able that no OCR was found in TRUNK in normals and only a small one in our patients.

In addition, it is understandable that no difference is observed (17) when the whole body or only the head is tilted: The contribu­tion to the OCR by the somatosensory system in the former case is replaced by the contribu­tion of the neck in the latter case.

It is an inevitable conclusion that OCR ex-

B. de Graaf et al

amination is almost impossible without con­tamination by proprioceptive stimulation. In order to sort out the real contribution of each of the systems in normals and patients, it is desirable to isolate the OCR-provoking sys­tems even more, for example, by measuring OCR during body tilt under water or during parabolic flight.

Subjective Horizontal

The gravitational reference frame proved to be less well determined by subjects devoid of vestibular function. Most of all they rely on the visual information even when slanted over 10° (VISUAL) and despite the fact that the body is in the upright position under these circumstances. Apparently this visual informa­tion is very useful to them. These data con­firm earlier observations in the tilting room (29).

Whole body tilt in the dark leads to Au­bert and Muller phenomena, as in normal subjects. When only the trunk is tilted, how­ever, the trunk is perceived as vertical and the head as tilted, in contrast to normal subjects. So, with stimulation of the neck the subjects without labyrinthine function take their trunk as reference frame, while normal subjects take their head.

Acknowledgments- We are especially obliged to Dr. Peter Roos for constructing the Tilting Chair and to Professor Hermann Schone and Professor Dick van Norren for providing useful suggestions for the preparation of the manuscript. This re­search was supported by a grant from PSYCHON, awarded by the Dutch organization for Scientific Research.

REFERENCES

1. Aubert H. (1861). Eine schein bare bedeutende Dre­hung von Objekten bei neigung des Kopfes nach rechts oder links. Virchows Arch. 20:381-93.

2. Muller GE. (1916). Uber das Aubertsche Phiinome­non. Z Psychol Physiol Sinnesorg. 49: 109-246.

3. Howard IP. (1982). Human visual orientation. Toronto: Wiley.

4. Witkin HA, Asch SE. (1948). Studies in space orien-

tation; 3: Perception of the upright in the absence of a visual field. J Exp Psychol 38:603-14.

5. Schone H. (1962). Uber den Einfluss der Schwerkraft auf die Augenrollung und auf die Wahrnehmung der Lage im Raum. Z vergl Physiol 46:57-87.

6. Miller EF, Fregly AR, Graybiel A. (1968). Visual horizontal- perception in relation to otolith-func­tion. Amer J Psychol 81 :488-96.

OCR and Subjective Horizontal in Normal and LD 29

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

Fischer MH. (1930). Messende Untersuchungen tiber die Gegenrollung der Augen und die Lokalisation der scheinbaren Vertikalen bei seitlicher Neigung des Gesamtkorpers bis zu 360 0

; 2: mitteilung Unter-suchungen an Normalen. Graefes Arch Clin Exp Op-thaI. 123:476-508. Miller EF, Fregly AR, Brink G van den, Graybiel A. (1965). Visual localization of the horizontal as a function of body tilt up to ±90° from gravitational vertical. NSAM-942 NASA Order NO R-47 Naval School of Aviation Medicine, Pensacola, Florida. Schone H, Udo de Haes HU. (1968). Perception of Gravity-Vertical as a function of Head and Trunk Position. Z vergl Physio!. 60:440-4. Bles W, De Jong JMBV, Rasmussens J. (1984). Postural and oculomotor signs of labyrinthine-defec-tive subjects. Acta Otolanyngol (Stockh) Supp!. 406: 101-4. CoIIewijn H, Van der Steen J, Ferman L, Jansen TC. (1985). Human ocular counterroIl: assessment of static and dynamic properties from electromagnetic scleral coil recordings. Exp Brain Res. 59:185-96. Diamond SG, Markham CH. (1981). Binocular counterrolling in humans with unilaterallabyrinthec-tomy and in normal controls. Ann NY Acad Sci. 374:69-79. Diamond SG, Markham CH. (1983). Ocular counter-rolling as an indicator of vestibular otolith function. Neurology. 33:1460-9. Miller EF, Graybiel A. (1974). Human ocular coun-terrolling measured during eight hours of sustained body tilt. Minerva otorinolaringol. 24:247-52. Goodenough DR, Sigman E, Oltman PK, Rosso J, Mertz H. (1979). Eye torsion in response to a tilted visual stimulus. Vision Res. 19: 1177-9. Howard IP, Templeton WB. (1964). Visually-induced eye torsion and tilt adaptation. Vision Res. 4:433-7.

L U o

12

10

8

6

4

2

o

? -2 UJ

-4

o

17. Krejcova JH, Highstein S, Cohen B. (1971). Laby-rinthine and extralabyrinthine effects on ocular coun-terrolling. Acta Otolaryngo!. 72: 165-71.

18. Fischer MH. (1927). Messende Untersuchungen tiber die Gegenrollung der Augen und die Lokalisation der scheinbaren Vertikalen bei seitlicher Neigung (des Kopfes, des Stammes und des Gesiimtkorpers). Graefes Arch Ophtha!. 118:663-80.

19. Wade NJ. (1968). Visual orientation during and af-ter lateral head, body and trunk tilt. Percept Psy-chophysics. 3:215.

20. Biemond A, De Jong JMBV. (1969). On cervical nys-tagmus and related disorders. Brain. 92:437-59.

21. Schone H, Udo de Haes HU. (1971). Space orienta-tion in humans with special reference to the interac-tion of vestibular, somaesthetics and visual inputs. Biokybernetik. 3: 172-91.

22. Graybiel A. (1974). Measurement of otolith function in man. In: Kornhuber H, ed. Handbook of sensory physiology, part V1/2. New York: Springer.

23. Winer BJ. (1962). Statistical principles in experimen-tal design. New York: McGraw-Hill.

24. Vogel H, Thumler R, von Baumgarten RJ. (1982). Ocular counterrolling: some practical considerations of a new evaluation method for diagnostic purposes. Acta Otolaryngo!. 102:457-62.

25. Bles W, de Graaf B. (1991). Ocular rotation and per-ception of the horizontal under static tilt conditions in patients without labirinthine function. Acta Oto-laryngol (Stockh). 1991; 111:456-62.

26. Mittelstaedt H. (1988). The information processing structure of the subjective vertical: a cybernetic bridge between its psychofysics and its neurobiology. In: Marko H, Hauske G, Struppler A, eds. Process-ing structures for perception and action. VCH Ver-lagsgesellschaft, Weinheim. 217-63.

27. Mittelstaedt H. (1983). A new solution to the prob-

-::f ~ j L--~~~5----2~!0~--~l~5----l~o~--~!5~~~~--~~--~1!~O--~1~!5~-2~!O~~2~5~

body position (deg)

Figure 13. Mean OCR data (filled circles) and mean SH data (open circles) from 5 subjects who always exhibited an E phenomenon, presented as a function of body position in the dark. The correlation between OCR and SH is for each of the individuals 80% or higher, and the mean OCR and SH data do not statistically deviate from each other.

30

lem of the subjective vertical. Naturwissenschaften. 70:272-81.

28. Bles W, De long 1MBV, De Wit G. (1984). Somato­sensory compensation for loss of labyrinthine func­tion. Acta Otolaryngol (Stockh). 97:213-21.

29. Bles W, De long 1MBV, De Wit G. (1983). Compen­sation for labyrinthine defects examined by use of a tilting room. Acta Otolaryngol (Stockh). 95:576-9.

30. Schone H, Udo de Haes HU. (1971). Space orienta­tion in humans with special reference to the interac­tion of vestibular, somaesthetic and visual inputs. Biokybernetik. 3: 172-91.

Appendix

Our final question was about a possible re­lationship between OCR and SH. As stated before, if the perceptual system fails to take ocular counterrotation into account, the re­sulting error (in the dark!) would be in the same direction as the E phenomenon. This overcompensation should then match the OCR exactly. When the subjects are taken together, a comparison of identical OCR and SH con­ditions in the dark (Figures 3 and 7) does not reveal such a relationship, as stated before by Miller and colleagues (6). However, when we compare the SH and OCR data of the sub­group who always exhibit an E phenomenon, the strength of the illusion appears to fit nicely with the strength of OCR (see Figure 13).

We therefore suggest that with tilts up to 30° the OCR could be responsible for the E phenomenon. With larger tilts the otoliths be­come less effective (will underestimate the tilt; Schone and Uda de Haes [30]), which at first will be masked coincidentally by OCR, and with still larger tilts (and a saturated OCR) will cause the A phenomenon in all subjects (see Figure 14). A consequence of this sugges­tion is that subjects who already show the A

B. de Graaf et al

-20

c '2 --------------------+OCR a > 0 (l)

u '0 ~ 10 en c 0 SH

20

30 L.5 60 75 90 body tilt (deg)

Figure 14. A model for the course of the su bjective horizontal (SH) as a function of lateral body tilt in the dark. For illustrative reasons the course of OCR is plotted as well. OH = objective horizontal; OCR = OCR as a function of body tilt; SH* = hypothetical SH, based on mere vestibular information (with a gain less than 1 with larger tilts) and without appearance of OCR. SH = subjective horizontal as found in the liter­ature (and in our data). A perceptive system that bases SH on mere vestibular information without feed­back about OCR could be responsible for the MUlier (E) phenomenon (an overestimation of body tilt caused by addition of OCR and SH*). With larger tilts, and a corresponding larger utricular inaccuracy, the addi­tion of SH* and OCR could not (over)compensate for this inaccuracy. As a consequence an Aubert (A) phe­nomenon will appear.

phenomenon with small tilts are considered to have a less accurate utricular functioning (a lower gain). There is no need for them to have sleepless nights about that, because OUf

suggestion is only valid for static tilt situa­tions in the dark. Anyway, latent A and E phenomena are simply overruled by a visual frame of reference.