General introduction

Certain findings point to an explanation for space sickness that entails the central processing of vestibular information. First, there is a clear cause-and-effect relationship between head movements and the appearance of symptoms, which is stronger with the eyes open than with the eyes closed (Oman et al. 1986). Second, astronauts have noted that visual reorientation illusions are potentially provocative, even though the subjective reorientation is not accompanied by actual motion. Thus, problems seem to arise from inconsistencies in the spatial orientation information, probably due to unusual stimulation of the otolith organs of the vestibular system. Under circumstances of normal gravity (1G), the otolith organs not only detect linear accelerations resulting from translational motion of the head, but also continuously register the direction of gravity relative to the head (see Figure 1). Without the constant stimulation from gravity, the otoliths still respond to translational aspects of head motion, but they no longer provide a reference to vertical. Obviously, this will affect the sensory interactions during head movements. It is assumed that the otolith function must adapt to the novel state of gravity (0G) as to reestablish the congruity between the orientation senses. In the same way it is assumed that, once adapted to 0G, the otolith system must re-adapt to 1G on return on Earth. The latter would explain the reappearance of symptoms after spaceflight (similar to the "mal d’embarquement" which is observed on the shore after adaptation to sea sickness). Money (1992) has suggested three possible paradigms by which otolith adaptation to weightlessness can take place: 1) suppressing the otolith system (gain reduction); 2) the interpretation that all otolith signals arise from head translations (tilt-translation reinterpretation); 3) the substitution of other senses for the otolith information (higher weighting of visual inputs). In general, studies in spaceflight seem to support the view that all three mechanisms may contribute to adaptation to 0G, although the evidence is sporadic and sometimes contradictory. The difficulty with research in space is that it is hampered by high expenses and small numbers of subjects.

Recently it was suggested that otolith adaptation may be studied after prolonged exposure to hypergravity in a human centrifuge (Bles et al. 1989; Albery and Martin 1994). After a centrifuge run with a G-load of 3G, postural instability and symptoms of motion sickness were found (Bles and De Graaf 1993), similar to the effects observed in astronauts on the first day post-flight (Kenyon and Young 1986; Bles and Van Raay 1988). The centrifuge run itself was not experienced as very stressful by the subjects, but symptoms emerged afterwards. Interestingly, European astronauts who participated in these experiments reported striking similarities with sensations in space (Ockels et al. 1990). Head movements induced illusory motion of the visual surround (oscillopsia) and, analogous to space sickness, triggered attacks of discomfort or nausea (Sickness Induced by Centrifugation, SIC). Especially head movements that changed the orientation of the head to vertical were provocative ("pitch" and "roll", see Figure 2 for convention on rotation axes). The subjects walked very carefully in an attempt to minimize head movements (resulting in a "robot walk"). In addition, it was found that the rank order in the astronauts’ susceptibility for SIC was the same as that for SAS, implying a common mechanism. Medical monitoring during and after the centrifuge run has shown that a cardiovascular cause was highly unlikely (Bles et al. 1989; Krol 1994), which is in favor of a vestibular mechanism.

Figure 2. The three main rotation axes with respect to the head. Rotation about the x-, y-, and z-axis are designated as "roll", "pitch", and "yaw", respectively.

Otolith adaptation to hypergravity

The first aim of the work presented in this thesis was to identify vestibular adaptation in human subjects after an one-hour centrifuge run with a G-load of 3G. Of special interest was adaptation of the otolith system. Otolith signals play a role in sensing the inclination of the head relative to vertical, and in controlling postural responses and eye movements. As mentioned above, previous studies have shown that a long duration centrifuge run produces destabilizing effects on postural balance. In addition, subjects noted a compelling sensation of illusory body tilt immediately after the run (Bles and De Graaf 1993), and also showed a measurable bias in the judgment of body orientation in a post-test about 30 minutes later (Bles et al. 1989). These findings provide indirect evidence for otolith adaptation.

A new study was designed to expand the body of evidence with more "objective" tests. In this study, the effect of centrifugation on the ocular torsion response to static tilt was evaluated. When the head or the whole body is tilted laterally (roll), the eyes rotate in the opposite direction about the line of sight. This is referred to as ocular counterroll or, more generally, ocular torsion (OT). In general, an important vestibular function is to generate compensatory eye movements (vestibulo-ocular reflex, VOR) during movements of the head, so as to minimize retinal blur. The OT response observed during static tilt is considered a static equivalent of this, which probably is used for orientation. It has been shown that static OT reflects the functioning of the otolith organs (Miller 1962; Colenbrander 1963; Miller and Graybiel 1971; Diamond and Markham 1981; Collewijn et al. 1985), and the response also has clinical relevance (Diamond and Markham 1983; Gresty and Bronstein 1992). It was hypothesized that a hypergravity-induced change in the sensitivity of the otolith system would be visible in the magnitude of static OT. In addition to the static measurements, the dynamic OT response was also measured during sinusoidally body roll, in order to evaluate changes in the interaction between the semicircular canals and the otolith organs. The results of this study are described in Chapter 2 of this thesis.

The magnitude of static OT in humans is rather small: on average, the maximum response amounts to about 6º. OT can not be detected by standard electro-oculography (EOG), because rotations about the visual axis do not produce changes in the electrical potential. Accurate measurements of OT can be achieved with the scleral search coils technique. However, this technique was discarded because it requires the subjects to wear contact lenses containing a small coil, that firmly adhere to the sclera and soon become uncomfortable. The OT measurements took about 45 min and were performed twice (once before and once after the centrifuge run), which was considered too long for the use of search coils. Furthermore, the metal frame of the rotating chair was considered unsuitable to contain the large Helmholtz coils, needed to generate a homogeneous magnetic field around the subject.

An alternative, non-invasive approach to measure OT was offered by the video-based technique described by Bos and De Graaf (1994). This comprises the recording of eye movements on video tape and quantification of OT afterwards by matching the iris patterns between one image of the eye and another image which serves as a reference. In the original method, the matching is performed semi-automatically by visual inspection of iris patterns that are digitized and displayed on a computer monitor. Because this procedure is too time consuming for time series analysis, an automatic pattern recognition algorithm was developed and implemented on a PC system, that automatically digitizes consecutive images from video tape into the computer memory, selects significant structures in the iris of a reference image, and relocates these landmarks by a method of template matching. The use of several landmarks, uniformly distributed over the iris, allows for correction of errors that result from miscalculation of the rotation center of the eye. This feature is usually not accounted for in other video-oculography methods. The new video-oculography method is described in Chapter 1 of this thesis.

To a great extent, the information concerning spatial orientation provided by the different sensory systems is redundant. On the other hand, sensory systems also differ in their characteristics so that they can supplement each other. For instance, the visual system operates in a low-frequency range, whereas the vestibular system operates at higher frequencies. This can be verified by a simple test: when you keep your head still and wave your hand in front of your face, the image of your hand will soon become blurred with increasing frequency; when you keep your hand still and oscillate your head, the image of your hand will remain clear up to much higher frequencies due to the VOR. This frequency dependency is reflected in the contribution of the visual and vestibular senses to orienting responses (as we will see in Chapter 5).

A similar interaction may exist within the vestibular system, between the semicircular canals and the otolith organs. The semicircular canals respond to angular accelerations of the head and are responsible for the generation of the angular VOR. The operational range of the semicircular canals is adjusted to frequencies of natural head movements. At very low frequencies, the canal-induced response becomes less effective (Carpenter 1988), and it was hypothesized that the otolith system may help to improve the low-frequency response dynamics for rotations about an off-vertical axis. The torsional VOR was considered appropriate for this purpose, especially because of the association of the otolith organs with OT. From studies on a linear track which applied linear accelerations along the interaural axis, it is known that otolith-induced OT can be characterized as a low-frequency response, being maximal in a frequency range up to about 0.3Hz (Young 1985; Hannen et al. 1966; Lichtenberg et al. 1980). Nevertheless, whereas otolith-induced OT has been well-studied during static tilt, its usefulness during dynamic tilt has received little attention. This is due to the simultaneous activation of the semicircular canals, which complicates examination of the otolith component. In Chapter 3 a study is presented in which the OT response was measured during sinusoidal body roll in a frequency range of 0.05-0.4Hz. By comparing the response to rotation about an earth-horizontal axis with that to rotation about an earth-vertical axis, some specific otolith contributions to the response were revealed.

The coupling between semicircular canals and the otolith organs is of interest for another reason as well. There is a fundamental difference between the two vestibular subsystems. Functionally, the semicircular canals can detect the rate of rotation of the head, but they can not detect the plane of rotation relative to gravity. Signals from the semicircular canals are coded in head egocentric coordinates. The otolith system, however, is able to detect the plane of rotation relative to gravity. Signals from the otoliths therefore may help to link the egocentric canal-induced VOR to an exocentric frame of reference. This issue will also be evaluated in Chapter 3.

Most of us are familiar with the experience that, sitting in a stationary train, watching a neighboring train leave may produce a sensation as if our train is moving. This example shows that our visual system plays an important, and often dominant, role in the perception of self-motion. In general, motion of the visual surroundings induces a sensation of self-motion ("vection") in the opposite direction. This is particularly true for yaw rotation about the vertical axis. An upright observer viewing rotation of a visual scene about an earth-horizontal axis, however, experiences a limited degree of self-tilt in the opposite direction, paradoxically combined with a sensation of continuous self-rotation. It has been shown that the magnitude of visually induced self-tilt is to some extent a function of stimulus area and also of stimulus velocity (Held et al. 1975; Howard et al. 1988). Nevertheless, the tilt effects seldomly exceed 20º, which is ascribed to the restraining influence of the otolith organs which do not register any change in the direction of gravity relative to the head. This hypothesis is supported by the observation that visually induced self-tilt increases when the head is inclined 90º or inverted, so that the urticles are out of their most sensitive position (Young et al. 1975; Bishof 1978; Howard et al. 1988).

Recently, Howard and Childerson (1994) distinguished that the sense of body orientation depends on three types of visual information: visual motion, visual frame, and visual polarity. The visual frame refers to a set of lines and surfaces (such as walls and floors) which are normally horizontal or vertical. Even a simple stationary tilted visual frame may lead to perceived self-tilt in the opposite direction (Asch and Witkin 1948c). Visual polarity is found in objects with a distinct top and bottom, indicating "up" and "down", such as chairs, tables, and people. When all polarized objects visible are tilted relative to gravity, the effect is a compelling illusion of self-tilt, especially if the objects are part of a tilted frame (Asch and Witkin 1948a). The usual stimulus in studies on visually induced self-tilt is a simple display of dots lacking visual frame or polarity cues, so that the effects are due to mere visual motion information. Using a rotating 7-foot cubic room with replaceable walls, Howard and Childerson (1994) showed that motion of a polarized visual frame elicited larger effects than visual frame motion, which in turn elicited larger effects than visual motion alone. They noted that erect subjects even experienced sensations of head-over-heels rotation when seated inside the "furnished" room (the polarized visual frame), while it rotated at constant velocity about the subjects’ roll-axis. Apparently, when certain requirements are met, a visual stimulus can completely dominate the conflicting otolith information in a stationary observer.

The so-called "Tumbling Room" of Dr. Howard’s laboratory (York University, Toronto, Canada) was later replaced by a larger and more realistic one, containing an even richer variety of polarized objects. Chapter 4 of this thesis describes a study that was performed in this new Tumbling Room to determine the percentage of subjects that perceive "full tumbling" (through 360º) about an earth-horizontal axis. Another objective of the study was to examine the effects of body orientation with respect to gravity (upright or inclined 90º) and with respect to the fixed rotation axis of the room (roll, pitch, and yaw). These variables have been shown to affect illusory self-tilt and vection, using displays that consisted only of dot patterns. It was expected that these factors become less important when the effectiveness of the visual stimulus increases.

Flight simulators, among other virtual reality applications, exploit visually induced self-motion and self-tilt. As long as the simulated motion is smooth and in horizontal direction, a visual display can be sufficient. However, when the simulated motion comprises fluctuations in velocity or direction (that is, linear and angular accelerations, respectively), the lack of confirming vestibular information will undermine the fidelity of the intended sensations. Flight simulators with a motion base therefore employ vestibular "onset" cues to reinforce the visual motion. For instance, the vestibular cue arising from forward linear acceleration in an aircraft can be recreated by pitching the motion base backwards, stimulating the utricles in a way that is approximately comparable to horizontal linear acceleration. Vestibular onset cues may also be useful to enhance visually induced self-tilt. As was mentioned above, illusory self-tilt is larger when the body is inclined 90º. However, a motion base has limited travel and must return to its original position. Despite the widespread use of motion base simulators, there seem to be no documented studies on the effects of a vestibular onset cue on illusory self-tilt. The experiment presented in Chapter 5 addresses the question whether sensations of self-tilt and self-rotation can be enhanced by actual body tilt at the onset of rotation of the visual scene. The original plan was to use the Tumbling Room (of Chapter 4) for this experiment but this room generated complete illusory self-rotation through 360º in most subjects even without a vestibular onset cue. Therefore, a less powerful stimulus was needed which only produces limited illusory self-tilt, so that any effects of an onset cue could be assessed. Such a stimulus was the rotating sphere of Dr. Howard’s laboratory, which had been shown before to generate a mean illusory self-tilt of about 20º (Howard et al. 1988).