Summary and Conclusions

Part I

Measuring perceived horizontal self-motion indirectly, using perceived object-motion

In Part I of this thesis a relatively new psychophysical method for measuring perceived self-motion (Wertheim, 1994) is presented and tested. The assumption on which this method is based, is that self-motion affects the perception of earth relative object motion. If this is so, one may be able to measure percepts of self-motion indirectly, by measuring percepts of earth relative object motion. Such a method would have the advantage of not directing the subject's attention towards the perceived self-motion, which would possibly influence the measured percept. In tile following 4 chapters the contribution of the vestibular system to perceived selfmotion was under investigation using the method at issue. Thus it was tried to exclude as much as possible other sources of formation to the self-motion perceptual system, such as visual, auditive or proprioceptive information.

Chapter I

The perception of ego-velocity (PEV) during sinusoidal linear acceleration (otolith stimulation combined with pressure cues) is investigated in chapter 1, with tile mentioned psychophysical method proposed by Wertheim (1990, 1994). Tile method is based oil thresholds for object motion perception measured during passive self-motion: subjects are oscillated on a linear horizontal motion device (called 'tile sled') and, whilst they are in motion, are required to judge whether a shortly presented (200 - 400 ms) grating is moving or not. Thresholds for perceiving motion of the grating in forward and backward direction are then assessed and tile midpoint between them is taken as the velocity at which the subject perceives the gating as stationary relative to the earth. This velocity is called tile Point Of Subjective Stationarity (PSS). Now, when one moves past a stationary object, it call be stated that the retinal image velocity of this object (corrected for its distance and visual angle) equals ones ego-velocity. In the following experiments this reasoning is turned around, to state that the retinal image velocity of the grating (corrected for its distance and visual angle) at the PSS may then be taken as a measure for PEV.

In a first experiment PEV is measured both with and without illumination of the surrounds. It is shown that, for the particular frequency (0. 15 Hz) and amplitude (109.5 cm/s) of self-motion used, PEV in the light approaches the actual velocity of self-motion. In the darkness PEV is found to be significantly smaller than the actual velocity of self-motion. The results show that without visual flow information the self-motion perceptual system lacks sufficient information to form a veridical estimation of ego-velocity. These results are compared to the so-called Filehne illusion (Filehne, 1922; Mack and Herman, 1978) which happens during smooth pursuit eyemovements. In a second experiment the PEV is measured again in the same way, but only without illumination of the experimental room. It is measured at different points of the presented self-motion velocity sine, in order to find out whether a PEV sine wave may be fitted to the results. The results show that in response to a sinusoidal self-motion stimulus, PEV is indeed sinusoidal as well, with a gain of 0.8 and a phase-lead of 4 deg. at 0. 15 Hz. It is concluded that this gain and phase relationship may be an indication of the otolith response to linear horizontal acceleration (see e.g. the model of Grant and Best, 1987).

Chapter 2

In chapter 2, using the same psychophysical method as in chapter 1, object motion perception was assessed in avestibular patients and normal controls. Two experiments were conducted, in which subjects were required to assess the motion of a visual stimulus with respect to earth. In the first experiment, we measured the velocity at which a briefly presented (200 ins) grating was perceived as earth fixed, while the subject maintained fixation on a visual target fixed relative to the body, during whole-body yaw rotation (VOR suppression). In this experimental setup, tile influence of the semi-circular canal signals oil object motion perception is evaluated. The avestibular patients judge the grating to be stationary with respect to earth, when it is actually moving at the saine velocity as their body, whereas normal controls perceive the grating stationary relative to earth when It actually moves at a velocity slower than their body, but greater than zero. The difference between the two subject groups is significant, and shows a strong contribution of tile horizontal canals to object motion perception. Similarly, a measurement of the velocity at which a gating was perceived as stationary relative to earth was obtained during smooth pursuit eyemovements (in this case the actual Filehne illusion as it is known from literature (Filehne, 1922; Mack and Herman, 1978) is measured). The contribution of the efference copy of the oculomotor signal to object motion perception is thus assessed. As with the first experiment, the normal controls display a more veridical sense of object motion perception than tile patients, although tile difference is only just significant. We suggest that the difference could be all adaptive change in the patients' perception of motion, which allows them to reduce tile effects of oscillopsia (see Bronstein, 1987).

Chapter 3

In chapter 3 a visual illusion is reported that comprises the following: When a monitor with a moving constant velocity grating is swayed in front of a subject, the grating may be perceived as freezing or decelerating oil the screen. This percept also appears in tile peripheral field and seerns very similar to what happens when one is passing a earth stationary monitor with a moving grating oil it (compare this latter example to the experiments in chapter 1). Tile illusion seems to depend oil tile magnitude and tile direction of the retinal grating velocity relative to that of tile retinal image velocity of the monitor. Various possible explanations are rejected. It is concluded that the illusion shows a resemblance to the phenomenon of 'motion capture' (e.g. Ramachandran and Cavanagh, 1987) but that it has certain new characteristics that need all explanation.

Chapter 4

Finally, in chapter 4, three experiments are presented that test different aspects of the validity of tile psychophysical method that has been used in chapters I and 2. In the first experiment a possible influence of visual grating size oil tile measured gain of the efference copy during smooth pursuit is investigated. Tile results indicate a significant effect of grating size on measured efference copy gain: tile greater the stimulus, tile greater the gain. The second experiment concerns the consequence of the existence of the illusion presented in chapter 3, for the results presented in chapter 1. Were the presented modulations oil the thresholds for perceived grating velocity in chapter I caused by vestibular stimulation, as was assumed? Or were they instances of tile illusion presented in chapter 3, which seems to be caused by retinal image velocity interactions between grating and monitor runs? The results of the second experiment in this chapter, strengthen the presumption that tile illusion of chapter 3 is at least partly responsible for the results of chapter 1.

To prevent interactions of the presented grating with the rims of the monitor, and thus the happening of the illusion presented in chapter 3, for the third experiment, a new visual stimulus is designed to measure PEV. It consists of one single moving stripe, which during its presentation, never appears from behind or disappears behind the rims of the monitor. With this stimulus in complete darkness, the monitor rims are absolutely invisible during the experiment To our amazement PEV during linear horizontal sinusoidal acceleration is found to be zero using this one stripe stimulus! Different explanations for this result are proposed. It may be, that the linear inertial sensors (otoliths, proprioception, etc.) do not influence the perception of earth relative objectmotion. Also it is possible that the size of this one stripe stimulus (see experiment 1 of this chapter) and, related to that, its estimated distance (see also Wertheim, 1994) have caused this result.

The question underlying all the experiments presented in Part 1 was: can perceived horizontal self-motion be measured indirectly, by attending to, and reporting, perceived object motion?

To answer this question in the affirmative the following conditions had to be fulfilled:

  1. Horizontal self-motion should modulate perceived earth relative object motion
  2. Features of the presented moving object should riot modulate perceived object motion.

The first condition was shown to be true with certainty only for angular motion, in chapter 2, but not for linear motion (see chapters 1, 3 and 4).

As for the truth of the second condition, severe doubt was raised in chapters 3 and 4. One of the measures that was taken in the presented experiments to prevent features of the object from influencing the results, was a very short presentation time. Apparently, even with presentation times as short as 200 ms, features of the presented object do influence the results.

In conclusion it has become doubtful that the proposed psychophysical method is valid to measure contributions of different sensory systems to the perception of selfmotion.

Part 11

Sensory interactions during the perception of horizontal linear self-motion

In Part II, because of the doubts that were raised in Part 1, especially concerning the measurement of linear horizontal self-motion a different approach to investigating the linear inertial sensors (otoliths, proprioceptive feedback etc) was tried. Instead of trying to eliminate all information other than from the inertial sensors, the interaction of the inertial sensors with two of such other sources of information was investigated. III chapter 5, mental set, and in chapter 6, vision, is intentionally interacting with the linear inertial sensors.

Chapter 5

In chapter 5 the influence of mental set on the perception of tilt during linear horizontal accelerations is investigated. III the first experiment 14 subjects are accelerated sinusoidally at 4 different acceleration amplitudes (0.04 g, 0.08 g, 0.1 g, 0.16 g), whilst they have no idea at all about the motion device (tile linear horizontal acceleration sled) that is moving them. Tile subjects have never been to our lab before, they have not seen any photographs of this and other motion devices present at our institute, they are blindfolded before being taken to tile linear acceleration sled, and they are blindfolded during tile experiment. In addition noise of the sled's motion and airflow cites are masked. It is shown that tinder such circumstances (uncertainty concerning the experienced motion) during half of the runs subjects experience tilt illusions in addition to horizontal motion, almost immediately (after I - 5 sinusoids). Similar data, from a setup in which Subjects did see and know the sled with its horizontal rails before the experiment started, showed that tilt illusions do not occur until after 53 sinusoids or more.

In a second experiment the sled's chair is adjusted so that It can tilt sinusoidaly in the XZ plane, during linear horizontal sinusoidal motion along the sled's rail (in the X direction). The subjects do see the device before they are blindfolded in this experiment, but because it is impossible for the subjects to sense whether the chair is tilting, or the sled is moving horizontally, or both, the situation in terms of mental set is again uncertain. Thresholds for chair tilt motion in the same direction as its horizontal motion, and in the direction opposite to its horizontal motion are assessed. By calculating the midpoint between these two thresholds it is possible to assess tile amplitude of tilt motion opposite to horizontal sled motion, necessary to create a percept of horizontal motion (in other words the size of the tilt illusion during horizontal motion is measured). This amplitude, on average, turns out to coincide with one fourth of the gravito inertal force vector amplitude of tile horizontal motion and does not depend on frequency or amplitude of the motion profile. This is in accordance with a model on the perceived vertical by Bos and Bles (1997).

In addition it is shown that tilt thresholds during horizontal sled motion are significantly higher than tilt thresholds without horizontal motion. It is suggested that this is due to more noise in the system during horizontal motion, as a result of the fact that for the linear inertial sensors (otoliths, proprioception etc) tile situation is ambiguous without the help of prior knowledge and vision. When the latter two lack, the final motion percept may depend oil the tilt thresholds and oil past experience of the perceptual system memory of similar sensations).

Chapter 6

In this last chapter interactions between proprioceptive meaning otolith, proprioceptive etc) self-motion stimuli and visual self-motion stimuli are investigated. The proprioceptive stimuli consist of linear horizontal sinusoidal acceleration at different amplitudes, induced by the same sled as was used in the previous chapter. The Visual stimuli consist of computer generated linear horizontal sinusoidal accelerations of a virtual environment presented on a monitor, that was mounted onto the sled's chair, and comprised the entire visual field of the subject. The two motion stimuli may concur with one another, but they call also be made to differ from each other, either in amplitude or in phase. Differences in amplitude or phase are introduced systematically in the following experiments, and subjects are asked to report whether they experience a 'normal' (veridical) linear horizontal selfmotion sensation, and if not, whether they perceive the surroundings as moving earth relatively. Three experiments are performed. In the first experiment the virtual environment consists of a room, and series of amplitude and phase discrepancies at two different frequencies (0.18 Hz and 0.36 Hz) are investigated. The second experiment is similar but the virtual room is replaced with a virtual endless corridor. Finally in the third experiment only amplitude discrepancies are investigated, but for a whole range of self-motion profiles with the saine frequency (0.16 Hz) and differing in maximum acceleration (0.0 1 g, 0.02 g, 0.04 g, 0.08 g and 0. 16 g). In addition the different visual self-motion stimuli are tested on their ability to produce a veridical self-motion sensation (vection) without the support of a proprioceptive stimulus.

Especially in the conditions in which phase discrepancies between proprioceptive and visual motion stimuli are induced (in experiments 1 and 2), a lot of motion sickness occurs. Almost all of the subjects tested, suffer some symptoms of motion sickness. It is proposed that phase differences may be more provocative because of a direction conflict between the proprioceptively sensed change ill tile GIF vector and the expected (on the basis of the visual stimulus) change in the GIF vector. In addition the results imply that a small phase lead of the visual stimulus improves the quality (veridicality) of the perceived horizontal self-motion (see also Mitchell, 1991). The same seems to be true for all endless corridor combined with an instruction to tile subjects to 'forget' the constraints of the motion device (i.e. its length), as opposed to a restricted room without this instruction.

With respect to the data resulting from discrepancies in amplitude between proprioceptive and visual stimuli, it is found that low amplitude proprioceptive stimuli result in the most veridical percepts of self-motion, when they are combined with visual stimuli of a higher amplitude (1.5 to 8 times higher). Tile ratio of visual/proprioceptive stimuli that is judged as most veridical, as well as tile uncertainty of this judgement, decrease with all increase ill tile proprioceptive amplitude. This brings with it tile fact that proprioceptive stimuli that are bigger than the visual stimulus are much more often judged as 'not normal' than when it is the other way around. In addition it is found that, even though small amplitude discrepancies already cause subjects to label their self-motion as 'not normal', it takes much bigger amplitude discrepancies before Subjects experience tile visual surround as not stable relative to tile earth. The latter is explained by a rather big JND (Just Noticable Difference) between perceived ego-velocity and retinal slip velocity, before instability of the Visual world occurs (Wertheim and Bles, 1984).

Finally, without proprioceptive stimuli, visual selfmotion stimuli with acceleration amplitudes Lip to 0.08 g may induce a veridical sense of self-motion (vection). With higher acceleration amplitudes, the quality of the self-motion percept gains from adding proprioceptive stimuli.


Bos J. E. and Bles W. (1994) A vestibular based i notion sickness incidence model. TNO report TM-97-A032, TNO Human Factors Research Institute, Soesterberg, 'nie Netherlands.
Bronstein A. M. (1987) Oscillopsia of peripheral vestibular onigin. Acta Otolaryngol. (Stockh) 104, 307-314.
Filehne W. (1922) Ueber das optische Walirnelimen von Bewegungen. Zeitschrift flir Sinnesphysiologie 53, 134-145.
Grant W. and Best W. (1987) Otolith-organ mechanics: Lumped parameter model and dynamic response. Aviat. Space and Environin. Med. 58, 970-976.
Mack A. and Hennan E. (1978) The loss of position constancy during pursuit eye movements. Vision Research 18, 55-62.
Mitchell. D. G, Hoh R. H, Atencio Jr. A. and Key D. L. (199 1) The use of ground based simulation for handling qualltied research: a new assessment. AGARD conference proceedings. Agard-CP-513, 23/1-23/14.
Mittelstaedt M. and Glasatier S. (1991) Idlothetic navigation In gerbils and humans. Zool. Jb. Physiol. 95, 427-435.
Ramachandran V. S. and Cavailagh P. (1987) Motion capture anisotropy. Vision Research, vol. 27, no. 1, 97-106.
Wertheirn A.H. (1990) Visual, vestibular and oculoniotor interactions in the perception of object-motion during egoniotion. Ill R. Warren and A. H. Wertheim (eds.), Perception and control of self-motion (pp. 171-216) Hillsdale, NJ: Lawrence ErIbaurn.
Wertheini A. H. (1994) Motion perception during self-inotion: The direct versus inferential controversy revisited. Behavioral and Brain Sciences 17, 293-355.
Werthefin A. H. and Bles W. (1984) A re-evaluation of cancellation theory: Visual, vestibular and oculomotor contributi oils to perceived object motion. Report EF 1984-8, Soesterberg, NL: TNO Human Factors Research Institute.

Go to the INDEX page