... interfaces[*]
Note, for instance, that this dissertation was written for a Mechanical Engineering degree. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... interface[*]
The use of the word ``virtual'' in ``virtual interface'' is related to the term ``virtual images'' in optics. A virtual image is one which appears to originate from a location which is not physically occupied by the source of the image. For instance, an object seen in a mirror appears to be behind the mirror, but is not [37]. Similarly, a virtual interface (or, somewhat more generally, a virtual environment) creates a strong impression that something ``really'' exists which in fact lacks a physical instantiation. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... construct[*]
In prior publications, I have referred to the ``rest frame construct'' as the ``rest frame hypothesis''. The change to ``rest frame construct'' was made on the recommendation of my doctoral committee, to emphasize that the intent is more to find a convenient summary of the existing literature on spatial perception than to form a testable hypothesis. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... sensations[*]
Another way to view the Class A/Class B distinction is that in the case of Class A measures the participant makes a response which is directly congruent to the input, whereas with Class B measures a mental transformation is required. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... stimulus[*]
In general, one needs to be careful when comparing an HMD to a non-HMD with a body-sway measure. The HMD may, simply by altering the effective mass of the head, affect body sway (particularly head sway). This would need to be controlled for, for instance by comparing body sway with eyes closed with and without the HMD. It is not clear from the available abstract whether the researchers did so. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... eye[*]
The term ``foreground occlusion'', seems preferable to ``peephole'', because the latter has strong monocular connotations. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... movement[*]
I recommend the faster parts of the videotape Over Washington, possibly played on fast-forward. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... ``inside-out''[*]
There is a refinement called a ``frequency-separated'' display which combines the ``outside-in'' and ``inside-out'' approaches. There are data pointing to the superiority of frequency-separated displays; see, for instance, [10]. However, frequency-separated displays are beyond the scope of the current discussion. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... nauseogenic[*]
For instance, the presentation of visual motion cues in the absence of inertial self-motion cues from the chair in which the observer is seated. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... disorientation[*]
It is puzzling that vomiting would be a response to motion sickness. An evolutionary hypothesis put forward by Treisman [102] suggests that food poisoning produces abnormal perceptions, which are therefore linked with vomiting. Motion sickness mimics the abnormal perceptions due to food poisoning, and therefore triggers the vomiting response. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... observer[*]
As an example of a reference frame which is not a rest frame, consider a train which is perceived as moving through a landscape. One might notice where some person is on the train. Since one is making a judgment with respect to the train, which is perceived as moving, the train acts as a moving reference frame. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... stars[*]
Mike Weissman, a HITL staff member, gives an extreme example of this. He reports that when on a boat in the Pacific Ocean, after a couple of days when looking up at the night sky the swaying of the boat was perceived as a swaying of the stars. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... ``illusions''[*]
These six conditions are termed ``illusions'' here because they are often so labelled, and for lack of a better term in general circulation. If one accepts the physics viewpoint that absolute motion, position and angular orientation are meaningless, then it is incorrect to think of these six conditions as reflecting illusions. They are simply different perceptions. (Although a case could be made for the ``corridor illusion'' being a true illusion, since relative distances are physically meaningful, at least for small velocities.) I would prefer to describe these six conditions with the terms ``rest frame phenomena'' or ``rest frame shifts''. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... frame[*]
In accordance with the argument given above that the selection of rest frame is driven by efficiency considerations linked to the current task. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... reversals[*]
However, an alternative explanation is possible, which is that the more detailed cues simply make the ground representation easier to interpret, for low-level visual perception reasons. The more detailed cues may make it easier to detect the angle between the plane and the ground. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... se[*]
As suggested by the standard sensory rearrangement theory. See Section 2.5. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... dissertation[*]
It should be emphasized that the IVB technique can only reduce the component of simulator sickness which is due to rest frame conflict (i.e., motion sickness). It can not reduce interface sickness. See Section 2.5 and Appendix F for a discussion of this distinction. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... audio[*]
The audio capability was not used for the experiments reported in this dissertation. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... head)[*]
The terms ``direct'' and ``peripheral'' FOV were invented for the purposes of this experiment. I do not know of established terms for these ideas. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... step[*]
Actually, in Chapter 4 I chose to report the range between the last amplitude tested and the next determined by the PEST procedure. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... measure[*]
See Section 2.3 for an introduction to psychological measures in general, as well as their previous application to presence and vection. See Section 3.7 for a description of the equipment. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... measure[*]
Test-retest correlation evaluates the self-consistency of a measure. A high test-retest correlation is crucial to a good measure, because a measure can be no more accurate than it is consistent. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... dependency[*]
See Section 3.8.2 for a further discussion of these issues. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... vertical[*]
See [45] for a discussion of the perception of the vertical, Section 2.3.2 for prior research on presence measures of this type, and Section 3.8.2 for a description of the rod-and-frame test and its relationship to other psychological measures. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
...FOV[*]
48$^{\circ}$ was the widest FOV available for the Virtual Research VR4 HMD used in these studies. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... ear[*]
Humans have other inertial motion detectors, for instance in the torso [104]. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... canals.''[*]
Without going into the details of vestibular physiology, the integration time constant refers to the time period over which a stimulus detected by the semicircular canals continues to be signaled. The integration time constant is partly affected by a recovery time within the semicircular canals (about 4 seconds) and partly, apparently, is due to ``the time taken for central neural events to subside'' [46]. The combined integration time constant is over 12 seconds. By comparison, the duration of one cycle of an inertial wave at .1 Hz, as used for the experiments described here, is 10 seconds. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.../sec[*]
In a few cases, only a resolution of 10$^{\circ}$/sec could be achieved due to equipment or scheduling problems. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... angles[*]
It should be remembered that .1 Hz is at the bottom of the ideal angular motion-detection range for the vestibular channels. Below this frequency, phase lead becomes an issue. See above. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... trials[*]
A possible reason for this incorrect perception will be given in Section 4.4.5, below. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... test[*]
Recall from above that the mean threshold for detection of whole body yaw rotations is 1.5$^{\circ}$/sec. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... dominance[*]
The accuracy of the inertial motion signaling was limited by human response, rather than equipment considerations. Under ideal conditions, simply watching the chair moving with no distractions, it was not difficult to signal the endpoints of the chair motion to within 5$^{\circ}$, often less. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... afterwards[*]
Within minutes. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... trials[*]
It is interesting that, judging from informal discussions, participants never had a clear perception of what the difference was between the inertial and visual amplitudes, or even whether there was a difference. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... AIP4[*]
Running more participants in AIE1 than AIP3 and AIP4 took advantage of the relative briefness of AIE1 to gather more data on reported presence, and served to gain practical experience with new equipment. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... taken[*]
Thus, the ``meaningful'' condition from the current experiment was identical to the ``48$^{\circ}$'' condition from Experiment AIE1. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... .06[*]
This was computed by comparing all data between the two measures with matched participant identifier, treatment condition, and session number. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... .38[*]
This was computed by finding the difference across treatment conditions for matched participants and session numbers, then correlating these differences across measures. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... occlusions[*]
See Section 3.7.6 for background. Related pilot studies and an experiment on inside-out displays are described in Appendix C. The description of Experiments AIIE1 and AIIE2 borrows from [83] and [82]. While I was first author, this research was carried out in collaboration with Dr. Hunter Hoffman. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... display[*]
Actually, due to the design of the dVisor and of the tanning goggles, part of the nasal edge of the screen was visible in the foreground occlusion condition. The effect found below would be predicted to have been even stronger if the foreground occlusion had perfectly blocked the screen boundary. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... formed[*]
See Section 3.3.5 for background, and Section 3.7 for a description of the equipment. This chapter borrows from [80] and [81]. While I was first author, this research was carried out in collaboration with Mark Draper. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... analysis[*]
Stance breaks are a discrete measure, and become meaningless in the case of continuous instability. The same problem does not afflict the other measures. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... 24[*]
The distribution is similar for the reported presence measure. Again from Table 4.7 (page [*]), for 9 matched pairs out of 24 there is no difference between conditions, for 13 matched pairs the meaningful condition was rated higher, and in 2 the random condition was rated higher. This suggests that the frequent lack of difference in the visual-inertial nulling measure data across conditions was due to the nature of the experiment rather than to a specific lack of sensitivity for visual-inertial nulling. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... problem[*]
In a few cases, participants reported feeling better after the experiment than before. This may have been a beneficial side-effect of the instruction to ``relax and close your eyes'' between trials. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... laboratory[*]
For myself, at least, the impression was that the CI and the background grid were mounted on independent (not visible) disks, which were free to rotate independently. The disk the background grid was on appeared to slide in the opposite direction during turns. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... there[*]
See, for instance, the discussion of ``illusions dependent on perceptual hypotheses'' in Levine and Shefner [60]. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... represented[*]
That is, new information can be thought of as being defined in terms of its projection onto known chunks, which act as basis vectors. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... screen[*]
It should be noted that this procedure might introduce noise into the cross-over data, since participants would be faced with a divided attention task. They would need to respond to both the inertial motion extremes and the appearance of visual targets. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... annulus[*]
For those interested in replicating AIIP3 or AIIE3, I should emphasize the importance of not providing useful reference frame cues surrounding the display in the non-foreground occlusion condition. In preliminary studies, not reported in this dissertation, the aircraft icon was kept fixed and the wide FOV condition allowed spatial orientation cues from the laboratory. These studies found a strong effect against the foreground occlusion condition. A possible interpretation, suggested by Dr. Parker, is that one needs to know two orientations to solve the inside-out roll correction problem: the angle of the ground and the angle of the plane. The spatial cues from the laboratory were in agreement with the aircraft icon, which made the orientation of the aircraft icon easier to interpret, which improved performance in the non-foreground occlusion condition. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... simulator[*]
See Section 3.7.2 for a description of the simulator. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
...)[*]
The application described here of foreground occlusions to the binocular rivalry problem was suggested by Furness (personal communication). 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
... argument[*]
Dr. Patterson, a visual psychologist at Washington State University specializing in stereovision, has taken the ``spatial frequency'' position in our own discussions. Ironically, I am in debt to him for the Shimojo and Nakayama [95] reference. 
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Jerrold Prothero
1998-05-14