| Publications Page | HITL Home |

The Adaptive Effects of Virtual Interfaces: Vestibulo-Ocular Reflex and Simulator Sickness
by Mark Draper

Chapter 7
Longitudinal VOR Adaptation Experiment

7.1 Objectives and Hypotheses

The purpose of this experiment was to gather initial data on VOR adaptation time course. Although re-adaptation data had already been collected in the previous two experiments to determine how long subjects must wait before they are considered fully re-adapted, there were no data on when meaningful adaptation responses begin to appear when a subject uses a virtual interface. This information could potentially support the formation of optimum exposure time guidelines to maximize time spent in the VE while minimizing any negative effects of that exposure (i.e., undesired physiological adaptations and simulator sickness). This experiment was conducted to ascertain when VOR adaptation likely appeared in the previous two experiments by testing VOR gain at 10, 20 and 30 minutes of a 30 minute VE exposure. As such, it addressed objectives 1 and 5 (see Chapter 1).

The hypothesis was that VOR adaptation to the virtual interface would follow a decaying exponential function, with initial adaptation occurring rapidly. Given the saliency of the specific interface used, a 5% change in VOR gain was expected to be noticed within 10 minutes of exposure, with an additional 3 to 10% change occurring between 10 and 30 minutes.

In addition, given the incomplete gain re-adaptation observed after 10 minutes in the Image Scale Experiment, this experiment examined the level of recovery after 20 minutes. However, to reduce potential contextual effects of the chair and HMD, subjects were released from the chair and allowed to move about the building naturally for the re-adaptation period. It was hypothesized that the additional 10 minutes of recovery along with natural movements would return the VOR gain completely to pre-exposure levels.

Lastly, this experiment provided another opportunity to investigate the issue of frequency specificity vs. generalized adaptation across frequencies. The literature generally favors the notion of frequency specificity but the results of the previous two experiments supported the notion of generalized adaptation across frequencies. It was hypothesized that generalized adaptation across frequencies would again occur, given that the same exposure protocol was used.

7.2 Subjects

Two adult subjects (both male, ages 27 and 35) were chosen to participate in this experiment based upon their demonstrated ability to adapt their VOR gain in the past (each had participated in the Image Scale Experiment and showed near average gain adaptation in each condition). Both subjects were in good health with no history of visual or vestibular medical problems. One subject wore corrective lens and had slight astigmatism. Neither subject had experienced the virtual interface for at least two weeks.

7.3 Experimental Design

A two factor, within-subjects design was used, with five levels of TESTTIME (Pre, 10 min, 20 min, 30 min, and 20 min post) and three levels of test frequency (FREQ: 0.2 Hz, 0.4 Hz, 0.8 Hz). The dependent variables were VOR gain and phase estimates at each TESTTIME and FREQ combination, averaged over two trials.

7.4 Experimental Set-up and Apparatus

Experimental set-up was as described in the Image Scale Experiment. To stimulate a significant VOR gain change, the virtual interface was configured with an image scale of 0.5X (minification) and a system time delay of 125 ms for the entire exposure. This configuration presented a significant VOR gain decrease demand.

7.5 Procedure

The procedure was the same as in the Image Scale Experiment, except for the following. In addition to the pre- and post-exposure tests (at 0 and 30 minutes, respectively), VOR data were also collected at 10 and 20 minutes into VE exposure. No sickness data or posture data were collected, though oral sickness reports were recorded during the exposure for health and safety reasons only. Subjects removed the HMD after the 30 min test and were released from the chair so that they could naturally move within the building for 20 min prior to the final VOR test. Eye calibrations were completed: 1) prior to collecting pre-exposure data, 2) immediately after collecting the 30 min data, and 3) prior to the 20 min post testing.

7.6 Statistical Analysis

Given that two subjects were run, only descriptive statistics are presented.

7.7 Results

Table 22 presents the VOR gain change over time for each subject, along with the average gain values and percent gain changes across subjects. Gain change over time per subject is presented graphically in Figure 79. The average percent gain change across subjects over the 30 minute exposure was 11.2%. Most of this change (7.8%) occurred within the first ten minutes. There did not appear to be any additional change in gain between 20 to 30 minutes. After 20 minutes of re-adapting to the real-world through natural movements, the VOR gain had returned to within one percent of its pre-exposure value.

 

 

VOR gain change at each test frequency over time (collapsed across subjects) is shown in Figure 80. All frequencies resulted in similar percent gain changes over the 30 minute exposure (between 10.5 to 13% gain decrease). The individual curves varied a bit, as the 0.2 Hz curve levels off after 10 minutes, the 0.4 Hz curve levels off at 20 minutes, and the 0.8 Hz curve actually suggests a slight gain increase between 20 and 30 minutes. However, the small sample size prohibits detailed speculations on curve peculiarities.

 

7.8 Discussion

Both subjects responded similarly and in accordance with the literature on adaptation time-course (Collewijn, et al., 1983; Welch, 1986). Subjects showed strong initial gain decreases within the first ten minutes, followed by less substantial changes thereafter. Using these data, a gain adaptation time constant can be roughly estimated to be 10 to 12 minutes. Note that gain adaptation would likely continue to occur beyond 30 minutes at a reduced rate.

However, it was interesting to note that neither subject showed any further gain decrease after 20 minutes. The reason for this is not entirely clear. The most likely explanation is that adaptation rate lessened after approximately 20 minutes and this decreased rate of adaptation was not picked up by the ISCAN system. Another possibility is that the images presented during the final 10 minutes were for some reason less conducive to VOR gain adaptation. This is less likely, as these images were successfully used in earlier experiments. Finally, some other uncontrolled factor may have been involved.

Implications of these data regarding the formation of optimal exposure time guidelines are quite interesting. Given that most of the gain change occurred early in the exposure period, it seems somewhat pointless to attempt to optimize exposure time to �avoid the onset� of adaptation. However, another alternative for avoiding long re-adaptation effects exists given the results of this experiment. If a subject is repeatedly exposed to the same virtual interface for short periods of time, his/her adaptive mechanisms would be forced to adapt to the VE and then re-adapt to the real world several times in succession. This process could serve to speed up both adapting and re-adapting processes through the formation of adaptation sets (Parker, personal communication, 1997; Shelhamer, et al., 1992). Re-adaptation time-course would thus be shortened, which lowers the risk of ill-effects due to the interface. Therefore, the subject�s ability to quickly adapt could be employed to facilitate re-adaptation and lesson post-exposure safety risks. This is not a novel concept; only the application of an existing technique to virtual interfaces.

Once again, these data support the notion of generalized adaptation across frequencies. The three frequencies resulted in nearly the same gain change (10.5 to 13.3% decrease) over 30 minutes. However, while most gain change occurred early at 0.2 Hz, adaptation appeared to be more gradual at 0.4 Hz and 0.8 Hz. Perhaps the majority of early adaptation occurs at the dominant frequencies of head movements while adaptation at lesser encountered frequencies requires more time. Clearly more research is required to test this assertion.

---

Human Interface Technology Laboratory