The Influence of Immersion in Virtual Environments on Performance of a Modelling Task: An Experiment

GR/M85692

Principal Investigator: Mel Slater

Visiting Research Fellow: David Mizell
The Boeing Company, Seattle, USA.

Research Assistant: Bernard Spanlang

An experiment was carried out to assess the impact of different virtual presentation and interaction styles on performance of a modelling task. Subjects were asked to physically reproduce a puzzle made of upright twisted wires positioned on a board. A reference puzzle was presented to them on a desktop monitor, through a head-mounted display (HMD) or physically in the real world. They carried out the modelling task with 3 different reference puzzles randomised over 3 display conditions: the real world, the desktop, the HMD, and then a repetition of the first condition again in order to estimate learning over time. The puzzles made by the subjects were scored as to their accuracy against the reference puzzle. The hypothesis was that other things being equal, the accuracy would be greatest for the real-world display and least for the desktop monitor display - in particular the HMD results would have a greater accuracy than the desktop monitor results. 36 subjects were allocated randomly over the design. Other information was collected such as spatial ability (a SAT test), reported level of nausea induced by the display types, and various subjective responses such as the sense of being in the same space as the reference model.

The results show that in fact the desktop display if anything resulted in less errors and took less time for completion than the stereo head-tracked HMD. The least error rate was for the real condition. Analysis shows that there was a tendency for females to do less well than males, and independently that the SAT score was positively associated with better task performance. Although the 3 puzzles used had been calibrated before the experiment to be of equal difficulty, in fact there is evidence to suggest that they were not the same. This led to some interaction between display type and puzzle, and for one of the puzzles subjects tended to perform better with the HMD than with the desktop. The reason for this is not known. Finally, analysis with a composite error score suggests that higher SAT is positively associated with better performance for the HMD condition only.

This research is supported by EPSRC under grant GR/M85692 and an associated Visiting Fellowship for Dr David Mizell (GR/M86200).