From: David Waller <dwaller@u.washington.edu>
Subject: Re: HUMAN-FACTORS: Spatial abilities in VR
Date: Wed, 4 Jun 1997 11:54:31 -0700
Message-ID: <Pine.OSF.3.95.970604085111.22125A-100000@saul7.u.washington.edu>
Organization: University of Washington



On Mon, 2 Jun 1997, Marilyn Salzman wrote:

> I was wondering if any of you have examined whether an individual's
> spatial ability influences his/her ability to interact with, navigate
> in, or learn in a VR environment.  Specifically, I am interested in 1)
> whether or not you found that spatial ability significantly predicted
> performance, and 2) what tests you used.

The results of a recent experiment we conducted have some relevance to
your question.  Some of the participants in this experiment repeatedly
entered a VE to learn about the spatial layout of a real-world maze
environment.  They were then tested on their knowledge of the real
world maze in number of ways.  These tests were:

1.  Time through the real maze while blindfolded after 1 training session
in the VE.  (Blindfold-1)  

2.  Time through the real maze blindfolded after 2 training sessions
(Blindfold-2)  

3.  Time through maze blindfolded after 3 sessions (Blindfold-3)  

4.  Blindfold-4 

5.  Blindfold-5 

6.  Blindfold-6 

7.  Whether the participant tried a familiar route after a path was 
blocked.

8.  Whether the participant tried the most direct route after a path was
blocked.  

9.  Number of items correct on a true-false test in which participants
were presented with correct or incorrect depictions of various routes in
the maze. 

10.  Number of items correct on a true-false test in which participants
were presented with correct or incorrect depictions of the relative
locations of various landmarks in the maze environment.


Before the VE training, we administered the Guilford-Zimmerman test of
spatial orientation.

Pearson product-moment correlations with the Guilford-Zimmerman were as
follows (n = 60) 

	Variable #	r
	1. 		-.388 **
	2.  		-.157
	3.  		-.176
	4.  		-.196
	5.  		-.199
	6.  		-.242
	7.  		-.114
	8.  		-.033
	9.   		 .368**
	10.  		 .269*
	11.  Gender:  	-.383**
	12.  Experienced sim sickness:  .231

** indicates significance at a 2-tailed .01 level
*  indicates significance at a 2-tailed .05 level

----------------------------------------------------------------------------

In summary, the Guilford-Zimmerman test of spatial orientation was
predictive of some of the more interesting abilities involved in acquiring
spatial information from a VE.  Its correlation with the first time
through the real-world maze blindfolded was particularly high, even after
controlling for gender differences.  However, it did not correlate highly
with subsequent learning.

I'm happy to provide you with more information about this on request.  By
the same token, I'll appreciate your letting me know about the other
responses you get to your question.  Take care,

Dave Waller

dwaller@u.washington.edu