Two Eyes, 3D

About a year ago I ran my first research study as part of my program here at Tufts. I spent a week at the Adler Planetarium in Chicago giving middle school aged kids spatial tests using their GeoWall. I'm presenting the results at NARST this April and will submit a paper to a journal soon. But I'll summarize it here, because the results were not what I expected.

Basically, I gave the kids 3 spatial tasks to complete, all taken from existing literature. One was completely 2D in nature (a letter rotation task). One involved a single 3D mental manipulation (block rotation task). And one involved lots of 3D mental manipulations (a paper folding task).
The students took one version of the three tasks presented on pieces of paper. Then they were given the same tasks, in a different order, via the stereoscopic GeoWall. I recorded accuracy, time on task and a short post-task interview.

My hypothesis was that the GeoWall would have no effect on the 2D task, a little effect on the moderate 3D task and a larger effect on the multiple 3D task. I thought the stereoscopic presentation would make it easier for the students to "see" the objects as 3D and therefore make 3D manipulations.

Wrong! Results show that there was no change in accuracy on any of the tasks between the paper and the stereoscopic presentations. (Note that no change means just that, not only did it not improve but accuracy didn't drop either.) However, completion time was affected by the difference in presentation formats. For the 2D task, there was no difference in time on task. For the moderate 3D task there was a moderate, but statistically significant, increase in completion time for the stereoscopic format. For the larger 3D task there was an even larger increase in completion time. In other words, completion time increased in line with the 3D nature of the task.

What does this mean? It means my hypothesis was waaaaayyyy off! The stereoscopic projection had no affect on accuracy and actually SLOWED the time it took to complete the task. Using cognitive theory and data from the interviews, I interpret that as meaning there is an increase in cognitive load when students use the stereoscopic system. Specifically, I think the kids are still thinking in 2D even when presented with a stereoscopic presentation. So their mind is taking the stereoscopic object, converting it to 2D and then RECONVERTING it to a 3D mental model.

If so, there are some things I may be able to do to resolve this. One, use cue theory to build in more depth cues into the image (i.e. make it more realistic). Two, give the students more training time to get used to the stereoscopic environment. I hope to do that in a small pilot study this spring at a local middle school.

Of course there are lots of details of this study I'm leaving out. My paper currently sits at 28 pages, not counting tables and figures. :) If you want more detail, wait for the NARST presentation or publication in a journal. Or e-mail me. :)