Three-Dimensional spatial memory graining using virtual reality : effects of reference frame and point-of-view

Abstract

Humans are usually able to imagine how a terrestrial environment will appear from a new perspective, and usually avoid disorientation because of their ability to transform spatial mental models of their environment. However, astronauts must transform their spatial mental models in three dimensions, and our previous experiments have shown that this is initially not an easy task, though most subjects eventually learn to orient themselves, particularly those who are taught specific strategies. The present experiments are the first to investigate the effects of using a Self-rotation vs. an Array-rotation assumption when transforming mental models of an environment, and to see whether subjects can change rotation modes easily. Also, does performance depend on the point-of-view (Inside vs. Outside)? Subjects (n=29) had first to memorize the arrangement of 6 animal picture icons painted on the interior walls of a virtual cubic room. The room was kept in a fixed orientation during the first four trials in order to allow the subjects to memorize icon locations in a canonical orientation. In each subsequent trial, they were then asked to imagine that either they (Self-rotation mode) or the array (Array-rotation) had rotated. Based on the identification of icons displayed in front and below them, they had to deduce their new orientation and indicate the direction of a specific unseen "target" icon. The speed and accuracy of their responses were recorded. The experimental design was blocked by point-of-view (Inside vs. Outside). Each of the four groups was assigned a specific rotation mode in the first ("training") phase of the experiment; they were then instructed to switch to the alternate mode in the second ("testing") phase. We looked for but found no strong evidence that the groups differed greatly in intrinsic skill. Most subjects managed to master their assigned point-of-view and first rotation mode by the end of the training phase, although learning rates differed between groups. During the training phase, Arrayrotation groups performed significantly better than Self-rotation groups (Kruskal-Wallis, p=0.05). However, after training, the group that trained Inside but was tested in Self-rotation mode performed significantly better than the others (Kruskal-Wallis, p=0.005). Even though they had no practice trials during the testing phase, most groups were able to switch rotation modes gracefully, without significant performance decrement. Targets on the left and right were less accurately identified than those in other directions. Implications for spatial memory training of astronauts are discussed.