We implemented a classroom experiment in an authentic physical-geology laboratory course in which different training conditions were tested for the teaching of rock identification and categorization. In Condition 1, students learned rock identifications by interacting with typical physical samples of the to-be-learned categories. In Condition 2, students learned by interacting with multiple three-dimensional (3D) virtual rock models displayed with a computer-controlled system. Condition 3 was a hybrid condition that combined use of the physical samples and the 3D virtual models. Training conditions that made use of the 3D models implemented procedures that were theoretically motivated by core findings from literature on the cognitive science of category learning. Following training, students in all three conditions were tested on their ability to identify novel physical samples from each of the to-be-learned categories. Although the magnitude of the effect was not large, final test performance was significantly better in the hybrid condition than in the physical-only and 3D-model-only training conditions. Notably, test performance in the 3D-model-only condition was not significantly different from the physical-only condition. Thus, results support that training rock identification with the present 3D-model based computer technology appears to provide an effective instruction tool even on its own, and that this technology could enable the geoscience education community to provide instruction to student populations unable to benefit from in-person physical instruction. An archive of 100 of the 3D virtual models is publicly available through UAB Digital Commons, an open access repository service of the University of Alabama at Birmingham Libraries.
