Training of rock classifications: The use of computer images versus physical rock samples

ABSTRACTA highly controlled laboratory experiment was conducted that suggested computer-based image training of rock classifications can provide a useful supplement to physical rock training. Two groups of participants learned to classify samples of 12 major types of rocks during a training phase. One group was trained using computer images of the rock samples, and another group was trained with physical rock samples. A third group that was familiarized with images of the samples but did not receive initial classification training served as a control. The participants' ability to generalize their training to the classification of novel, physical rock samples from the 12 types was then assessed in a test phase. All groups received trial-by-trial feedback during this test phase; still, the image-based and physical rock training groups maintained a significant performance advantage (75.2% correct) compared to the control group (37.5% correct). The group that received physical rock training performed only sli...