ON HOMOGENEITY OF STATES AND TRANSFER IN LEARNING TO CATEGORIZE

Two groups of subjects (A and B) were first trained on samples of states with different homogeneity and then tested on common samples. It was found that B learned faster with the more homogeneous samples of items. This group also performed better in terms of transfer. Both groups did better than the control group (C) on the test series. A fourth group (D) had items which, according to Weber's law, would appear less similar than those of the control group. As was to be expected, C performed better than D. The outcome may be interpreted in terms of categorization. When states leading to the same event become more homogeneous, categorization of the input information is easier, and both learning and transfer facilitated.