Effects of Proportion of Positive Instances upon Concept Learning

Abstract

The purpose of this study was to determine the effects of varying proportion of positive instances in the stimulus sequence over a wide range (0.1, 0.2, 0.4, 0.6, and 0.8). The results showed that number of trials to criterion generally increased as proportion of positive instances decreased. The function for number of errors was convex, however, reaching a maximum at 0.4 and decreasing toward the extremes. Detailed examination of the data disclosed that Ss in the 0.1 and 0.2 conditions were learning the probabilities and adopting a strategy of “when in doubt say no.” The effects were attributed in part to variation in information rates as proportion of positive instances changed and in part to differences in difficulty of utilization of information from positive and negative instances.