Four male and five female Negro children (ranging in age from 5 to 9 years) served in an experimental comparison of the discriminative control exerted by sex and race aspects of other children. A baseline was established in which color photoslides of a Negro girl and a Caucasian boy differentially controlled responding on two push buttons. Responses during test probes picturing children from each race-sex grouping indicated predominant control by the stimulus dimension of sex. Six of the subjects were then reinforced for differentially responding to photoslides of Negro and Caucasian children. During subsequent training, photoslides of new individuals were introduced. Responses during test probes which followed each of these phases showed no generalization along a stimulus dimension of race. Effects of racial stimulus class labeling upon generalization were studied in a final phase following a multiple baseline design across subjects. It was found that for four subjects a verbal stimulus class labeling requirement, when paired with differential reinforcement for button-pushing, was sufficient to shift control to a dimension of race. (Author)