In Experiment 1 eight five-year-old children received an arbitrary matching-to-sample task with sample stimuli A1 and A2 and comparison stimuli B1 and B2. This task was mixed with a simple, two-choice discrimination task with stimuli A1 and A2. In subsequent tests with B1 and B2 presented alone, the number of responses to B1 was greater when it had been paired with the correct stimulus of the simple discrimination task than when it had been paired with the incorrect stimulus. In Experiment 2 incorrect comparisons were omitted during the matching-to-sample task. Eight children were taught to point to either A1 or A2 first, and to either B1 or B2 second. Eight other children were taught to point to either B1 or B2 first, and to either A1 or A2 second. The effect of Experiment 1 was replicated with the first but not with the second group. The results of both experiments suggest that the functions of S+ and S- in a simple discrimination task can transfer to test stimuli if the training stimuli precede the test stimuli during paired presentations.