A study was undertaken to investigate the effects of directed response training to focus attention and to assess the impact of pictorial integrated stimuli on incidental learning. A total of 140 second- and fifth-grade children were administered a two-choice discrimination learning task consisting of three parts: original learning, overtraining, and a nonreversal shift task. The original learning task was the same for all participants. During overtraining trials, children either made a response that the experimenter said to make (response training) or responded without help. A second task was administered in which, for some children, "real-life" pictorial integrated stimuli replaced geometric forms. Other children saw the same stimuli as in original learning. Results indicated that response-trained second graders learned the shift task more rapidly than did untrained second graders. The poorest shift task performance was made by response-trained fifth graders. In addition, learning was more rapid when pictorial stimuli were used on the shift task. These findings supported the conclusion that younger children selectively attend when an attentional strategy is available and generalize that attention to pictorial integrated or more realistic stimuli. Poor learning by response-trained fifth graders raised questions as to the appropriateness of the response training strategy for older children's learning. (Author/RH)