The purpose of the study was to test the discriminative cue hypothesis by attempting to train pupils to increase effort following a high expectancy condition and to reduce effort following a low expectancy condition. Fourth grade high-IQ low achievers were used as subjects in the study, because the authors felt that while they would learn fast, they would not be as likely as high-IQ high achievers to have previously learned responses to expectancy conditions. The results of the experiment support the hypothesis that children can learn to behave differentially to different adult expectancies when the expectancies are followed by consistently different outcomes; thus the "discriminative cue" function proposed for expectancy conditions is supported. The study provides a valuable model for an analysis of teacher behavior, implying that many kinds of teacher statements and non-verbal cues may potentially serve as discriminative cues for children. Used positively, such cues can increase responding and even bring about initial responses in new situations. (SES)