These experiments tested the hypothesis that an instructional strategy involving guided discovery will promote meaningful learning of problem solving concepts by activating an assimilative set during learning. In order to investigate the effect of discovery training and concrete manipulation, preschool children were given training in one to one correspondence. In the first study (involving 19 subjects) a discovery method was compared to a matched expository method, both of which involved active manipulation of objects. The second study (involving 24 subjects) replicated the first study using different materials and included a third group (observation training) in which subjects did not manipulate objects. The learning outcomes were evaluated by tests of short-term recall, short-term transfer, long-term recall and long-term transfer. Performance on the posttests administered by a "blind" experimenter revealed a pattern in which groups did not differ on short-term recall but the discovery group excelled on far transfer (conservation) and delayed recall. The effect of guided discovery on the acquisition of broader learning outcomes was discussed. (MS)