Building response classes: a comparison of two procedures for teaching generalized pouring to learners with severe disabilities.

Five individuals with profound mental retardation received instruction on liquid pouring. The effects of two training strategies on performance with nontrained pitchers and receptacles were compared within a split multiple baseline design across subjects. One training approach followed general case instruction guidelines in which both difficult and easy teaching examples were selected that sampled the range of relevant stimulus and response variation. The second training approach approximated the early steps in an easy-to-hard sequencing strategy and involved using only a set of easy teaching examples that did not sample the full range of relevant stimulus and response variation required for successful pouring in natural settings. Results indicated that only after training with the full set of general case examples did the learners become competent with the nontrained generalization probe tasks. In addition, the set of easy examples resulted in learners performing topographically predictable errors with the generalization probe tasks.