Re-evaluation of a programmed method to teach generalized identity matching to sample.

Programmed training in identity matching to sample was given to six participants who had severe mental retardation, mental age-equivalent scores of 3.0 years or less, and histories of failures in prior assessments and training attempts with standard procedures. An intermediate goal of the training program was to establish one-trial discrimination learning (OTDL), where new discriminations are acquired after a single training trial, OTDL was included because an analysis of the task requirements for identity matching suggested that it could be a prerequisite skill. One participant was eliminated from the experiment when stimulus control by prompting procedures broke down relatively early in training. Only one of the remaining participants achieved OTDL. When the program was modified to eliminate OTDL as an intermediate goal, for participants completed it and passed tests for generalized identity matching with high accuracy scores. The program was partially successful with the sixth participant in that it established highly accurate and reliable identity matching when different stimuli were displayed on every trial (nonconditional-function matching), but not when the same set of comparison stimuli was displayed on every trial (conditional-function matching). The results showed that (a) one-trial discrimination learning appears to be sufficient but not necessary for identify matching, and (b) the program successfully established identity matching in a majority of difficult-to-teach students who had well-documented failures to learn by standard teaching methods.