Examining Generalization Performance in a Conditional Discrimination Task for Learners with Moderate to Severe Intellectual and Developmental Disabilities

University of Minnesota Ph.D. dissertation. July 2016. Major: Educational Psychology. Advisor: Joe Reichle. 1 computer file (PDF); x, 193 pages.This study examined how differences between teaching stimuli in a conditional discrimination task impacted the discrimination/generalization outcomes among learners with intellectual and developmental disabilities (IDD). Three color sets were identified in a pilot study among 30 typically developing adults for the experiment. Subsequently five young children diagnosed with Down syndrome and moderate to severe intellectual disabilities participated in the experiment that employed an adapted alternating treatment single case experimental design. Prior to the intervention, participants’ performance within each color set of nine stimuli was baselined to ensure room for improvement. During intervention, each participant received three intervention conditions in an alternated manner. The three experimental conditions included two conditional discrimination conditions, that differed only in the degree of color difference (maximal or minimal) between two teaching stimuli (i.e., SA and SB), and a simple discrimination condition that involved only one teaching stimulus (i.e., SA). The assignment of the three color sets to the three intervention conditions was counterbalanced across five participants. As soon as the participants reached the mastery criterion on an intervention condition generalization testing was implemented involving all nine stimuli within the color set. Results suggested that the simple discrimination was acquired most quickly. The comparison of the generalization gradients collected during baseline and generalization testing demonstrated that a simple discrimination was not less effective than the two conditional discriminations in enhancing the generalization performance subsequent to intervention. The minimal-difference conditional discrimination was relatively more effective in enhancing the generalization performance at least within one stimulus class than the maximal-difference one (and the simple discrimination). Limitations and implications of this study are discussed