Association for Behavior Analysis International

Children with autism often have deficits in generalization of skills and generalization has to often been directly programmed. Matrix training and multiple-exemplar training are two ways to effectively program for generalization. This symposium has two studies that will demonstrate the application of these two generalization strategies. Specifically, the first study will use matrix training to establish a generalized repertoire of direction following. This study will also discuss the implication of the size of matrix and efficiency of generalization. The second study will examine the effects of multiple exemplar training to establish a generalized repertoire of play sequences in young children

Graduate students and professionals

The purpose of the current study was to investigate the effects of implementing a diagonal matrix training layout to teach a generalized repertoire of direction following to four male preschoolers with autism spectrum disorder (ASD) using two different sized matrices (i.e., 3x3 and 5x5). A multiple probe across participants experimental design was implemented to compare generalization efficiency. Participants were taught trained combinations using a constant prompt delay and a model prompt. Correct responding was defined as the participant performing the correct action and then selecting the target stimulus (e.g., “Jump to bowtie”) within 5 s of the discriminative stimuli. In addition, generalization was assessed on untrained direction combinations for within and across matrices. Interobserver agreement, procedural integrity and social validity was also assessed. The results demonstrated that matrix training led to generative responding across all participants and maintained up to four weeks after training was completed. Additionally, it was demonstrated through an efficiency measurement that a 5x5 matrix produced the fastest rates of generalization acquisition when compared to the 3x3 size matrix.

Children with autism (CWA) often lack play skills. CWA engage in repetitive manipulation of play materials (MacDonald et al., 2005) and are more likely to respond invariably to leisure materials than their typically developing peers (Bancroft et al., 2016). Current play research shows a variety of teaching methods have been used to teach play behavior, however play is often lacking novelty and generalization. The purpose of the current study was to examine the effects of combining an in-vivo model and multiple exemplar training to teach 9 dress up play sequences across 3 themes (fire fighter, chef, and doctor). One participant was trained on 9 dress up play sequences using in-vivo modeling. Probes across the 9 dress up play sequences were conducted to test for mastery and generalization within and across play themes. Results indicated increases in differentiated play, pretend play, and scripted play from that observed during baseline sessions.