Association for Behavior Analysis International

Practitioners make decisions about skill-acquisition programming for individuals with autism spectrum disorder (ASD) based on best-practice recommendations in the literature. However, there is still much we do not know when providing recommendations on how to best develop and implement skill-acquisition programming. This symposium is a group of presentations that look to refine instructional procedures when teaching skills to individuals with ASD. The first presentation will evaluate the types of prompts used during discrete-trial instruction. The second presentation will compare the materials used (flashcards compared to tablets) used during discrete-trial instruction. The third presentation will examine the level of instruction necessary for preschool children to learn social skills. Finally, the fourth presentation will compare chaining procedures to teach cooking skills. The presentations will include implications for clinical practice. A discussant will review the presentations and provide suggestions for future research and clinical considerations.

Discrete-trial teaching is an evidence-based teaching strategy that may be individualized to each learner. One way to individualize discrete-trial teaching is the type of prompts which are used. Prompts may include: modelling the correct response, gesturing, providing verbal cues, and physical guidance. There is limited research on the relative efficiency and effectiveness of these different prompt types making it difficult to identify which prompt will be best for each learner (Seaver & Bourret, 2014). In addition, the learners preference for how they are taught is not always considered. The present study compared the relative effectiveness of three different prompt types to teach a receptive identification task for three boys with autism. An adapted alternating treatments design was used to compare a gesture, modeling, physical guidance, and a no-prompt control condition. For one participant, the physical prompt was the most effective. For a second participant, the model prompt was the most effective. Following mastery, a concurrent chains preference assessment was conducted to assess individual preference for prompt type. Both efficacy and client preference may be used to determine prompt selection during discrete-trial teaching.

Children with developmental disabilities are at increased risk to develop problem behavior in the absence of appropriate social and communication skills. Hanley, Heal, Tiger, and Ingvarsson (2007) created a classwide preschool life skills (PLS) program that taught young children to request teacher assistance, tolerate delays or denials in the delivery of materials, friendship skills, and functional communication skills. The purpose of the current study was to assess the effectiveness of the PLS program when implemented with children with developmental disabilities and at risk for developing classroom problem behavior. A multiple-baseline across-behaviors design was used to demonstrate the effects of instruction, differential reinforcement, and error-correction procedures that systematically increased as necessary for participants to acquire the target skills. Probes were also conducted in the classroom with adults and peers following acquisition to test for generalization. The level of instruction varied across participants. Four participants acquired the skills with least-to-most prompting and praise. Three participants required individualized instruction and reinforcers. Overall, there was little generalization of the acquired skills to peers and adults in the participants classrooms.

This study evaluated the utility of a clustered forward chaining (CFC) procedure for teaching a 23-year-old male with autism to follow written recipes. CFC incorporated elements of forward chaining and total task chaining by teaching a small number of steps (i.e., clusters) using total task chaining, and introducing new clusters sequentially, contingent upon sustained mastery of previous clusters. For each of three recipes targeted in a multiple probe design, we organized 45 total steps into 15 functional triads (i.e., read, do, record). We then organized the triads into three clusters (five triads each). We used a 5-s constant time delay to prompt responses for every step within a training cluster and immediately prompted responding for all untrained clusters. We probed for mastery of the entire response chain after demonstrations of mastery of each individual cluster (and before initiating training for a new cluster). Results showed that CFC successfully established independence of all three response chains and required considerably fewer training trials than what would have been required had we implemented a traditional forward chaining procedure. Maintenance probes 3-5 weeks after training demonstrated continued independence of all three cooking recipes.