Association for Behavior Analysis International

This symposium explores how tacting correct and incorrect implementation of behavioral training procedures may enhance treatment integrity when carrying out those procedures. Train to Code (http://www.ai2inc.com) is an internet-enabled program that uses multiple exemplar training to teach trainees to tact behavioral events as an expert does, as they occur sequentially in a video. By definition, when trainees have achieved expert tacting performance with the targeted events, high inter-observer-agreement with the expert has been attained. Early in the evaluation of this program, we noted that when trainees learn to consistently tact steps an expert therapist takes when carrying out a behavioral training procedure, they show improved skill in carrying out the same procedure. This may be conceptualized as tacting if-then relationships in the environment (i.e., deriving self-rules) and then following those rules at a later time under similar circumstances (Fryling, Johnston, and Hayes L, 2011). We explore what behavioral principles and phenomena might be involved in such transfer, what practical considerations impact the success of this training, and provide two examples of the use of this approach to enhance skills in carrying out behavioral training procedures.

Train to Code is an internet-enabled program that uses multiple exemplars to teach trainees to tact behavioral events as an expert does, as they occur sequentially in videos. When the events to be tacted involve implementation of behavior analytic training procedures, we have found that, after trainees have achieved mastery (high inter-observer-agreement), they often carry out those procedures more correctly compared with their baseline performance (i.e., with improved treatment integrity). Since ABA interventions must be implemented with high treatment integrity to be effective, and this is a major issue in the field, we are excited that TTC may play a role in achieving this goal. In this presentation, we seek to identify behavioral principles that may account for this transfer, which we view as delayed observational learning. We also discuss behavioral phenomena that appear to be related to this finding, such as abstract control in simple and more complex (higher order) verbal relations, rule-governed behavior, self-talk in skilled performance in judged sports, and the role of tacting a model's behavior in imitating that behavior at a later time.

The purpose of the present study was to assess the feasibility of an observation and coding software system (i.e., Train-To-Code [TTC]) to teach university students implementation of Phase 3A of the Picture Exchange Communication System (PECS). The software was used to develop a customized program that coached participants by presenting multiple exemplars of correct and incorrect implementation of each step of this phase of PECS. Participants were required to code specific behaviors in accordance with a taxonomy developed by the experimenters. The training program provided prompts and feedback in real time based on participants' level of performance and required mastery of seven levels of training, each with fewer prompts and feedback, until an expert level of unprompted coding performance was demonstrated. Four undergraduate students with no prior experience in the implementation of PECS were recruited. A multiple baseline design across participants was used to evaluate the effects of the training on treatment integrity performance during role-play with a confederate learner before and after training. Results showed improved performance relative to baseline following training, and maintenance of performance at 2-4 weeks follow-up. Implications of these findings for staff training in applied settings will be discussed.

ABA staff are typically taught quite differently from how those same staff members teach clients. Clients are individually evaluated for existing skills and knowledge, then Individualized Educational Programs (IEPs) are developed that take into account a wide variety of important variables that impact the educational program designed for each individual client. Staff, on the other hand, are more typically taught using techniques more commonly found in classroom instruction and is not individualized. That is, they are taught in groups, are presented slide presentations and limited narrated video examples, and may participate in some discussion and/or even role playing. Ignored are individual differences in prior training or content exposures, the pace of delivery, and most other teaching/learning variables. Train-to-Code (TTC) is a software system designed to help individualize this process through highly adaptive and interactive process pacing, prompting, and feedback or error correction procedures. But that is only a beginning in individualization. The design of the content and task demands of those who supply content for TTC presents a whole new set of instructional challenges that should also be very carefully considered and individualized for each trainee. This presentation reviews data showing how critical this requirement really is.