Association for Behavior Analysis International

Error-correction procedures are ubiquitous in instructional programs. A number of unique error-correction strategies have been shown to be efficacious and subsequent research suggests that a single procedure may not be optimal under all conditions. Complicating this line of research are the numerous strategies that include seemingly minor procedural differences or use unique terms to describe similar conditions. Further, few systematic evaluations of common error-correction procedures can be found in the extant literature, which limits the generalizability of findings in this area. This symposium includes three studies examining the potential variables underlying the effectiveness of error-correction procedures previously described in this literature. Participants include children and adolescents with and without developmental disabilities receiving specialized intervention in educational or clinical settings. The findings of these studies may contribute to the broader understanding of the effects of error-correction procedures and to the refinement of instructional methods by increasing the effectiveness, efficiency, or acceptability of these procedures. Future research and implications for applied practice are discussed.

Practitioners, graduate students, and board certified behavior analysts.

The onset of differential reinforcement of unprompted responding has not received adequate attention in the skill acquisition literature and, perhaps more importantly, has not been sufficiently controlled in comparative studies. Take for example recent research on error-correction procedures, which have produced unique findings potentially resulting from unique onsets of differential reinforcement across conditions. The current study serves as a systematic extension of this research by comparing the efficiency of acquisition in immediate and delayed onset of differential reinforcement conditions for three participants with developmental disabilities using an adapted alternating treatments design. In all four comparisons, the delayed onset condition required fewer trials to mastery, although superior performance was observed early in acquisition in the immediate onset condition in two comparisons. These findings failed to replicate those of prior research on differential reinforcement onset, possibly due to differences in participant characteristics, target tasks, or other procedural modifications. Nevertheless, the current findings also suggest that onset of differential reinforcement is a critical variable to consider in skill acquisition programming and should be controlled in applied research.

We conducted a component analysis of learn unit (LU) instruction that included delivering antecedent instructions and differential consequences for correct and incorrect responses. Six preschoolers with and without disabilities participated in the study. In the LU condition, researchers praised correct responses and implemented a correction procedure contingent on incorrect responses. LU instruction was compared to conditions that omitted consequence portions of instruction. In the reinforcement-only (RO) condition, researchers applied positive reinforcement operations to correct responses and ignored incorrect responses. In the correction-only (CO) condition, researchers ignored correct responses and implemented the correction procedure following incorrect responses. We manipulated this independent variable across learning educational and abstract stimuli and measured acquisition rates, cumulative duration required until mastery, and maintenance of responses. The results showed that the learning procedures of LU and CO were both effective on teaching listener responses for all participants and were more effective than the RO procedure. Furthermore, LU instruction that involved both reinforcement and corrections was not necessarily more effective than the procedure of CO on teaching listener responses. The results also suggested that the correction procedure was more effective than the reinforcement procedure on the maintenance of learned skills.

Multiple response repetition (MRR) is a frequently used error-correction procedure that requires the learner to repeatedly emit a response following an error. Several studies have found the MRR procedure to be effective; however, it is unclear if these effects result from the repeated emission of a response, similar to positive overcorrection procedures, or the period of signaled extinction, similar to time out from positive reinforcement. The current study extended prior research by comparing the effects of three MRR conditions (i.e., relevant, irrelevant, and yoked-delay) in an adapted alternating treatments design. A 5-year-old female with autism spectrum disorder participated. Across all three comparisons, targets were acquired in the yoked-delay condition in the fewest number of exposures. This finding suggests that timeout from reinforcement may serve a prominent role in the effectiveness of MRR procedures. This result is particularly promising as it may provide early evidence that a less-intrusive version of an effective procedure may produce similar or even better discrimination outcomes.