Arithmetic accuracy and stability as a function of token reinforcement



Journal Title

Journal ISSN

Volume Title



The effects of token reinforcement procedures on the accuracy and stability of arithmetic performance of eleven children in an intermediate public school learning disabilities classroom were assessed using an intrasubject replication design. Standard computational problems were employed as the tasks for which accurate completion earned tokens exchangeable for monetary reward. In addition to a baseline period three different types of pay conditions prevailed such that the requirements for earning tokens were varied. In one condition, one correct response produced a penny token, and in the other two conditions a penny token was contingent on two and four correct responses, respectively. Incorrect or incomplete problems were not penalized. It was consistently observed that the levels of both accuracy and stability of students' performance on the token contingencies was significantly lower than baseline levels. Consistent shifts were observed in the students' performance as a function of reinforcement contingencies. While the findings appear contradictory to the results of most token studies reported in the literature, some design and implementation weaknesses could have confounded the analysis of contingencies. Uncontrolled extraneous variables are discussed, and suggestions for future research in the area of accuracy and academic performance emphasize methodological issues, with recommendations for convergent studies. The argument is offered that, should deficits in the present research be corrected and repeated studies continue to yield similar results, the possibility be considered that token procedures as established in controlled settings may not readily transfer to naturalistic ones.