Examining the "punishing rewards" assertion: Faded token reinforcement and children's intrinsic motivation for an academic task

John Christopher McGinnis


Almost 30 years of study have overwhelmingly supported the notion that naturally occurring interest in an activity will be destroyed if the activity is rewarded. The idea that "rewards punish" has taken its toll in applied settings where structured reward programs are valuable and, in many cases, necessary to allay human suffering. Mostly within the last decade, a cohort of scholars has critically examined the assertion that the systematic use of rewards in applied settings is a harmful practice to be avoided. The methodology and theoretical underpinnings of previous studies have been reevaluated to some extent, and more research has been conducted. The present work contributed to the forum an investigation employing a multi-element within-subject experimental design. The amount of time spent on each of three tasks (one of which was a series of math problems), the quantity of work produced, the quality of the work, and subjective interest were ongoingly assessed for 5 children over 26 15-minute sessions. Following baseline, the math task was reinforced with tokens (i.e., adhesive foil stars) redeemable for secondary reinforcers (i.e., edibles, school supplies, and small, inexpensive toys). Token reinforcement was then faded and eventually withdrawn. Conclusions drawn from the present study corroborate those from previous studies employing contingent repeated reward delivery. No evidence was found to suggest a detrimental nature of rewards using procedures reflecting those typically implemented by behavior analysts. Children with initially high levels of engagement and interest in the math task generally exhibited higher levels of engagement and interest once reinforcement procedures were withdrawn. Children with initially low levels of engagement, unsurprisingly, were observed to engage in the math task as a means to obtain the rewards; however, interest was generally rated as moderately high throughout the study. Token reinforcement conditions resulted in a significantly greater number of math problems solved than in the no reinforcement conditions, suggesting that children holding little interest in an academic task may benefit from a significantly greater amount of practice attained as a result of token reinforcement procedures.