Abstract
The behavioral technology used to establish the relationship between what someone says about what they did has been termed correspondence training. Previous research (Cortez et al., 2013; Domeniconi et al., 2014) has not identified the specific training parameters responsible for the establishment of high levels of do-say correspondence. Moreover, the lasting effects of correspondence training have yet to be evaluated. Thus, the purpose of the present study was to evaluate the effects of three different types of correspondence training on children’s self-report accuracy over time. The behavior of doing consisted of reading aloud a written word presented on a computer screen, and the behavior of saying consisted of participants reporting on their performance accuracy after receiving automated feedback by the computer. Sessions consisted of 20 or 30 trials depending on the experimental condition. Each trial started with the simultaneous presentation of a visual stimulus (word, letter, syllabus, or figure) and the instruction, “What is this?” Immediately after participants’ response (doing), they were presented with the instruction “if you said [correct stimulus], click the green button, if not, click the red button.” Correct correspondent reports consisted of clicking the green button when their previous (doing) response was correct, and clicking on the red button when their response was incorrect. Following report accuracy evaluation sessions, in which the probability of errors was manipulated, participants were assigned to different correspondence training procedures. Four children were exposed to Training A in which points were delivered contingent upon correct correspondent reports in the absence of instructions (CRF). Three participants were assigned to Training B in which points were initially contingent upon correspondent reports on CRF, followed by intermittent reinforcement (VR2, VR3, VR6), with specific instructions describing the contingencies in effect. Finally, three participants were exposed to Training C, which replicated the previous condition in the absence of instructions. After training criterion was met, participants were re-exposed to report accuracy evaluation sessions. Follow-up sessions were then conducted after 30 and 60 days. All training methods were effective in maintaining accurate self-reports for seven participants, independently of the training parameters employed. Low levels of corresponding reports during initial sessions (report accuracy evaluation) seemed to predict low maintenance over time. In contrast, intermediate or high percentages of correspondence during the initial sessions of evaluation would be predictive of higher maintenance levels over time.