Abstract
Procedural fidelity is the extent to which procedures are implemented as prescribed. Research shows that fidelity errors readily occur in applied settings and that they can delay and/or impede learning. This study consisted of three human-operant experiments to evaluate how varying levels of fidelity errors affect performance on an arbitrary matching-to-sample task (MTS). Experiments 1 and 2 investigated how implementing the MTS task with varying levels of fidelity affects learning and how initially exposing learners to varying levels of fidelity affects ongoing performance once the errors were corrected. The purpose of Experiment 3 was to approximate the minimum level with which fidelity can vary before individuals can no longer master the task. We found that the varying fidelity levels can delay learning for some learners but not all. However, all participants mastered the task once the errors were corrected. Moreover, the results suggest that the MTS procedure remains efficacious when administered within a relatively narrower range of varying fidelity.