In the present study, we used feature lists in a categorization task and required participants to report their subjective confidence to examine the dissociation between explicit and implicit knowledge. Whereas our findings replicated the effect of prior knowledge on learning, our results challenge the role of explicit and implicit knowledge suggested by previous research using a similar paradigm. We argue that measures that consider all confidence responses, referred to as calibration indices, are required to understand the contributions of implicit and explicit knowledge in any task.