The comparative analysis of metacognition may answer fundamental questions about the evolution of cognition. Although a substantial amount of research has been directed toward this goal in the last two decades, the recent development of quantitative nonmetacognition models has raised questions about the existence of metacognition in nonhumans. Kornell (2014, pp. 143-149) proposes that advances in animal metacognition may be made by following emerging trends in human metacognition research, namely that animal metacognition may take the form of drawing inferences from metacognitive cues without directly assessing the strength of memories. A problem with this approach is noted. Because the metacognitive status of certainty judgments in animals is at the center of the dispute in the field, demonstrations of the inferential view would not provide evidence that putative metacognitive cues are indeed based on metacognition. I argue that any preparation that claims to tap into metacognition needs to be tested against leading nonmetacognition hypotheses such as Le Pelley's (2012) reinforcement-learning model. Progress in animal metacognition will come from the development of new assessment techniques that offer predictions contrary to nonmetacognition hypotheses. Animal metacognition will advance by applying skepticism about methods and interpretation while letting the animals (and their data) settle the debate.