This paper considers methodological problems in animal metacognition studies. After summarizing a variety of experimental designs to study animal metacognition, we review recent literature that were not included in Fujita (2010)'s more comprehensive review. We then examined Carruthers (2008; Carruthers and Rithie 2012)' skeptic argument against animal metacognition. He argued that no experiments thus far successfully exclude the possibility that animals, as well as human infants, elicit "metacognitive" responses without explicitly and consciously monitoring internal representations. We argued that newly developed experimental paradigms to study explicit memory processes could be a key to reject Caruuthers' skeptic argument and thus facilitate understanding how we access to the internal representations.