Matching-to-sample tasks have been a useful method in visual cognitive studies on non-human animals. The use of touch panels in matching-to-sample tasks has contributed to cognitive studies on terrestrial animals; however, there has been a difficulty in using these devices underwater, which is one of the factors that has slowed the progress of visual studies on underwater animals. Cetaceans (e.g., dolphins and whales) are highly adapted to underwater environments, and further studies on their cognitive abilities are needed to advance our understanding of the interactions between environmental factors and the evolution of cognitive abilities. In this study, we aimed to develop a new experimental method in which a captive killer whale performed a matching-to-sample task using a monitor shown through an underwater window as if a touch panel were used. In order to confirm the usefulness of this method, one simple experiment on mirror image discrimination was conducted, and the pairs with mirror images were shown to be more difficult to identify than the pairs with other normal images. The advantages of using this method include (1) simplicity in the devices and stimuli used in the experiments, (2) appropriate and rigorous experimental control, (3) the possibility of increasing the number of individuals to be tested and interspecies comparisons, and (4) contributions to animal welfare. The use of this method solves some of the problems in previous visual cognitive studies on cetaceans, and it suggests the further possibility of future comparative cognitive studies. It is also expected to contribute to animal welfare in terms of cognitive enrichment, and it could help with the proposal of new exhibition methods in zoos and aquariums.