Avian brood parasites and their hosts provide model systems for investigating links between recognition, learning, and their fitness consequences. [1][2][3][4] One major evolutionary puzzle has continued to capture the attention of naturalists for centuries: why do hosts of brood parasites generally fail to recognize parasitic offspring after they have hatched from the egg, 5-9 even when the host and parasitic chicks differ to almost comic degrees? 7 One prominent theory to explain this pattern proposes that the costs of mistakenly learning to recognize the wrong offspring make recognition maladaptive. 10 Here we show that American coots, Fulica americana, can recognize and reject parasitic chicks in their brood by using learned cues, despite the fact that the hosts and the brood parasites are of the same species. A series of chick cross-fostering experiments confirm that coots use first-hatched chicks in a brood as referents to learn to recognize their own chicks and then discriminate against later-hatched parasitic chicks in the same brood. When experimentally provided with the wrong reference chicks, coots can be induced to discriminate against their own offspring, confirming that the learning errors proposed by theory can exist. 10 However, learning based on hatching order is reliable in naturally parasitized coot nests because host eggs hatch predictably ahead of parasite eggs. Conversely, a lack of reliable information may help to explain why the evolution of chick recognition is not more common in hosts of most interspecific brood parasites.The puzzling absence of chick recognition as a host defense against avian brood parasites has fuelled a long-standing and unresolved debate over what constrains the evolution of such a seemingly obvious host adaptation, producing many alternative hypotheses. 8, 9,[11][12][13][14] Learning is an important component of host defenses such as egg recognition, 2,15,16 , and theory suggests that it could be fundamental to the presence or absence of chick recognition in some taxa as well. 10,17 One prominent theory 10 shows that if chick recognition were learned, as it is with eggs, 2,15,16 then the evolution of parasitic chick recognition in hosts of common cuckoos, Cuculus canorus, could be constrained simply by the costs of errors in learning. Shortly after hatching, cuckoo chicks evict all host eggs and chicks, 7 and hosts are assumed to learn the features of their own eggs and chicks in their first breeding attempt. 2,15 According to the theory, a host parasitized during its first breeding attempt would falsely imprint on the lone parasite as its own chick, and thereafter reject its own offspring for the rest of its life. The extreme fitness costs of such misimprinting 10 could potentially explain why so many cuckoo hosts, and perhaps the hosts of other parasitic species, 17 do not recognize parasitic chicks. Because this misimprinting hypothesis was proposed to explain the absence of adaptation, it cannot be tested directly, but two corollary predictions are tes...