How do scientists confirm dynamic compositional hypotheses, such as that rats navigate mazes using cognitive maps or that humans parse sentences using a grammar? I propose that they use a type of hypothetical reasoning, a type of abduction. Schematically, the idea is that if H were true, then H would explain E, so we may infer that H. This paper sets out a theory of this reasoning based on an example drawn from the experimental work of Alan Hodgkin and Andrew Huxley. It also presents some examples where there is prima facie reason to think that scientists are using this reasoning.