Inspired by the work of Ivan Pavlov, and seeking a scientific approach to human psychopathology, from the 1930s, American psychiatrists turned to the animal laboratory. Experimental neurosis could be reliably generated in a variety of animals – most commonly dogs, but also sheep, goats, pigs, and cats – by changing the conditioning routine and creating irresolvable conflict between stimuli. From the 1960s, however, the study of experimental neurosis was criticised for being too artificial in method, lacking in interpersonal and social context, and the disorders created too readily mapped onto the human. Focusing on the work of W. Horsley Gantt at Johns Hopkins, this paper will revisit the use of the conditional reflex method in the study of experimental neurosis. It will show how Gantt was, firstly, less concerned with modelling a specific syndrome, than a functional relationship between emotional disorder and its physiological bases; and secondly, was acutely aware of social factors such as the relationship between dog and human experimenter. Gantt observed individual dogs throughout their lives and took account of their personalities and social relationships. We will explore the influence of his case-based approach in behaviour therapy and clinical medicine, and examine the problems it generated for the growing field of experimental psychopathology.