In 2016 Deborah Gordon proposed that the term ‘division of labor’ was misleading and ought to be abandoned. The problem with the term, as she saw it, was that it implied a division of labor among specialized castes of workers to explain colony behavior. In a series of experiments in the 1980s, Gordon discovered that observed colony behavior could not be caused by such a division. Gordon proposed ‘task allocation’ as an alternative description that allowed for a broader range of possible explanations, including inter-individual interaction in dynamic networks. These interactions can happen on short time scales, and Robert Jeanne took Gordon’s ‘task allocation’ to mean just those brief interactions. Jeanne has fiercely rejected Gordon’s proposal on the mistaken grounds that she has called for abandoning developmental explanations, and that her account of division of labor distorts the productive history of the term as researchers have used it. In fact, the term has been used to mean a great many things since Oster and Wilson associated it with optimality modeling and sociobiology in 1979. The reasoning in this controversy can be clarified with a framework Elisabeth Lloyd has called “The Logic of Research Questions.” This framework involves identifying what answers are possible and responsive to a given research question, and can be used to distinguish what restrictions the terms ‘division of labor’ and ‘task allocation’ imply. With this framework I will disentangle historical uses of the term, show how Gordon’s proposal has continued to be mischaracterized.
