Horan is aware that robustness is a valuable feature of any scientific account, but proposes that the inductive process of building up a functional account from one species to others through the comparative process will secure realism as well as robustness. But, for reasons I outlined above, I see no reason to think that the inductive approach is more likely to identify true functions. Ultimately, both methods are fallible. But perhaps each can compensate for the liabilities of the other. Each provides a way of further testing proposals developed from the other method. I have already indicated how results of an optimality analyses can tested by use of the same kind of reasoning as Horan outlines in her account of the inductive method of the comparative analysis. But the model building in optimality analyses also provides an important test on functional analyses developed from a comparative approach. The importance of actually developing a formal model is that all assumptions of the functional account must be made explicit in the model, and when we calculate predictions from the model, we are able to determine precisely what the implications of the analysis are. The predictions may turn out different from those we might make without developing a detailed formal model. My suggestion is that the two approaches, optimality analyses and comparative models, may actually complement each other and arguing for one rather than the other may represent a false dichotomy.