I thank Daniel Hausman for the generous discussion of my work. Since the journal permits a brief reply, I offer a few comments on Hausman’s discussion of my position in his paper. It is great news for philosophy of economics that there is now such a lively debate about the interpretation of preference, and I believe readers will be in a good position to adjudicate for themselves by reading the literature discussed in this paper. I would just briefly like to comment on one aspect of Hausman’s discussion. Hausman’s response to criticisms by me and others concedes that the constraints of decision theory are idealizations, but stresses that these idealizations may be entirely justified for a particular explanatory or predictive task. I agree, of course, and more particularly agree that this often allows us to ascribe to agents preferences with relatively coarse-grained objects: it may be perfectly fine to, in this way, assume away dependence on context irrelevant to our explanatory and predictive purposes. However, a way of rephrasing my argument is to say that there is simply no guarantee that the level of grain at which we can ascribe consistent and stable preferences that reliably track an agent’s observed choice behaviour within our context of interest is also a level of grain at which it makes sense or has explanatory value to ascribe mentalistic preferences to her. And if that is so, then insisting on a mentalistic interpretation of decision-theoretic preferences creates a potential tension between achieving fit with both an agent’s explanatorily relevant mental states and her observed behaviours with the same model. Whether we care most about prediction or about (folk-psychological) explanation, the less restrictive way forward either way is the approach argued for in my papers: that we should ascribe preferences and choose a level of grain of the objects of preference with a view only to capturing choice behaviour in our context of interest. This allows us to reveal patterns in agents’ choices, e.g. which aspects of a consumption bundle make an agent more likely to choose it. We can then further ask what mental states best explain these patterns. The mentalistic interpretation of preference prejudices rather than helps this inquiry: Most of the time, I contend, the best answers will involve mental states with objects coarser than the decision-theoretic preferences we ascribed to them (e.g. not entire consumption bundles but rather specific aspects of them) and partial rather than total evaluations.