It is common to speak of brain regions for particular cognitive functions: regions for reading, for seeing faces, or for doing math. This suggests that brain regions have a single function which they always and uniquely perform. Advances in neuroscience show that this simple picture cannot be correct. Each brain region participates in a wide variety of cognitive functions, and cognitive functions can be performed by distinct brain regions at different times. Drawing an analogy to similar problems in philosophy of biology, I suggest that we ought to read 'region for' as designating brain areas which make a difference to the performance of personal-level activities. While this is a permissive reading of 'region for,' it also allows us to distinguish brain regions that make sensitive and orderly differences to activities from those which have only weaker sorts of difference-making relations. I argue that a difference-maker account lets neuroimaging indirectly constrain cognitive theories, by providing evidence about the activities that cognitive theories are meant to explain. This kind of indirect test is continuous with current methodologies in cognitive science, and does not rely on contentious claims about the relationship between the cognitive and the neural.