When evaluating the truth of the modularity thesis about the brain, it’s important to distinguish between two forms modularity can take: a strong form and a weak form. The strong form is the view that the brain is organized along the lines of a swiss army knife, with hundreds or thousands of modules like the “mate selection module”, “food detection module”, or “cheater detection module”, with each module running a dedicated task. The weak form is simply the thesis that you can turn off or take out some parts of the brain without shutting down the whole system. For example, weak modularity is the idea that if you removed the auditory cortex your visual system would not completely crash and vice-versa.
The strong form is usually committed to things like “information encapsulation”. But there are two forms encapsulation might take: strong and weak. The stronger form says that any given module runs completely independently from other modules and when it is running its processes it uses its own internal knowledge to process it. This is supposed to be why the Müller-Lyer can’t be turned off even if you know it’s an illusion. The weak form views encapsulation a little different. On the weak view, each module is “talking” to a lot of other modules, and the idea is that when you have different modules talking to each other, new functions arise. The weak form thus sees modules built out of other modules, like a nested hierarchy. On this view, “encapsulation” has the wrong metaphorical connotations. Encapsulated seems to mean something like “isolated”. But on the weak interpretation, modules are not isolated at all; they are situated in a complex causal network of different modules. Moreover, the stronger form usually says that each module only really runs one process e.g. the cheater detection module only detects cheating. On the weak view however, it’s theoretically possible that a module could do more than one thing.
So when we look at task-based fMRI data using subtraction logic and are tempted to talk about a “theory of mind module” at one particular loci, we need to think about both the weak and strong forms of modularity and the weak and strong forms of information encapsulation. For the weak view of modularity, the theory of mind module is only modular because you could lesion it without shutting down the rest of the brain. And on the weak view of encapsulation, it’s more likely that theory of mind capacity stems from the powers of a distributed network of modules with the one particular loci that is “subtracted” out also being capable of helping out in other things beside theory of mind. The strong view of modularity and encapsulation would say the particular loci that is “most active” is the place where theory of mind happens. Michael Anderson has recently done meta-analyses of fMRI data and concluded that what’s going on often is cases where cortical areas are redeployed to perform new tasks, so the idea that any given brain loci does just one thing is mistaken. Since the brain constantly recruits old circuits to do new tasks, the strong form of encapsulation is going to be wrong: each loci can participate in different tasks in a slightly different way.