Jerry Fodor has a lively and thoughtful review of Andy Clark's new book Supersizing the Mind in the latest issue of the London Review of Books. The paper is in effect a critique of the extended mind thesis, targeting Andy's and my joint paper "The Extended Mind", Andy's book, and my foreword to the book. Fodor makes two or three interesting objections to the extended mind thesis.
Fodor's first objection is that we haven't given any clear content to the claim that extended objects (such as iPhones and notepads) can be parts of the mind. Here it's worth noting that we never put things in exactly this way in the original paper: at a couple of points we talk of external objects being parts of cognitive processes, but not parts of minds. I do talk about parts of the mind in the foreword, but this is mainly to quickly give a sense of the thesis for general readers. In any case, I think that talk about "parts" here (and of "minds", for that matter) is quite dispensable.
I think it is best to formulate the extended mind thesis in terms of the role of the environment in determining a subject's mental states. The key claim in the original paper (at the start of section 4) is that subjects' beliefs can be partly constituted by features of the environment, when those features play the right sort of role in driving cognitive processes. A bit more generally and precisely, the claim is that sometimes a subject has a mental state (e.g., believing that the Museum is on 53rd Street) partly in virtue of external processes, and in particular in virtue of these external processes playing the right sort of role in driving cognitive processes. Here, "in virtue of" should be understood constitutively rather than causally, and "the right sort of role" needs to be understood as a certain sort of active synchronic role, as opposed to the sort of diachronic role that is typically relevant in Twin Earth cases and the like.
However one fleshes out the thesis, the rubber hits the road in theses concerning supervenience. The key claim here is that there can be physical twins who are in different mental states, because of differences in the way that their synchronic environments are coupled to their internal cognitive systems. The Otto case is supposed to yield a case like this: Otto and Twin Otto believe different things in virtue of differences in their notebooks (and in particular in virtue of these differences being coupled in the right way to their brains). If there can be cases like Otto and Twin Otto, it follows that the supervenience base for mental states must be extended to include these environmental elements, for reasons quite different from those that are relevant in other forms of externalism. So in practice, the claim that coupled external states are part of the supervenience base for mental states is where the action is at.
Fodor's second objection is that the intensionality of external items (such as entries in a notebook or an iPhone) is always derivative: it always depends on the intentionality of internal thoughts. I'm happy to concede this point, but I think that something similar is true of many (perhaps all) dispositional beliefs. Typically, a dispositional belief that P is a belief that P partly in virtue of a subject's dispositions to have occurrent judgments that P. Correspondingly, the intentionality (and intensionality) of the dispositional belief depends on the intentionality of occurrent judgments. Likewise, the intentionality of Otto's dispositional belief about the Museum depends partly on his notebook-induced disposition to have occurrent judgments about the Museum. So the Otto and Inga case seem to be on a par in this respect. To deny Otto's dispositional belief on these grounds, it looks like one will have to deny Inga's dispositional belief too.
In discussing this objection, Fodor makes a third objection that is related but raises some distinct issues. Here he suggests that there is a disanalogy between the Otto and Inga case, in that Otto needs to remember to look in his notebook, whereas Inga doesn't need to remember to consult her memory -- indeed, she doesn't have to actively "consult" her memory at all. This suggests that there is really an extra step in the Otto case, and the need for tha step that suggests that Otto didn't really believe that the Museum was on 53rd Street before the process began.
In response: first, I'm not sure that this gets the Otto case right. The case we have in mind (like many actual cases, I think) is one in which Otto's consulting his notebook is an automatic process that needn't be preceded by occurrent thoughts or rememberings concerning the notebook. Second, there are certainly cases in which ordinary subjects don't retrieve information automatically, but instead need to deliberately attempt to remember. It doesn't seem that we deny that the subject has relevant dispositional beliefs just because retrieval requires conscious effort in this way.
Still, I think there is something to Fodor's objection here, which is closely related to something I say in the foreword (perhaps not quite as clearly as I could have). The most important objection to the extended mind thesis concerns the role of perception and (especially) action in cases of extended retrieval: Otto has to perceive and act to retrieve the relevant information, whereas Inga does not. Correspondingly, the most consistent way to deny that Otto has the relevant belief is to deny dispositional beliefs in any case in which the subject has to perceive or act in order to retrieve the relevant information -- and here we must include inner perception and mental action, in order to avoid begging any questions. So on this view, cases in which conscious effort is required to remember that P won't be cases in which the subject antecedently believes that P. The same goes for cases in which calculation is required, and cases in which external action is required. On this view, Otto won't have a dispositional belief about the Museum, and neither will a version of Inga who has to make a conscious effort to recall the information, but the version of Inga who retrieves the information without acting is OK.
Of course there is a lot more to be said about this view. For a start, the view (like the extended mind view) seems somewhat revisionary in going against our intuitive classifications. Further, a proponent of the extended mind thesis can reasonably hold that all of these cases should be counted as dispositional beliefs, at least for many explanatory purposes. A proponent might even suggest that as Otto's retrieval is effortless and automatic, it is more apt to count as a dispositional belief than cases requiring mental effort. Still, I think the view is at least coherent, and I think that it is probably the best way to resist the extended mind view. It's certainly a much less costly view than the increasingly common form of opposition (put forward by Brie Gertler and others) that denies that there are any dispositional beliefs, or that denies that dispositional beliefs are mental.
In any case: I take the upshot of Fodor's objection to be that if one denies a dispositional belief to Otto on the grounds that he has to consult his notebook, then we should also deny a dispositional belief to a version of Inga who has to wilfully consult her memory. And in the other direction, if one allows dispositional beliefs in the latter case, then there is no longer any objection here to allowing dispositional beliefs in extended cases such as Otto's. I think that at least for many purposes, it makes sense to allow dispositional beliefs even in cases involving mental action. If so, the case for the extended mind remains strong.