Brie Gertler

Brie Gertler (University of Virginia), “A Fregean Argument against Externalism,” with commentary by Sanford Goldberg (University of Kentucky).  The paper and commentary may both be found here.

Comments

[Benj]

Hi Brie, interesting paper.

You argue in support of the switching victim's being unable to reject that water = twater on the grounds that "the concept twater plays the same narrow conceptual role, for Toscar, that water plays for Oscar". If the narrow conceptual roles are *exactly the same*, I wonder how this is compatible with the victim's possessing both concepts simultaneously.

[Justin Fisher]

I had a similar concern to Benj's. Here's an example:

I'm pretty sure that, a couple of times as an undergrad, some professor told me about some chemical with properties PQR. I don't anymore recall anything that helps to distinguish between the one occasion of a professor saying such things and the other. I don't know whether they were talking about the same chemical or just similar chemicals. I can't remember the names of the chemicals, and I don't know any way to find out.

Question: Is it possible that I (rationally) now have two concepts - call them C1 and C2 - such that C1 is the concept I formed in response to the first professor, C2 is the concept I formed in response to the second professor, and yet the internal roles of these concepts are exactly symmetrical? (NB: I'm thinking of concepts as mental particulars, which fits Brie's usage, but not Sanford's.)

Suppose you answer yes: Then, it seems that I could rationally affirm "C1=C1" (and "C2=C2") while at the same time doubting "C1=C2". And it could be that the first professor was talking about H2O (so, according to the externalist, C1 refers to H2O), while the second was talking about XYZ (so C2 refers to XYZ). Hence, this case would be a case where the content-difference posited by the externalist matches up to a difference in dubitability, despite exact symmetry in internal roles; and hence the externalist *can* pass your Frege-test afterall.

If you answer no: Then the reason that the externalist can't find a case that passes your Frege test apparently is that you can't simultaneously have two distinct concepts that are exactly symmetric in their internal roles. But this suggests that your Frege test is a bad test, because you can't even set up the sort of case that you'd need to look at in order to test the externalist.

[Brie Gertler]

Benj and Justin: Thanks for your comments on the paper. I’m sorry for the long delay in responding.

Benj said “If the narrow conceptual roles are *exactly the same*, I wonder how this is compatible with the victim's possessing both concepts simultaneously.”

I agree that it’s difficult to see how one could simultaneously possess two distinct concepts with precisely similar narrow conceptual roles. (In fact, I don’t think that one could do this.) But notice that I am allowing this as a concession to the externalist, and only for the sake of argument. It is a commitment of externalism (at least, of the brand of externalism that my paper targets) that two narrow duplicates could differ in their concepts. For if concepts were individuated by narrow features (whether this is conceptual role or something else), then they would be narrow as well.

The dialectic is this. My externalist target claims that sometime after I am moved unknowingly to Twin Earth, my water concept will be either replaced or supplemented by a twater concept. Now if the new concept simply replaces the old one, then we run up agains the ‘memory’ objection: I cannot remember what I was thinking about last week (when the water concept was still in place). For this reason, most externalists—including Burge—allow that the old concept remains in place, and is somehow supplemented by the new concept. This means that the slow switching victim does have two concepts that differ only in their wide features. My argument then aims to show this: two thoughts which differ only in that one employs the water concept and one the twater concept are cognitively equivalent.

Now depending on how one understands concepts, this result may mean that these aren’t different concepts after all – and hence, perhaps, that one cannot simultaneously possess two different concepts with equivalent conceptual roles. I’d be happy to embrace this conclusion. But notice that, given that the concepts share narrow features but differ in wide features, this weighs against externalism. For the best way to explain why one cannot simultaneously possess two distinct concepts that are internally alike is that concepts are individuated narrowly.

Justin has a similar worry: he says that if two distinct concepts can share an internal role, then they will pass my Frege test; if they cannot, then the Frege test is not a good test for cognitive equivalence. I think that the externalist is committed to the claim that two distinct concepts can share an internal role. For as I explain (in Section 3, addressing Objection 2), to deny this is to give internal features precisely the sort of individuating role that an externalist should repudiate.

But now why does Justin think this means they will pass the Frege test? He says: “Suppose you answer yes: Then, it seems that I could rationally affirm "C1=C1" (and "C2=C2") while at the same time doubting "C1=C2".”

I don’t see the argument here. Perhaps the reasoning is this: if C1 and C2 really are distinct concepts, then parallel thoughts involving them must be differentially dubitable; hence, these thoughts would pass the Frege test. I am allowing that they are distinct concepts, for the sake of argument—but in allowing this, I am of course not allowing that parallel thoughts involving them are differentially dubitable. For that would put me in the untenable position Justin describes. Another way to construe the dialectic is this: the externalist claims, or should claim, that sameness of narrow features isn’t sufficient for sameness of concepts. Allow that this is correct. Now ask: are two thoughts that are alike in their narrow features differentially dubitable? If not, then even if there is some notion of wide content that renders them distinct, this notion isn’t the Fregean notion of content that Burge is after.

[Justin Fisher]

Brie, thanks for the response. I'd like to follow up with respect to the part of my argument that you had trouble seeing, but first, I think it might help to get clear on how we are using our 'concept'-talk. I'll describe how I use this talk (which is basically the same way as Millikan 2000), and then articulate my concern in those terms. I hope I won't beg any questions.

I think of a concept as being a mental particular, the psychological equivalent of a manilla file-folder that one can use to gather useful information about an interesting target in the world. On this usage, my twin and I have numerically distinct concepts, just as we have numerically distinct hearts.

The debate between the internalist and the externalist is a debate about whether my concepts and my twin's corresponding (but numerically distinct) concepts have to have the same *intentional*contents*, or whether I might have a concept that has one content (say H2O), even while my Twin's corresponding concept has a different content (say XYZ).

Many externalists, including the opponent you have in mind, hold that, by unknowingly spending enough time on Twin Earth, I could come to have a concept with content XYZ, even though I previously lacked such a concept. You mention two ways in which one might suppose this could happen. Either, my earlier concept that once meant H2O is subtly transformed into a concept that now means XYZ, or my earlier concept that meant H2O is supplemented by a new concept that means XYZ.

My preference is for the former 'tranformative' view. An intuitive picture is this: I grow up on earth using a particular manila file folder - call it WATER - to store information about H2O. Then I'm secretly transported to twin earth. I continue to use folder WATER, but now to gather information in my encounters with XYZ, and to bring this information to bear in further encounters with XYZ. Given enough time at this, it seems reasonable to say that, now, my folder WATER is being used to track XYZ, and not, anymore, to track H2O. Unbeknownst to me, my folder WATER has transformed from being a tool for tracking H2O to being a tool for tracking XYZ. And on my preferred externalist view, that means this concept has shifted from meaning H2O to meaning XYZ. After this transformation, I may still be able to form thoughts of H2O, but only in a much more roundabout way than I could before: e.g., now I have to think, 'what my concept WATER meant when I was a child'. I'm inclined to bite the bullet regarding the 'memory objection' -- our tools for thinking about the world can gradually change in meaning, and that's a concern we have to live with as we try to employ memories of the distant past.

You consider an opponent (e.g., Burge) who takes the other route, and says that, during my time on Twin Earth, my old H2O-concept is supplemented by a new XYZ-concept. This route fits very poorly with the metaphor of concepts as manila file-folders -- if remaining on earth wouldn't have led me to create a new file-folder, how could being secretly transported to an indistinguishable place lead me to do this? So I think there are good reasons for the externalist not to take this route. But you're willing, for the sake of argument to lay these concerns aside, and instead rely on your Frege-test to raise problems for the externalist.

Now getting back to my earlier posting, regardless of what the externalist says about slow-switching cases, it does seem plausible that the externalist might allow that a single person could have two distinct file-folders C1 and C2 that are exactly alike in their internal roles. As a plausible example, I proposed two of my concepts, each of a chemical discussed in a long-ago and mostly-forgotten chemistry course. These seem plausibly to be two simultaneously-possessed concepts that are now exactly alike in their internal roles, but which could turn out to differ in their wide content -- exactly the sort of case you'd need to apply your Frege-test.

You seem willing to grant that a situation like this could occur, but you were hopeful that this won't cause problems for you. I argued that it would cause problems, because this, by itself would be enough to allow the externalist to pass your Frege-test.

I took it that, to pass your Frege-test, for any simultaneously-possessed concepts C1 and C2 that are exactly alike in their internal roles but differ in their wide content, the externalist would need to point out a predicate such that one could rationally affirm that predicate's applicability to C1, while doubting its applicability to C2. I proposed the predicate "being identical to C1". Whatever concepts C1 and C2 turn out to refer to, it's clear that the thought "C1=C1" will be true; but it's doubtful that the thought "C2=C1" will be true. So there's a difference in dubitability between a predication involving C1 and a parallel predication involving C2, which is enough to pass your Frege test.

As far as I can tell, this concern will arise whenever you admit that C1 and C2 are distinct concepts which, from the agent's perspective, might turn out to refer to distinct targets. It doesn't really matter whether the externalist takes the 'transformative' route, or the 'supplementative' route regarding slow-switching cases. Insofar as you can set up the test-case at all, the externalist should be able to pass it.

(Incidentally, I still think you might want to go to the other horn of the dilemma posed in the earlier post, and insist that, so long as C1 and C2 are distinct concepts, then they are not relevantly alike in their internal roles. E.g., I can trivially infer C1(x) from C1(x), but I can't infer C2(x) from C1(x). But I still think that, if you take this horn, this just means that your Frege-test is *never* applicable, and hence poses no problems for the externalist.)

[Brie Gertler]

Thanks for your clarificatory comments, Justin. I’m first going to register a bit of speculation about your version of externalism, and then reply to your objection to my use of the Frege test.

(i) Your externalism

It may be that the version of externalism you have in mind is less radical than the version I target in the paper. For instance, you say: “Given enough time at this, it seems reasonable to say that, now, my folder WATER is being used to track XYZ, and not, anymore, to track H2O. Unbeknownst to me, my folder WATER has transformed from being a tool for tracking H2O to being a tool for tracking XYZ.”

Your picture may be underspecified here, but as it stands, I think it’s consistent with internalism. Suppose that my concept is a narrow description, which is rigidly indexed to my environment: e.g., ‘the waterish stuff around here’. Then, the move to Twin Earth could change its referent from H2O to XYZ without my knowledge; and, in that sense, my concept could ‘track’ a different substance. By contrast, Burge claims that environmental features partially determine the subject’s ‘epistemic perspective’: “how things seem to him, or in an informal sense, how they are represented to him”. And it’s quite possible that there is variation in what one tracks, while how things seem to one remains constant.

Another indication that your externalism may be less radical than Burge’s is this: “if remaining on earth wouldn't have led me to create a new file-folder, how could being secretly transported to an indistinguishable place lead me to do this?” Of course, I agree with this doubt. But it seems to me that the doubt stems from an intuition that is internalist in spirit, viz., that when it comes to rational relations among (and changes in) thoughts, a purely environmental change is irrelevant.

In any case, a more moderate version of externalism isn’t threatened by the Frege test—not because it would allow for differential dubitability, but because it needn’t require that two thoughts that differ in content be differentially dubitable. Burge is vulnerable to the Frege test because he makes the more radical claim that we must construe thoughts widely in order to capture the ‘epistemic perspective’.

(ii) The Frege test

Regarding your dilemma about the Frege test: I do want to claim, for the sake of argument, that one can simultaneously have two concepts that are precisely alike in their narrow features. This claim functions as a concession to the externalist; I think it is in fact false.

Onto your objection. Your characterization here is correct: “I took it that, to pass your Frege-test, for any simultaneously-possessed concepts C1 and C2 that are exactly alike in their internal roles but differ in their wide content, the externalist would need to point out a predicate such that one could rationally affirm that predicate's applicability to C1, while doubting its applicability to C2.” (Well, this is sufficient, though perhaps not necessary—there may be other ways of establishing differential dubitability. But it’s the sufficiency that’s at issue here.)

Still, I don’t think that your case works. It is this:
“I proposed the predicate "being identical to C1". Whatever concepts C1 and C2 turn out to refer to, it's clear that the thought "C1=C1" will be true; but it's doubtful that the thought "C2=C1" will be true.”

First, note that the terms used to denote C1 and C2 will not differ. For instance, when moved unknowingly to Twin Earth, one does not begin saying ‘twater’. So the corresponding thoughts will be expressed by “water=water” and “water=water”. And this feature is not specific to the water/twater case: externalists don’t deny that differences in terms are internal.

Might there be another difference that could deliver the result you describe? Returning to the ‘water’ case, one option would be this. If I entertain the idea that I was switched last week, I may wonder whether ‘water=water’, where the term ‘water’ on the left side of the identity sign employs an old concept of mine, and the term on the right employs a new concept of mine. (The Frege test would then be run by comparing this with the thought ‘water=water’, where these both employ the old concept, OR both employ the new one.) However, this reflection smuggles in a descriptive difference—the difference between being an old concept (or a concept I exercised last week) and a new one. And, just as in the case of terminology, that’s the sort of difference that even externalists treat as an internal difference. So this reflection would mean that these thoughts were not internally similar, and hence could not serve as test cases for externalism. And I expect that any such maneuver would yield a similar result: it would secure differential dubitability only by smuggling in the sort of descriptive difference that is ordinarily taken (even by externalists) to be a narrow difference.

As before, thanks very much for your thoughts on these issues.

[Benj Hellie]

Hi Brie, thanks for the response.

My concern is that two distinct concepts possessed by the same individual simultaneously could not share a narrow functional role. Externalists don't commit to that, rather to the view that two distinct concepts possessed by distinct individuals, or by an individual at distinct times, could share a narrow functional role.

[Brie Gertler]

Hi Benj,

I think that the externalist – at least, the externalist who is inclined towards functionalism – should allow this. For (i) it would be ad hoc, and contrary to the spirit of externalism, to deny it; and (ii) it would commit the externalist to an implausible position regarding memory.

(i)
According to the (functionalist) externalist, the difference between a ‘water’ thought and a ‘twater’ thought is purely external. So, as you say, these could be instantiated by two internal duplicates – or by a single person at two different times, whose internal state remains constant. But then why couldn’t a single person, at a single time, instantiate them both? To deny this would be, it seems, to assume that the two thoughts were somehow in competition. But this would accord the kind of significance to purely internal features that the externalist denies. More to the point, it would deny external features the kind of significance that the externalist claims.

(ii)
If the subject unknowingly transported to Twin Earth could (eventually) no longer have ‘water’ thoughts, then she could not remember what she was thinking last week. For this reason, Burge maintains that one can simultaneously have the concepts ‘water’ and ‘twater’. Some externalists do deny this, of course – but then they face the memory objection.

Justin, in an earlier comment, said that he thinks the ‘water’ concept will be replaced, not supplemented. But Burge is my externalist target. So even if the memory objection can be somehow finessed—which I doubt—the basic objection still remains against Burge.

[Brie Gertler]

Reply to Goldberg

First, I want to thank Sandy for his detailed comments on the paper. He has offered some forceful objections, and raised at least one point where I commit myself to more than I should. I’m very grateful for his careful attention to my arguments.

I confess that I find it very difficult to feel the force of externalist arguments. In fact, the desire to achieve a clearer understanding of externalist intuitions partly motivated me to write the paper. But I’m not at all confident that I have achieved this. It seems to me that the conflict between internalism and externalism is (even) deeper, and less tractable, than most other philosophical disputes. My replies to Sandy’s criticisms may reflect this fact.

I reply, in turn, to each of his five objections.

Reply to (1)
My intention, in distinguishing Fregeanism about thoughts from Fregeanism about propositions, was simply to define my target in the paper. I did commit myself more than I needed to, for this purpose, in saying that my view was consistent with rejecting a particular form of Fregeanism about propositions. But the intention was to express my neutrality about that latter position (which Goldberg helpfully labels ‘P-Fregeanism’).

Now as Goldberg notes, accepting Fregeanism about thoughts (‘T-Fregeanism’) is consistent with denying P-Fregeanism. For instance, P-Fregeanism holds that propositions should be individuated by cognitive value. And one may simply deny this, taking ‘Superman is formidable’ and ‘Clark Kent is formidable’ to express the same proposition, independent of one’s position on how the corresponding thoughts are to be individuated.

Goldberg worries that anti-Fregeanism about propositions, of the sort I describe, conflicts with the internalism that I defend. He says “Those who deny P-Fregeanism – those who hold a view we (following Gertler p. 5) might call ‘anti-Fregeanism’ about propositions – typically do so on the basis of ‘externalist’ considerations.” That may be so. And certainly, those who deny P-Fregeanism on Russellian grounds are a good example of this. But I don’t see that this fact threatens my argument. For one may reject P-Fregeanism on other grounds: for instance, one may individuate propositions in a way that is unrelated to their cognitive value.

In describing the rejection of P-Fregeanism, I refer to a view outlined by Salmon, in which a thought is determined by the mode of presentation of a proposition. This reference suggests a particular alternative to P-Fregeanism, one that—according to Goldberg—conflicts with my internalism about thoughts contents. Now I don’t quite see how this conflict arises. Goldberg says “But presumably these modes of presentation would not be the modes of presentation they are, if the object presented were other than Superman/Clark Kent.” But this would follow only if modes of presentation were partly individuated by the objects presented. Now it may well be that that is a feature of the specific view Salmon describes—and I should surely drop the reference to Salmon, to avoid this confusion. But it doesn’t seem to be a necessary feature of any ‘mode of presentation’ view. For it seems coherent to claim that someone other than Clark Kent (and, hence, other than Superman) might share the ‘Caped Crusader’ mode of presentation.

The upshot is this: if there is a conflict of the sort Goldberg describes, it arises from specific features of a view that I mention only for illustration. It is not part of every reason to reject P-Fregeanism; nor is it, I think, even part of every ‘mode of presentation’ approach to distinguishing thoughts from propositions.

Reply to (2)
Goldberg raises a concern about simultaneously employing two homonyms, or two internally similar concepts. (This concern was later raised by Benj Hellie and Justin Fischer, in the comments.) I share this concern: for it amounts to the worry that externalism has implausible commitments. If it’s not possible to simultaneously have, or exercise, two internally similar concepts, the best way to explain that fact is to adopt internalism: if concepts are individuated by their internal features, then one simply couldn’t have two concepts with precisely similar internal features. In the paper, I allow that one could do this, as a concession to the externalist. And it seems that the externalist should allow this; it is contrary to the spirit of externalism to claim that internally similar concepts are somehow in competition with each other.

The real issue raised by this objection concerns whether, in a move to Twin Earth, the new use of “water”—i.e., the twater use—takes over, and eradicates the previous, water use. If so, then the Frege test simply can’t be run in the way my argument requires. But then the externalist faces two problems. First, he must explain why a move to Twin Earth prevents the subject from remembering her previous (Earthian) thoughts. (This worry is pressed by Boghossian.) Second, we seem to need some special story about why this happens with water and twater, and not with ordinary homonyms, like (river) bank and (financial) bank.

The worry, that use of one concept will somehow cancel out or render irrelevant another concept, suggests that these are (at least) more closely related than non-twin concepts. But this puts pressure on the externalist claim that they are distinct: a clear and obvious way to explain why, in discussing thirst-quenching liquids, the thinker doesn’t have access to (or can’t simultaneously employ) 2 distinct concepts water and twater, is that she simply doesn’t have 2 distinct concepts; her term “water” expresses only one.

Reply to (3)
Goldberg argues that my central thought experiment may not yield the result I envision. I report that, when imagining that I have been slowly switched, and that my ‘water’ and ‘twater’ concepts are internally identical, I find myself unable to believe that water quenches thirst while doubting that twater quenches thirst (or vice versa). Note that this is my own reaction to the thought experiment—I am not simply assuming that this is the only reasonable reaction, but rather am using the only evidence about the case that I have, viz., my own response to the envisaged scenario. So I am not begging the question as Goldberg implies here: “it begs the question against the externalist to suppose that any switching case of the sort Gertler is describing will be one in which adopting different attitudes to what in fact are distinct ‘twin’ thoughts will be imagined to be ‘equivalent to simultaneously endorsing and denying a single thought.’”

Here is Goldberg:
“Consider the following possibility (or what I will argue is a possibility): one works oneself into a position in which one suspects, of two of one’s applications of what appears (introspectively) to be the same concept, that two distinct but subjectively-indistinguishable concepts are nevertheless in play.”

To consider his objection, I will modify this case, so we can use the familiar ‘water’ / ‘twater’ terminology. Suppose that I think to myself, “I might have been switched to Twin Earth.” I then wonder: “does my current use of ‘water’ express a different concept from my use of that term last week?”

There are two possibilities here. The first is that, in wondering this, I merely mention the term ‘water’, but do not use it; in that case, this wondering does not yield any result about my water concept. The second is that this is truly a use of the term, which does allow for genuine differential dubitability, of the sort relevant to the Frege test. But in that case, my concept has changed: I now use ‘water1’ to express “the waterish stuff in my current environment”, and I now use ‘water2’ to express “the waterish stuff in last week’s environment”. The difference in these descriptions is an internal difference; even externalists believe that such descriptive concepts differ internally. So these two ‘water’ concepts cannot be used to run the Frege test on externalism.

Reply to (4)
Goldberg argues that an internal difference between two concepts may not suffice to individuate them. He imagines a subject named Schmidt, who thinks to herself:

“When I was just now thinking: water quenches thirst, I was thinking of the watery stuff around here; years ago when I was thinking: water quenches thirst, I was thinking of the watery stuff around wherever I was then. In both cases (and even now, in recollection), the way I was/am thinking of the stuff in question was/is via the concept(s) expressed by my use of ‘water’ on the relevant occasions. But I now think that that these uses express the same concept only if the watery stuff around here is the same kind of stuff as the watery stuff around there. And I wonder whether that’s true…”

Goldberg continues: “It seems to me that this describes the sort of case I need: it is a case in which there is an ‘internal’ difference between Schmidt’s two ‘twin’ thoughts, but – because this ‘internal’ difference is an ‘indexical’ difference – this internal difference does not suffice, by itself, to individuate the two concepts in question.”

The internal difference does suffice, according to the internalist, to individuate the concepts in question. The first concept is ‘the waterish stuff in my current environment’; the second is ‘the waterish stuff in my past environment’. Now of course this isn’t enough, on its own, to determine the referent of those concepts: we have to look at the corresponding environments to do that. But if I believe that the waterish stuff here ≠ the waterish stuff there, my doubt is due to the internal difference between these concepts. (That is what I called ‘the same intuition that drives my original argument’.)

Reply to (5)
As with the previous objection, there seems to be a confusion here between thought content and referent. Suppose that, as Goldberg says, the deference to community standards about ‘sofa’ is internal. Then, the actual community standards will determine the referent of ‘sofa’, but not the concept of ‘sofa’. Now as for ‘application conditions’: community standards determine which things the term ‘sofa’ applies to, so they determine application conditions in that sense. But internal facts about me—the fact that ‘sofa’ applies to ‘whatever my community standards dictate’—determine application conditions for ‘sofa’, in another sense. And the key internalist claim is that this second sense is primary. That is, it’s only because I (internalistically) defer to community standards that those (external) standards are relevant to fixing the referent of my term ‘sofa’.