Saturday, 27 January 2007

An Exchange with Adrian Haddock

Note: The following was originally supposed to be just a reply to Adrian's comments (See McDowell, Knowledge and Credit below), but I decided to feature it as its own separate post so that it doesn't distract from the other important issues raised by Adam's post. Of particular interest are Adrian's elucidative comments that follow.

As is usually the case with his reflections on McDowell, I find Adrian's comments very illuminating. While I agree with his point that locutions like 'reflectively accessible' are 'philosophical terms of art', I'll like to play devil's advocate by arguing that there is something problematic about McDowell's employment of the concept. McDowell's commitment to 'reflective accessibility' is typically instantiated in his claim that a subject's epistemic standing is determined solely by factors that constitute 'how thing are with her subjectively.' I take this expression to be an umbrella term for the kinds of things we typically take to be reflectively accessible, such as beliefs, experiences etc. But by using this expression, McDowell doesn't define what he means by 'reflectively accessible'. Rather, he merely implicitly appeals to a list of items we typically take to be such. But that is a far cry from providing us with an account of what 'reflective accessibility' consists in.

Adrian's locution: 'accessible to the subject in the way in which the subject's mental states are accessible to the subject' is equally unhelpful. Again, all it does is list one item, namely mental states, that is typically taken to be reflectively accessible, while failing to say precisely what is meant by 'reflective accessibility.' The question then becomes, what does it mean for mental states to be 'reflectively accessible'? This question is particularly pertinent given that McDowell's brand of singular anti-individualism seems to entail that a subject may often lack knowledge about her own mental states, such that she may think she is entertaining a certain thought when in fact she is having an illusion of thought. This picture of mental states, if not at odds with our putative understanding, is certainly not how we would typically think about reflectively accessible items. Consequently, it is misleading to simply define 'reflective accessibility' via an ostensive reference to items we take to be paradigmatic cases of reflective accessibility, when these very items have been re-described so that they no longer match our putative understanding of what reflective accessibility consists in.

In sum, the McDowellian take on the question of reflective accessibility amounts to the following rhetorical slight of hand: to redefine mental states etc, in a way that makes reflective access to them problematic, and then to simply define 'reflective accessibility' via an ostensive reference to mental states, all the while pretending to talk about them in the way we typically do.


Adrian said...


Just a few comments on your recent post. Remember that McDowell never uses the expression "reflectively accessible". Also remember that when he appeals to "how it is with one subjectively" [in 'Criteria...'] what he says is that "the obtaining of the fact is ... not blankly external to [one's] subjectivity". As far as I can see, there is no reason to read this as meaning anything more than: one sees that P. Seeing, for McDowell, is a configuration in one's subjectivity (or, if you prefer, is a state of one's mind), and when one sees that P one has the obtaining fact itself as the content of one's subjective (i.e. mental) state; in that sense - and that sense alone, I think - the fact is not external to one's subjectivity. What licenses one to read it in any other way?

Also, you say at one point that we have an "understanding" of the notion of reflective accessibility on which X is not reflectively accessible to us if we "may lack knowledge" of X. Now I do not think I understand this. What do you mean "may lack knowledge"? We "may lack knowledge" of facts that we can know infallibly, simply by not having formed any beliefs about them.

Do you mean to equate being reflectively accessible with being infallibly knowable? If so then we know that McDowell does not think the fact that we see that P is reflectively accessible (because he does not think it is infallibly knowable - see 'Singular Thought and the Extent of Inner Space').

The more important point is that these worries about "reflective accessibility" may be chimerical. McDowell never uses the term, so he can hardly be charged with failing to define it. More importantly, in trying to understand the kind of knowledge we have of our mental states, it is not a good idea to fixate upon a term that at best we do not understand, and at worst serves as a repository for dodgy philosophical ideas.



Secretly, I’m very sympathetic to all you have said. But I wouldn’t be much of a devil’s advocate if I gave up that easy! I grant that McDowell never uses the expression “reflective accessibility” (the truth is, I only used the expression myself in order to maintain some continuity with your (and Adam’s) earlier comments). However, while he never uses the words “reflective access” he does come close. For example, consider footnote 7 in “Knowledge and Hearsay”:

“I agree…that we lose the point of invoking the space of reasons if we allow someone to possess a justification even if it is outside his reflective reach.”

Now assuming that “reach” as employed here, is a close synonym to “access” then it seems as though one could substitute all talk of “reflective access” in my previous post with “reflective reach” and still run my argument as before.

Now the worry I raised previously can be put like this: Consider the following statement of what it means for a thought (or mental content) to be within a subject’s reflective reach:

For any subject S, some thought, M, is within S’s reflective reach at some time t if and only if M is readily accessible to S in the sense that she can readily know, at t, that she is having M.

Now, McDowell’s first line of defence would be to point out that the mere fact that one can mistakenly believe oneself to be awake when one is dreaming does not mean that when one is awake one cannot tell that one is not dreaming. Likewise, the mere fact that one may mistakenly take oneself to be entertaining a certain thought when one is not, does not mean that when one is entertaining a thought one does not know that one is not merely suffering an illusion of thought. String of double negatives aside, so far so good.

But notice that this apparently innocuous idea has strikingly counterintuitive implications regarding a subject’s mental life. For example, consider a subject who is the victim of a perceptual Getteir-type deception [see my post Un-discriminating Reliabilism (part 1) for an example of such a Gettier case]. According to McDowell, the Gettier subject in such a case merely has an illusion of entertaining the relevant demonstrative thought, “that cube on the table is red”. As such, she does not have the reason for believing the embedded proposition that she takes herself to have. The upshot of this picture is that in a Gettier-type case, the subject has no justificatory reason for believing the corresponding existential proposition (e.g., “there is a red cube on the table”, when she is actually looking at a hologram of a cube, etc.). But that flies in the face of a common sense assessment of the Gettier subject. Common sense says that the subject does have a justificatory reason. In fact, that’s what intelligibly makes it a Gettier case.

Now, if McDowell is willing to admit that by denying the intelligibility of such Gettier cases (i.e., justified true beliefs that don’t amount to knowledge) he is proposing a philosophical revision of our putative notions, then fine. I, for one, think that common sense is often mistaken, and perhaps it is the task of good philosophy to uncover such errors. But that would mean that McDowell (self-proclaimed champion of non-constructive philosophy) would have to give up the pretence of being some sort of new Thomas Reid, defender of the intuitions of the ordinary man.

P.S.: That last bit is meant as a joke, I doubt McDowell ever claimed to be the new Thomas Reid. :-)

Adrian said...

Hi Avery—

This is fun. I’ve studiously avoided the “blogosphere” up to now, and just hope that taking part in this exchange doesn’t lead me to form what seems to be a very popular habit…

So, let us grant that the subject in your Gettier case cannot essay the singular thought in question. I am not sure why it is supposed to follow that she does not have the reason she takes herself to have. What reason does she take herself to have? That she sees that there is a red cube in the box? That she sees that red cube there? Let us grant that she does not have those reasons. Still, she surely has some reason (which she takes herself to have): namely, that it looks to her as if there is a red cube in the box. Of course, this reason is not enough to ensure that she knows the proposition in question. But she still has some justification for believing it. And I do not know why you think it is not open to McDowell to say this.

Of course, McDowell does think that the subject who sees that P has a justification for believing that P that the subject in the Gettier-case (who at best can have it look to her as if P) does not have. And this flies in the face of a standard way of describing Gettier cases, according to which the subject who knows that P has exactly the same justification (exactly the same reasons, we might say) for believing that P as the subject in the Gettier-case. But I cannot see that the person who adheres to this standard description of Gettier-cases – itself a classic example of a “highest common factor” conception – and denies McDowell’s different view, can claim a monopoly on common-sense.

You are, of course, right about the “reflective reach” passage – even though it is, I think, important that McDowell does not use the familiar (and tainted) phrase “reflective access” – so perhaps I can put the point a different way. We are self-conscious subjects, and when we leave the study, and hit the streets, we do take ourselves, not only to see things, but also to know that we are doing so. McDowell is hardly going to deny that (because he *does* want to show that we are entitled to our pre-reflective view of things). That, I think, is what the “reflective reach” passage signals. If we are going to stand a chance of making sense of this second-order knowledge, it is not a good idea to begin by uncritically employing a phrase (viz. “reflectively accessible”) that philosophers typically use to mean (something like) “accessible even by someone who doesn’t see that P”, and thereby rule out the pre-reflective view from the outset.

Of course, I realise that you are only playing devil’s advocate. But I think a case for the devil has yet to be made.



I think I’ve pretty much exhausted my ability to play devil’s advocate on this question; good job Adrian!

Thankfully, your replies suggest that I’m not totally off regarding my own interpretation of McDowell on this score. Specifically, I agree that McDowell need not be committed to the idea that the Gettier subject has no justification. I have tended to unpack this claim via an appeal to our quotidian distinction between having some reason for believing such and such, and having a good reason for believing such and such. In our ordinary speech we often describe someone who has carried out a valid inference, in which (unknown to them) one or more of the premises fails to obtain, as still having some reason for believing the conclusion. This “reason” is embodied in the valid logical form of the inference itself. Saying they have “some reason” is our way of tipping our hat to the fact that they have acted in a manner in keeping with what is expected of a rational agent; they have, to use a McDowellianesque expression, discharged their doxastic reponsibility. But they do not have a “good reason”. Though the inference is valid, it is not sound. But essentially the unsoundness of the inference is due to no fault of their own as responsible epistemic agents.

The inference the Gettier subject makes from the attempted demonstrative, “that cube on the table is red” to the existential claim, “there is a red cube on the table” is a valid one, and displays an adequate understanding of the relevant evidential relations. The Gettier subject therefore has some reason for believing the embedded proposition, instantiated in the valid logical form of inference itself. However, it is not a “good reason” because the inference rests on an empty premise – i.e., there is no actual demonstrative thought being entertained. This distinction between having “some reason” and having a “good reason” seems quite natural, and appears to be in keeping with how we ordinarily talk about these types of cases. Admittedly, our ordinary application of this distinction is often not as explicit and systematic as here articulated, but I believe the formalisation of our murky putative distinctions is well within the prerogative of non-constructive philosophy.

However, I wonder how persuasive the considerations that Adrian (and now I) have raised are for individuals who aren’t already enamoured with the McDowellian view. Will a real anti-McDowellian please stand up?