The Rules of Thought: Philosophy and the a priori

I'm going to live up to the blogger stereotype and set a few posts on autofocus. The shameless project is to make the case that you might have good reason to read The Rules of Thought, the book that Benjamin Jarvis and I recently wrote. (OUP catalogue page) (my webpage)

I think that there are three possible hooks into our project. One of them -- the one that represented our own way into the project -- concerns the epistemology of the a priori in general, and the epistemology of philosophy in particular. Ben and I trace this interest pretty specifically to 2005, when, while PhD students at Brown, we took Joshua Schechter's seminar on the a priori, and also attended Timothy Williamson's Blackwell-Brown lectures, which eventually became The Philosophy of Philosophy. We were attracted by traditional idea that in many paradigmatic instances, philosophical investigation proceeded in some important sense independently from experience, but came to appreciate that (a) there were deep mysteries concerning the explanation for how this could be, and (b) there were strong challenges that suggested that the traditional idea couldn't be right. For example, the traditional idea has it that judgments about thought experiments constitute appreciate of facts that are both a priori and necessary; but Williamson gave what is now a somewhat famous argument that this can't be so: thought experiments don't include enough detail to entail the typical judgments. So the best they can support is something like a contingent, empirical counterfactual: if someone were in such-and-such circumstances, he would have JTB but no K, etc.

We wrote a defensive paper in response to Williamson's argument, explaining how one can understand the content of thought-experiment judgments in a way that renders them more plausibly necessary and a priori, invoking the notion of truth in fiction. ("Thought-Experiment Intuitions and Truth in Fiction" -- (draft) (published)) That paper did two useful things: it gave an objection to Wiliamson's treatment, and it defended a traditional aprioristic picture from Williamson's particular critique. But on the latter score, it was purely defensive; it did little to explain how a priori justification or knowledge was possible, or to articulate just what apriority could consist in. Another paper, "Rational Imagination and Modal Knowledge," (d) (p) gave a bit more epistemological background, and a focus on modal epistemology in particular. By the time of that paper, we were underway on the book.

What we needed, we realized, was a much fuller story about apriority, including detailed engagement with extant critiques of the notion. We give this in Part II of The Rules of Thought. Some of the critiques -- in particular, some of those from Williamson and Hawthorne, as well as some similar challenges from Yablo and Papineau -- show that a characterisation of apriority in terms of more psychological states like knowledge and justified belief is extremely difficult, perhaps impossible. (Here's a related blog post from last year.) Our general characterisation of the a priori is a negative one, given in terms of propositional justification. A subject has a priori propositional justification for p just in case she has justification for p, and this isn't due in constitutive part to any of the subject's experiences. We explain how this approach avoids the challenges to the a priori that are in the literature, and argue that there is strong reason to think that philosophical investigation is often a priori in our sense. The focus on propositional justification requires a fairly strong version of the traditional distinction between warranting and enabling roles for experience, which we attempt to explicate.

The negative characterisation is thin by design. We are explicitly open to a kind of pluralism about apriority, according to which various positive epistemic states can realise apriority. The state we focus on most is what we call 'rational necessity' -- certain contents are, we think, by their nature such that there is always conclusive reason to accept them. (Much more on this idea in another post on another motivation for the project.) But we allow that other states may realise apriority as well; we are open, for example, to the idea that it is a priori that perception is generally reliable, even though this isn't rationally necessary. Perhaps some kind of pragmatic explanation for these a priori propositions may be found.

In the context of our theory of the a priori, and our more detailed positive story about rational necessity, we rehearse the main ideas from our two previous papers on philosophical methodology: thought-experiment judgments, properly understood, often have contents that are rationally necessary, hence a priori; so likewise for many judgments in modal epistemology concerning what is metaphysically possible. This all happens in Part II of the book.

So that's the first hook for our book: understanding the a priori and the epistemology of philosophy. We tell a story that is able to vindicate a number of pretty traditional ideas about how philosophy works (but without problematic focus on words or concepts). The other two hooks will each get another post -- one concerning Fregean ideas about mental content, and one about the role of intuitions.

The Rules of Thought

Benjamin Jarvis and I have been working for some time now on a book manuscript on mental content, rationality, and the epistemology of philosophy. I posted a TOC of our first draft last summer. Since then, we've received some helpful comments from reviewers, and have revised extensively; we now have a full new draft, which we feel ready to share with the public. If you're interested, you can download the large (2.3 MB, 331 page) pdf here. Comments and suggestions are extremely welcome.

I'm including a table of contents of the new draft in this post, to better give an idea of what we're up to.

Concepts and Survey Results

I'm thinking about a point that Ernie Sosa has made in response to survey-based experimental philosophy challenges. As we all know, some critics have argued that certain experimental results challenge traditional armchair philosophy. In particular, for example, Weinberg, Nichols, and Stich found that there seemed to be a systematic divergence of epistemic intuitions depending upon the ethnic background of the subjects studied: students of East Asian descent were more likely than students of European descent to, for instance, describe Gettier cases as cases of knowledge.

Here's a line that Ernie has pressed a few times now:

And the disagreement may now perhaps be explained in a way that casts no doubt on intuition as a source of epistemic justification or even knowledge. Why not explain the disagreement as merely verbal? Why not say that across the divide we find somewhat different concepts picked out by terminology that is either ambiguous or at least contextually divergent? On the EA side, the more valuable status that a belief might attain is one that necessarily involves communitarian factors of one or another sort, factors that are absent or minimized in the status picked out by Ws as necessary for “knowledge.” If there is such divergence in meaning as we cross the relevant divides, then once again we fail to have disagreement on the very same propositions. In saying that the subject does not know, the EAs are saying something about lack of some relevant communitarian status. In saying that the subject does know, the Ws are not denying that; they are simply focusing on a different status, one that they regard as desirable even if it does not meet the high communitarian requirements important to the EAs. So again we avoid any real disagreement on the very same propositions. The proposition affirmed by the EAs as intuitively true is not the very same as the proposition denied by the Ws as intuitively false.

(That's quoted from his contribution to the recent Stich and His Critics volume.)

As I'd understand it, the core suggestion here is this: maybe there's no real disagreement here; some group of subjects say that such and such 'is a case of knowledge,' while philosophers and other subjects say that such and such is not a case of knowledge, and there's no genuine disagreement, because the former subjects don't mean knowledge by 'knowledge'.

So here's my question. (One question, anyway. I have a few more.) What does any of this have to do with concepts? As I understand it, it's a question about meaning and reference: what does the word 'knowledge' refer to in a given subject's mouth? One can run a little detour through concepts if one wants: word meanings are concepts; the concepts are different; so the word is ambiguous. But what, if anything, does this 'conceptual ascent' contribute? I rather suspect that it does more to distract than to help. Steve Stich's response to Sosa emphasizes concepts in a way that looks to me largely irrelevant:

There is a vast literature on concepts in philosophy and in psychology (Margolis and Laurence 1999; Murphy 2002; Machery forthcoming), and the question of how to individuate concepts is one of the most hotly debated issues in that literature. While it is widely agreed that for two concept tokens to be of the same type they must have the same content, there is a wide diversity of views on what is required for this condition to be met. On some theories, the sort of covert ambiguity that Sosa is betting on can be expected to be fairly common, while on others covert ambiguity is much harder to generate. For Fodor, for example, the fact that an East Asian pays more attention to communitarian factors while a Westerner emphasizes individualistic factors in applying the term ‘knowledge’ would be no reason at all to think that the concepts linked to their use of the term ‘knowledge’ have different contents (Fodor 1998).

But Fodor's theory of concepts is not a theory of word meanings. What bearing does it have on whether there might be an Asian-American idiolect in which 'knowledge' means something other than knowledge? (I do mean this as a serious question; I'm less fluent in Fodor than I'd like.)

To my mind, the sort of view that Ernie needs to be worrying about is not Fodor's but Burge's. More on that in a future post, I think. For now, just this question: is anything usefully gained by thinking about Sosa's suggestion here in terms of concepts?

Generality of Gettier Judgments

I'm teaching a contemporary epistemology course with Yuri to Honours students this year. We started with Linda Zagzebski's "The Inescapability of Gettier Problems", which, to my mind, helpfully turns attention away from attempts to analyze knowledge on which students may have spent much of their intro epistemology courses. I read it a few years ago, and found it totally convincing; I read it again this week, and found it totally convincing again, but noticed that the argument wasn't nearly so straightforward as I'd thought it was. In fact, I'm not sure what it is. (But I still find it compelling.)

Here's what Zagzebski says. She understands Gettier as having refuted the JTB theory thus: imagine a case in which JB but not T. Now change the case so that T, but just by luck -- not in a way connected to JB. Now you have a Gettier case -- an intuitive counterexample to K = JTB. That's what she said Gettier did. Then she says we can generalize the argument. Her target is any view that tries to analyze knowledge as T + X, where X doesn't entail T. Do just the same thing, she says, as Gettier: take a case in which X and not T (guaranteed possible), then tweak the case so as to make T true in a way unrelated to X.

(One might worry here as to whether this latter step is always possible. Juan Comasaña told me via Twitter that he wants to resist the argument here. I have a hard time seeing how it couldn't be done, for any X that's plausibly natural enough to figure into an analysis. We'd need X to be consistent with not-T, but for X & T together to entail that X and T are closely connected. That seems, at least, really weird. Maybe there's an argument lurking that this is impossible? Or maybe it's possible after all? I'm not sure. Thoughts? Anyway, this isn't the point I wanted to press.)

Ok, so, modulo the parenthetical, we've generated a case according to Zagzebski's recipe. Now, she tells us, we have a counterexample to the K = T + X theory. She offers:

...a general rule for the generation of Gettier cases. ... Make the element of justification (warrant) strong enough for knowledge, but make the belief false. ... Now emend the case by adding another element of luck, only this time an element which makes the belief true after all. The second element must be independent of the element of warrant so that the degree of warrant is unchanged. ... We now have a case in which the belief is justified (warranted) in a sense strong enough for knowledge, the belief is true, but it is not knowledge.

What's interesting about this passage is that she's making a general claim about the ultimate outcome of all instances of her argument schema. But the original Gettier argument, it is traditionally thought, depends on a particular sort of judgment about a particular case; we think about the story about Smith and Jones and Brown in Barcelona, and see that this is a case of JTB without K. If that's right, then it's totally mysterious how Zagzebski or anyone could be confident that the same pattern will hold of other attempts to analyze. But the argument isn't a non sequitor; it's (at least) prima facie compelling. Why?

At a workshop on thought experiments I attended in Brazil this summer, Anna-Sara Malmgren suggested that thought experiment judgments carry with them a kind of implicit generality that is best explained by their being products of nonconscious inferential reasoning. This, it seems, might be just the sort of case to support her suggestion. Our initial Gettier judgment constituted a kind of commitment to a general principle that rules out the kind of luck that Zagzebski is focusing on. If that's right, then metaphilosophical emphasis on cases may be misplaced; lots more of our thought experiment judgments may be more based on theory than is always realized. Without a move like that, it's hard for me to see how Zagzebski's argument could make any sense.

How rich is truth in fiction?

According to orthodoxy, what's true in a fiction goes beyond what's entailed by the text making up the story. Although fictions are gappy (there's no fact about whether Hamlet had an even number of hairs), some things are determinately true without being stated, or being entailed by thugs that are stated (Hamlet was not a leprechaun). This orthodoxy is pretty much universal, I think, and I've relied on it in my work on thought experiments.

In the past few months, I've worried a bit about that orthodoxy. I don't think orthodoxy here should be abandoned, but I do think it faces an important challenge that hasn't, to my knowledge, been articulated before. The challenge begins with a consideration of non-fiction.

Not all non-fiction is true; some works of non-fiction are mistaken, and some are fraudulent. (All biographies are non-fiction, but not all biographies are true.) What determines whether a non-fiction is true? The key to the challenge is this: we can and should distinguish between whether a work of non-fiction is true, and whether it is merely misleading. I could write a very deceptively misleading biography of David Lewis, such that anyone who read it would walk away with rampant false beliefs about him. But if I did so using only true sentences, relying on pragmatic implicatures and natural assumptions to generate the misleading nature of my non-fiction, then, I claim, the biography I have written is true.

Now take a fiction made up of just the same sentences I used in my misleading autobiography of Lewis. This is just the sort of situation where, according to orthodoxy, principles of generation for truth in fiction will generate false propositions and add them to the set of fictional truths. But this, given what we've said in the previous paragraph, is inconsistent with the truism that contents of fictions don't work in ways radically different from those of non-fictions. A non-fiction's content is true if its sentences are. Can we really deny that a fiction, sentence-by-sentence identical with a non-fiction, has true content if its corresponding non-fiction does? That's the puzzle.

Here, as I see them, are the options:

1. Reject orthodoxy. What's true in the fiction does not, after all, go beyond what's given in the literal text.


2. Posit a stark disanalogy. Their obvious forms of similarity notwithstanding, fictions and non-fictions get content in radically divergent ways.


3. Bifurcate 'content'. (Brian Weatherson suggested this to me when I posed the puzzle to him.) Agree with the conclusion about 'content' of fictions in some sense, while insisting that there's a richer 'true in the fiction' that goes beyond content.

I guess I'm inclined to agree with Brian that, of these choices, (3) is the best way to go. But I'd be interested to hear if anyone thinks I'm selling the other possibilities short, or have overlooked additional possible solutions.

Thought Experiments Lecture Slides

I'm kicking off the Arché Summer School this year; here are the slides for my talk. (PowerPoint) (pdf)

(This is mostly designed for the attendees, although I guess it's conceivable that others could find them interesting. I don't have a handout; instead, I have a URL where interested parties can look at the slides, quotes, references, etc. in more detail.)

Real-World Deviant Gettier Case

Something cool happened in our methodology seminar last week. Some people like to remark on real-world Gettier cases they find themselves in. I found myself last week in the presence of a real-life deviant Gettier case.

A deviant Gettier case (what Ben Jarvis and I have also called a 'bad Gettier case') is a situation in which the literal text used to describe a Gettier situation is satisfied, but in such a way so as to fail to provide a counterexample to JTB=K. Deviant Gettier cases play a central role in a disagreement Ben and I have with Timothy Williamson. What's cool about this deviant Gettier case is that (a) although I played a central role in producing it, I did so entirely without design, and (b) it's deviant with respect to one of the standard paradigms of Gettier cases.

Here's what happened.

Knowing the Intuition and Knowing the Counterfactual

Knowing the Intuition and Knowing the Counterfactual, (2009) Philosophical Studies, 145(3), September 2009: 435-443. Please refer to published version here. For a Philosophical Studies book symposium on Timothy Williamson's The Philosophy of Philosophy. See also Williamson's response here.

I criticize Timothy Williamson’s characterization of thought experiments on which the central judgments are judgments of contingent counterfactuals. The fragility of these counterfactuals makes them too easily false, and too difficult to know.

Thought-Experiment Intuitions and Truth in Fiction

Thought-Experiment Intuitions and Truth in Fiction, with Benjamin Jarvis. (2009) Philosophical Studies 142 (2), January 2009: 221-246. Please refer to published version, available online here.

What sorts of things are the intuitions generated via thought experiment? Timothy Williamson has responded to naturalistic skeptics by arguing that thought-experiment intuitions are judgments of ordinary counterfactuals. On this view, the intuition is naturalistically innocuous, but it has a contingent content and could be known at best a posteriori. We suggest an alternative to Williamson’s account, according to which we apprehend thought-experiment intuitions through our grasp on truth in fiction. On our view, intuitions like the Gettier intuition are necessarily true and knowable a priori. Our view, like Williamson’s, avoids naturalistic skepticism.