What is purism? What are epistemic standards?
Dec 27th, 2009 by Jonathan
I’m reading Fantl and McGrath’s new knowledge book. An important thesis of the book is that of Impurism. Impurism is defined in chapter one as the denial of Purism, given thus:
(Purism about Knowledge) For any subjects S1 and S2, if S1 and S2 are just alike in their strength of epistemic position with respect to p, then S1 and S2 are just alike in whether they are in a position to know that p.
Impurism is also defined, a bit differently, in chapter two:
(Impurism) How strong your epistemic position must be — which purely epistemic standards you must meet — in order for a knowledge-attributing sentence, with a fixed content in a fixed context of use, to be true of you varies with your circumstances. (35)
Impurism, Fantl and McGrath think, is a counterintuitive claim; adopting it is, according to the central argument of the book, the heavy cost that is worth paying for fallibilism. I have a hard time understanding why, if it really is so counterintuitive, people find it so. I’m also far from convinced that people do find it so. Purism and impurism are technical notions that require a fairly sophisticated background in epistemology to understand. Furthermore, they are given here in terms of the far from explicit notion of ‘purely epistemic standards’ and ‘strength of epistemic position’. What factors influence the ‘strength of one’s epistemic position’? Intuitively, what one knows is of great relevance to the strength of one’s epistemic position, but Fantl and McGrath cannot be using the term in a way that licenses this intuitive verdict; otherwise purism would be trivially true. They must have some notion other than the intuitive one in mind. What is it? And do we really have intuitions about it?
At points, Fantl and McGrath describe the factors that count towards strength of epistemic position as ‘truth-relevant’ factors. (I think that DeRose also used this gloss in his characterization of ‘intellectualism’, which I think is just meant to be the same thing as F&M’s ‘purism’.) This is meant to rule in facts about the actual or probable truth of the indicated belief, and to rule out facts like what is salient to the subject or how much is at stake for her. That’s some progress — but is it clear enough? It is, I think, meant to be consistent with purism that whether a subject knows depends on whether she is proceeding responsibly in forming her belief. (It’d better be, because that’s a traditional view, and purism is supposed to include that tradition.) Is this factor ‘truth-relevant’? I guess it’s supposed to be. We could rely on a principle like this: if a belief is responsibly formed, then it is likely to be true.
Similarly, it’s meant to be consistent with purism that whether a subject knows depends on features of her environment — even those that don’t affect the truth value of her belief. For example, whether a subject knows that there is a barn in front of her depends in part on whether there are barn façades nearby. Presumably, this is roped in under truth-relevance by the effect of such circumstances on the reliability of (a certain specification of) the subject’s belief-forming process, which is correlated with truth.
But if a connection that weak is sufficient to count responsibility and environmental features as truth-relevant, then it’s hard to see why it shouldn’t also count in knowledge, and thus, if the views developed by e.g. Fantl & McGrath, Stanley, etc. are right, make practical situations truth-relevant. Your stakes, like your environment, play a role in determining what you know, and knowledge, like responsibility and reliability, is strongly connected to truth. In what sense is such a stakes-sensitive view ‘impurist’? In what sense are stakes disconnected from ‘purely epistemic standards’?
I don’t even understand what purism amounts to, if it’s not the triviality that this line of reasoning suggests. And I certainly don’t have intuitions about purism. Therefore, I have a hard time seeing what all the fuss is about. (Interesting sidenote: it looks like common ground among at least many SSI types and at least many contextualists that one of the motivations for contextualism is to maintain purism. I’m a contextualist who has never had anything like that motivation; indeed, it looks pretty incomprehensible to me.)
Hi Jonathan. When I’ve suggested to epistemologists that stakes can make a difference to knowledge, they have usually reacted like this: well, stakes can make a difference to belief, of course, but that’s the only way they can make a difference to knowledge. The reaction I get is not at all like the reaction an internalist might get in front of an externalist crowd, or vice versa. Epistemologists think that stakes are the wrong *kind* of the thing to make a difference to knowledge, except by affecting belief. There’s a “wrong kind of factor” intuition. A sort of category error intuition. People didn’t react to Nozick by simply saying scornfully that whether you track the truth *clearly* makes no difference to whether you know.
Suppose we say that someone is in a position to know p iff either they know p or the only thing standing in the way of their knowing p are the doxastic conditions regarding p. (We are factoring out not only belief here, but properly based belief as well as responsible belief – since those are at least partly doxastic.) Why does it seem to so many epistemologists that stakes are the wrong kind of thing to make a difference to whether one is in a position to know?
I don’t know what to say other than to look at what people seem to agree are right kinds of things to make a difference: evidence, availability of reliable processes, availability of a safe indicator, etc. All these things seem to me to be truth-relevant in one or another sense.
I can see why you might hope for more illumination about an umbrella notion of truth-relevance, but maybe you can agree that there is a “wrong kind of factor” intuition that we can attempt to articulate and to find a deeper basis for.
One can come at all this from a different angle, by working off a particular case. Consider the Cohen airport case. We have Smith, who doesn’t care much about whether the flight stops in Chicago, and Mary and John who care a great deal. Smith is relying on his itinerary in believing that it stops in Chicago. Mary and John overhear Smith saying “yeah, it stops there; that’s what my itinerary says.” Jonathan, I assume you think it sounds bizarre to say about this case: “Smith knows it’s stops in Chicago but Mary and John don’t.” I don’t mean merely that ordinary speakers won’t say this. I mean: it sounds just false that Smith knows in this case while Mary and John don’t.
Now of course Mary and John don’t believe it stops there, so maybe that’s why it sounds weird. But that can’t be it. It seems wrong to say about this case, “Smith is in a position to know but Mary and John aren’t.” Why does this seem wrong? A natural idea here is that there are certain features that Smith shares with Mary and John which would suffice to put all or none of them in a position to know. What are these? A first thought is: evidence.
So, the first thought is: subjects with the same evidence for/against p are alike in whether they are positioned to know p.
This first thought is a “purist” principle. Of course, this principle even if true (which I don’t think it is, independently of anything about stakes) won’t cover all the relevant cases. Suppose Smith is not relying on evidence in any obvious sense – he is relying on memory. We can suppose that Mary and John, too, are relying on memory, and that their memories are just like his (pretty strong but not clear and distinct). Even if we want to say that Mary and John, because of the stakes, should check further, while Smith need not, it seems wrong to say that Smith is positioned to know while Mary and John aren’t. Why? If memory isn’t evidence, we should say something perhaps like this:
Subjects with the same profile of evidence and non-doxastic justification with respect to p are also alike with respect to being in a position to know that p.
But some will complain about the notion of nondoxastic justification, and prefer something else: the presence of a safe indication, or the availability of a belief-independent reliable process. Etc. etc. Talk of strength of epistemic position is a way of attempting to bypass internecine disputes like this.
Sorry to go on so long. I could put my response simply as follows: the key intuition is that stakes are the wrong kind of factor; the attempt to formulate a doctrine of purism is an attempt to try to figure out what about stakes makes them the wrong kind of factor.
Do you share the *wrong kind of factor* intuition at least?
Hi Matt — please forgive the long delay in responding. Thanks for the comment.
So is your view that what characterizes purism is merely that the factors that influence whether someone knows are not very surprising to some favored subclass of individuals? That’s a coherent notion, of course, but it doesn’t strike me as a very interesting one. “Purism” is a funny name for it. And I don’t see that it’s terribly intuitive. Furthermore, I don’t see that it gets all the cases to line up the way you want. I think you mean it to be consistent with purism that whether I know p can depend in part on how close the nearest fake barn is to me. But that’s a surprising, counterintuitive claim — many people will react by saying it *clearly* doesn’t make a difference and is the wrong *kind* of thing to make a difference.
On this:
“Suppose we say that someone is in a position to know p iff either they know p or the only thing standing in the way of their knowing p are the doxastic conditions regarding p.”
I would have thought that knowledge is a ‘doxastic condition’ in your sense. Isn’t knowledge ‘partly doxastic’?
Then you ask:
“Why does it seem to so many epistemologists that stakes are the wrong kind of thing to make a difference to whether one is in a position to know?”
But I take that to be a separate question. We can argue on the merits of a particular thesis, like the thesis that stakes make a difference to knowledge, or whether one is ‘in a position’ to know. And you’re right, that strikes many people as counterintuitive, and you’re also right that there are interesting arguments in favor of it anyway. I just don’t see that it’s any different in kind from arguments about, e.g., whether purely external factors can influence justification facts, or whether elements outside one’s control can make a difference to moral facts, or whatever. I also don’t see that this suggestion has anything very interesting in common with other forms of ‘impurism’, such as a view that says that what possibilities are salient make a difference to knowledge facts.
“I don’t know what to say other than to look at what people seem to agree are right kinds of things to make a difference: evidence, availability of reliable processes, availability of a safe indicator, etc. All these things seem to me to be truth-relevant in one or another sense.”
Well, ok. But my point is that knowledge seems to be truth-relevant in at least the same senses. So if you’re trying to characterize what the unsurprising features have in common, this doesn’t seem to do the trick.
“The key intuition is that stakes are the wrong kind of factor; the attempt to formulate a doctrine of purism is an attempt to try to figure out what about stakes makes them the wrong kind of factor.
Do you share the *wrong kind of factor* intuition at least?”
I don’t think I do. I don’t really have strong commitments on what kinds of factors can be relevant to these cases. It strikes me as a somewhat surprising suggestion, but no more so than various suggestions that you do want to count as purist.
>So is your view that what characterizes purism is merely that the factors that influence whether someone knows are not very surprising to some favored subclass of individuals?
No, the way I think about it is like this. First, we notice that there is a strong intuitive pull, felt at least by many philosophers, against the idea that stakes matter to knowledge (independently of mattering to belief) and a similar pull against the idea that salience of error possibilities does, too. Second, we notice that there is a *wrong kind of factor* intuition in these cases. Third, we try to figure out why these factors seem so wrong, and one first stab at an explanation here is to see that they are not truth-relevant. Then we formulate purism as the thesis that only the truth-relevant stuff matters to being in a position to know.
>I think you mean it to be consistent with purism that whether I know p can depend in part on how close the nearest fake barn is to me. But that’s a surprising, counterintuitive claim — many people will react by saying it *clearly* doesn’t make a difference and is the wrong *kind* of thing to make a difference.
I know Hawthorne makes this point, and Stanley, too. But I never found it convincing. DeRose had a nice reply in his book:
“…it just doesn’t seem so implausible to suppose that whether a subject knows can depend on whether his way of telling is reliable in the particular surroundings he finds himself in” (198)
I’d add that this doesn’t seem so implausible even when we add that the subject doesn’t know his way of telling isn’t reliable in his particular surroundings.
>On this:
“Suppose we say that someone is in a position to know p iff either they know p or the only thing standing in the way of their knowing p are the doxastic conditions regarding p.”
I would have thought that knowledge is a ‘doxastic condition’ in your sense. Isn’t knowledge ‘partly doxastic’?
Right. It’s just a matter of getting an accurate and convenient terminology. It’s not true that knowledge is purely a matter of truth-relevant factors; it’s also a matter of doxastic factors. So, we introduce purism as explicitly about being positioned to know. We call this purism about knowledge, partly to avoid having to use ‘position to know’ hundreds of additional times in the book!
> “I don’t know what to say other than to look at what people seem to agree are right kinds of things to make a difference: evidence, availability of reliable processes, availability of a safe indicator, etc. All these things seem to me to be truth-relevant in one or another sense.”
Well, ok. But my point is that knowledge seems to be truth-relevant in at least the same senses. So if you’re trying to characterize what the unsurprising features have in common, this doesn’t seem to do the trick.
Now you’re wondering whether we can specify more precisely what we mean by ‘truth-relevant’ in such a way as to make purism non-trivial. For one thing, as you noted above, knowledge is partly doxastic and not pure truth-relevant. So it is out. You might ask, though: what about being positioned to know? We do say a few things about truth-relevant factors that would help rule out being positioned to know. We say that truth-relevant factors are standings on truth-relevant dimensions. If we think of truth-relevant dimensions as measures of probability of truth in an externalist objective sense or in an internalist sense, then a standing on a truth-relevant dimension corresponds to a particular probability level. I take it that “positioned to know” doesn’t do that unless it gives epistemic probability 1, and so if fallibilism of our kind is true, “positioned to know” won’t be a standing on a truth-relevant dimension. (We can talk about this more if you want.)
Another move one could make here is to talk about the factors explanatorily relevant to whether one is positioned to know, and the claim would be that these are purely truth-relevant. Knowledge and being positioned to know would be ruled out as not being explanatorily relevant to whether one is positioned to know. One possible drawback of going this way would be worries about closing off a knowledge-first epistemology.