Williamson on Evidence

In the previous post I mentioned that Williamson clearly opposes in chapter 7 a broadly psychological conception of philosophical evidence. But it isn’t exactly clear just what his target is. At times he seems to be arguing against psychological evidence ever being philosophically worthwhile. For example, consider the following batch of quotes.

“For now I face the challenge of arguing from a psychological premise, that I believe or we are inclined to believe the Gettier proposition, to an epistemological conclusion, the Gettier proposition itself. That gap is not easily bridged.”

“Since psychological evidence has no obvious bearing on many philosophical issues, judgment scepticism is also encouraged in ways that do not depend on the consequence fallacy.”

“In explaining why we have intuitions, analytic philosophy has a preference for explanations that make those intuitions true over explanations that make them untrue, but the justification for that preference remains unclear”

In those quotes his opponent seems to hold the relatively weak view that psychological evidence can (sometimes) be useful evidence for philosophical conclusion. But other times he seems to take his opponent to be the person who holds the much stronger view that only intuitions are evidence. For instance, he says

“One result [of EN] is the uneasy conception many contemporary analytic philosophers have of their own methodology. They think that, in philosophy, ultimately our evidence consists only of intuitions.”

I think it’s hard to believe that’s really a widespread view in philosophy. Does Singer’s argument for vegetarianism rest (even ultimately) on intuitions about the nutritional value of a vegetarian diet? Does the well-known argument from special relativity against presentism rest on intuitions about whether special relativity is true? Nevertheless, Williamson does attribute it to many (unnamed) philosophers. And yet some of Williamson’s arguments seem directed particularly against this position. For instance, he says

“Taken far enough, the psychologisation of philosophical method becomes self-defeating”

And he cites approvingly Joel Pust’s conclusion that it is self-defeating to hold that

“Aside from propositions describing the occurrence of her judgements, S is justified in believing only those propositions which are part of the best explanation of S’s making the judgements that she makes”

Probably Pust and Williamson are right here, but it hardly tells against anything but a strawman version of the psychological view of evidence. Finally, Williamson objects to a version of Reflective Equilibrium that just attempts to get our intuitions into equilibrium with the following argument.

“The reflective equilibrium account, as usually understood, already assigns a proto-evidential role to at least one kind of non-psychological fact. For it treats philosophers as relying on logical relations between theories and intuitions, in particular their consistency and inconsistency.”

The theme again is that we need some evidence other than intuitions, something that should be common ground. (For reasons I’ll suggest in the next post, I’m not sure this is a good argument for that conclusion though.)

We haven’t got very far by trying to characterise what Williamson’s opponent says. Perhaps it is better to look at his positive proposal for what is evidence in philosophy. We get one statement of what that positive conclusion.

“Our evidence in philosophy consists of facts, most of them non-psychological, to which we have appropriate epistemic access.”

That, in conjunction with the quotes above, suggests he is defending the following three theses.

  1. Not all philosophical evidence is psychological.
  2. Having p be part of your evidence requires appropriate epistemic access to p.
  3. The intuition that p, or the fact that one has that intuition, is weak evidence, perhaps no evidence at all, that p.

A position that denied all three of those would clearly be among the targets of Williamson’s chapter. But that would be crazy, since (1) is obviously true. But a position like the one I sketched at the end of the last post, which was neutral on (2) and denied (3), would still seem to be at odds with the bulk of what Williamson says, and I think is meant to be among the positions ruled out by the considerations he raises.

Now such a position does not seem especially related to the sceptical positions that are the targets of sections 3 and 4 of Williamson’s chapter 7. But that’s as things should be. The question of what our evidence is doesn’t immediately settle the question of what knowledge we have. Taking perceptual evidence to be psychological might be a precursor to defending external world scepticism. But it might also be a precursor to adopting indirect realism. Of course Williamson, by accepting knowledge as evidence, has effectively ruled out classical forms of indirect realism, where we know about the world on the basis of purely phenomenal evidence. But that shouldn’t be presupposed here I think. A position that holds that psychological states, or facts about them, are often crucial evidence for us is opposed to the core doctrines of Williamson’s chapter, even if it is also opposed to some of his other opponents.

5 Replies to “Williamson on Evidence”

  1. I’m not sure that the view you say is crazy is quite as crazy as you suggest it is. One might think that “ultimately our evidence consists only of intuitions,” but think that Singer’s position has evidence even though we don’t have direct intuitions about the nutritional value of a vegetarian diet. After all, one might claim that even the empirical and scientific claims that enter into philosophical arguments depend on intuitions about how best to interpret perceptual evidence, and intuitions about the rationality of some form of inductive reasoning, and intuitions about which theories are the simplest explanations of various phenomena. I suppose one has to grant that at least some observations are also part of the evidence, in addition to intuitions, but one might still claim that every claim that isn’t directly given to us in perception ultimately has intuitions as evidence (perhaps along with some perceptual evidence).

    Now, this doesn’t strike me as an especially helpful way to think about things, but it also doesn’t seem as crazy as the claim that we have direct intuitions about the nutritional value of a vegetarian diet.

  2. Right, there is a position that says all of our evidence is ultimately psychological. We might think the evidence we have for claims about nutrition science ultimately rests on observations.

    But (a) I don’t think that means philosophical evidence is all intuitive. The conclusions, and not just the grounds, of work in nutrition science can play an evidential role for us. And (b) I don’t think all those observations are really intuitions. (As you say in the last line of the first paragraph.) Perhaps the point (a) here is arguable, but I think (b) is clear enough.

    So I don’t think the “all evidence is psychological states” position is crazy. Perhaps false, but not crazy. I do think the “all evidence is intuition-centered” is pretty crazy.

  3. I’m convinced by your examples that nobody thinks that ALL philosophical evidence is psychological. “Eating meat is not necessary for a healthy diet” is a counterexample.

    I guess I took Williamson to mean to be restricting his target to a particular subclass of philosophical premises — roughly, the class where there are philosophers running around saying things like “it is intuitive that such and such, therefore so and so”. The sort of cases Pust lists in chapter 1 of his book. The way to know that Gettier cases aren’t knowledge is to argue out from the premise that it is intuitive that they aren’t knowledge. The way to know that XYZ isn’t water is to argue out from the premise that it is intuitive that XYZ isn’t water. Etc.

    The opposing view is that we needn’t use these facts about intuitions as premises; we (can) come to acquire the evidence that XYZ isn’t water in some other way.

    The way I like to think about the issue is this: introspection is a skill; people can be better or worse at telling what their intuitions are. Suppose someone is really bad at introspecting; must he thereby be really bad at running this class of philosophical arguments? Williamson can say that he might nevertheless be very good at them, if he has the right sort of conceptual competences.

  4. I think the ‘intuitionist’, for want of a better word, shouldn’t accept as much of the Williamsonian setup as you’ve done here.

    So I agree in several cases our evidence includes both p and Intuitively, p. Not always, I’m going to say, but sometimes. (Actually I think this isn’t that common – Gettier cases are unusual in lots of ways. But let that slide.)

    But I also want to hold on to the following four claims.

    1. The evidence that p is intuitive is quite good philosophical evidence; it can seriously move the debate.
    2. Even when p is quite controversial, the intuition that p can be good evidence.
    3. When p is quite controversial (among respected philosophers at least), p is not part of our evidence, but Intuitively, p is.
    4. Cases like 3 are normal in philosophy; Gettier cases, where it’s more or less common ground what to say, are relatively rare.

    That sounds like a position that’s consistent with what you say, but also the kind of view Williamson is attacking. Perhaps this is wrong about the exegesis.

    The other thing is about the argument from introspection. I don’t think the intuitionist should accept, at least uncritically, the argument that we need to know p for it to be part of our evidence. Perhaps having mental state M is sufficient to make having mental state M part of our evidence. Maybe that’s false, but I don’t think it is obviously false. I think once you buy into knowledge constraints on evidence, you’re accepting a big chunk of the Williamson story that should be up for debate around here. And I think without buying that, the argument of the last paragraph isn’t that compelling.

  5. Yes, I see; E=K has been my operating assumption. If we relax that, then the picture might well end up pretty different.

    Incidentally, I’m not totally convinced that Williamson disagrees with your (1)—(4).

    (Edited by BJW to remove strikethrough!)

Leave a Reply