Skip to main content.
December 2nd, 2009

Evidence and Inference

My position on evidence is I think fairly similar to the position Clayton Littlejohn takes when he says evidence is non-inferential knowledge. I think, as I say in this old paper, that evidence is basically knowledge that is the output of a Fodorian module. The differences between our positions aren’t great. But I think there are some differences. This kind of case brings out some of these differences.

Graham, Crispin and Ringo have an audience with the Delphic Oracle, and they are told ¬ p ∨ q and ¬ ¬ p. Graham is a relevant logician, so if he inferred p ∧ q from these pronouncements, his belief in the invalidity of disjunctive syllogism would be a doxastic defeater, and the inference would not constitute knowledge. Crispin is an intuitionist logician, so if he inferred p ∧ q from these pronouncements his belief in the invalidity of double negation elimination would be a doxastic defeater, and the inference would not constitute knowledge. Ringo has no deep views on the nature of logic, but has accepted the classical theory he learned in an undergrad intro class because he doesn’t know there’s any dispute about it. Moreover, in the world of the story classical logic is correct. So if Ringo were to infer p ∧ q from these pronouncements, his belief would constitute knowledge. Now Graham’s and Crispin’s false beliefs about entailment are not p ∧ q-relevant evidence, and Ringo doesn’t have more evidence about logic than Graham or Crispin. So all three of them have the same p ∧ q-relevant evidence, but only Ringo is in a position to know p ∧ q.

This case is meant to do two things. First, it is an argument against a kind of evidentialism about knowledge. It isn’t true that what you know, or even what you’re in a position to know, supervenes on the evidence you have. Graham, Crispin and Ringo have the same evidence, but only Ringo is in a position to know that p ∧ q. That’s because Graham and Crispin’s false beliefs about entailment are defeaters in this context. In general, knowledge doesn’t supervene on evidence because defeaters don’t supervene on evidence.

The other thing it is supposed to do is draw out the idea that there is something problematic about treating logical knowledge as evidence. Ringo knows that ¬ p ∨ q and ¬ ¬ p entail that p ∧ q. Graham and Crispin don’t know this. But this isn’t an extra piece of evidence that Ringo has. Indeed, it isn’t an extra piece of evidence Ringo has for two reasons.

First, Graham and Crispin have all the evidence that Ringo has. They know that logic professors in intro classes say that ¬ p ∨ q and ¬ ¬ p entail that p ∧ q. They just don’t believe that these professors are right. This doesn’t seem like the kind of thing that destroys evidence.

Second, there are reasons tracing back to Lewis Carroll’s “Achilles and the Tortoise” for distinguishing between logical rules and logical axioms. Similar reasons suggest that we should distinguish between between empirical evidence and inferential rules that licence inferences from that evidence. That’s especially true when the inferential rules just are logical rules. So I don’t think that what licences Ringo’s inference of p ∧ q is an extra bit of evidence, it’s rather a rule of logic.

But Ringo knows that rule is correct. Indeed, it might even be a non-inferential piece of knowledge for him. (If need be, make Ringo smart enough in the example that he can simply see that certain inferences are valid.) Basic inferential rules aren’t things we know inferentially, they are things we use to get inferential knowledge. So if evidence is non-inferential knowledge, they are evidence. But I suspect they are not evidence for the reasons I’ve given here. So I think there’s a downside to the equation of evidence with non-inferential knowledge.

Posted by Brian Weatherson in Uncategorized

46 Comments »

This entry was posted on Wednesday, December 2nd, 2009 at 8:00 am and is filed under Uncategorized. You can follow any responses to this entry through the comments RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.

46 Responses to “Evidence and Inference”

  1. clayton says:

    In writing this I know I’ll regret it, but maybe my evidence that I’ll regret it is this and so I’ll say it anyway.

    Someone could say that one of the rationales for the idea that non-inferential knowledge is sufficient for evidence is the idea that any p we know non-inferentially is a starting point for enquiry that is proper in the sense that you’re not under some obligation to start with something more basic. So, even if there are things that are more epistemically secure than, say, some contingent proposition about my surroundings known to be true because of my perceptual experience, if those propositions are known non-inferentially, there’s no need to provide evidence for beliefs in those propositions and so the propositions believed can be evidence for without needing evidence for them in turn. So, one question is whether we can say something similar about what Ringo knows about the rightness of the rule. I think you’re right that the rightness of the rule won’t serve as a premise in an argument that uses that rule and had better not play the evidence role in reasoning that uses the rule, but might (here’s what I’ll regret saying) what Ringo knows non-inferentially serve as a premise in other reasoning where the right to treat that proposition as a premise doesn’t rest on some further bit of evidence Ringo has? RIngo hears Paul telling George that ¬ p ∨ q and ¬ ¬ p don’t entail q and so Ringo has some evidence that Paul isn’t great at logic. He hears John tell Yoko that ¬ p ∨ q and ¬ ¬ p do entail q and so given what he knows about logic, he has evidence for the hypothesis that John is a better logician than Paul. He has evidence that good logicians make good musicians, so he has evidence that John is better than Paul musically, etc…

    [I’m sort of groggy so I don’t expect to not get pummeled for that one, but I thought I’d float it to see what happens.]

  2. barrylam says:

    Brian, let me play the Harmanian here.

    Rules of logic are principles stating the facts about the consequence relation, they are not rules for proper reasoning. You can reason about logic, you can reasons about other things by recognizing or knowing about logic (the consequence relation), but the explanation of the rationality/normativity of such reasoning cannot be appealed to on the basis of the fact that such reasoning is an instance of “logical reasoning”. So the statement “what licenses Ringo’s inference of p and q is not an extra bit of evidence, its rather a rule of logic” at worst makes a category mistake, at best makes a substantive (but false) claim that the mere fact that p and q is a logical consequence of not p or q and not not p is what explains the rationality of Ringo’s inference.

    For such a Harmanian, if you are reasoning about other things by using your knowledge about the consequence relation, (as opposed to just reasoning about things, like simply inferring p and q from not not p and not p or q) then its hard to see how that knowledge isn’t evidence, supporting or defeating.

    On the other hand, if you are simply reasoning about things, then the proper explanation of why such reasoning is rational, or licensed, or any other normative feature, shouldn’t appeal to the fact that the reasoning is an instance of a “logical rule”, on pain of, once again, conflating principles of consequence with principles of reasoning.

    Many stories about what licenses Ringo’s inference are consistent with this type of Harmanian view. You can tell some kind of coherentist story, and say he is licensed in virtue of the fact that the piece of reasoning is not incoherence-generating, which explains Graham and Crispin’s unlicensed inference.

    I don’t know how this relates back to your original concern about evidence.

  3. Brian Weatherson says:

    Clayton,

    I think in these cases, you don’t need the logical claim as part of your evidence. The evidence that Paul isn’t great at logic is that he doesn’t believe that ¬ p ∨ q and ¬ ¬ p don’t entail q. Since that entails that Paul has a mistaken logical view about an entailment involving just two variables it entails that he isn’t great at logic. So the fact of Paul’s belief is enough I think.

    If you want to add in as extra evidence the fact that ¬ p ∨ q and ¬ ¬ p do entail q, I start to wonder (for familiar Carrollian reasons) why we should stop there. Why not add the extra fact that ¬ p ∨ q and ¬ ¬ p entail q, and Paul believes that ¬ p ∨ q and ¬ ¬ p don’t entail q, together entail that Paul has a false logical belief. Or the fact that the three claims mentioned in the previous sentence entail that Paul has a false logical belief. It seems that we either include none of these logical claims, we include infinitely many of them, or we make an arbitrary looking choice. I think including none of them is best.

    Barry,

    I was actually trying to be fairly sympathetic to the Harmanian line here. Let’s do the original argument Harman style.

    There are situations where an agent is entitled, given the state she’s in, to infer p. Question: Should we include as part of her evidence that she’s entitled, in that state, to infer p? I say we shouldn’t, even if she knows this is true. And I say this because it seems to confuse inferential propriety with things that are the proper basis of evidence, and making this confusion leads to infinite regresses of Carroll’s sort.

    Getting back to the original case, I thought it was perfectly consistent with what Harman said to say that in some cases, it is proper to infer p iff you know p is implied by some evidence. This isn’t always true, but there’s no reason it couldn’t actually be true in a particular case. If you think it is ever true, just redraw my example to the case you have in mind. I think cases where you are given some evidence by an unimpeachable source fall into this category, but you may prefer some other kinds of examples.

  4. barrylam says:

    Brian,

    I see now. The view you are advocating is that, two subjects, both of whom are entitled to infer that P, but one and only one of whom knows that she is so entitled, do not differ in the evidence they possess for p.

    I am sympathetic, but I have a question. How does the opposing position confuse inferential propriety with the proper basis of evidence, and run into the Carrollian regress problems? I take an opposing position to be that what explains the two subject’s entitlement to infer in both cases to be the same, but the presence of knowledge about the entitlement is further support for p for the subject who has such knowledge (a “higher-order evidence is evidence view”), just like higher-order defeaters can be defeaters for p.

  5. John Turri says:

    Hi everyone,

    For those attracted to the style of view Clayton and Brian find attractive, and more generally for those who find attractive the view that evidence is factive, I have a question.

    Some evidence is good evidence for Q, and some evidence is bad evidence for Q. Doesn’t it seem that one way for evidence to be bad is for it to be false?

    Suppose I offer evidence for Q. It goes like this, “If R, then Q. R. So, Q.” Someone might object to me, “R’s bad evidence [a bad reason] to believe Q, because R is false.” And that sounds like a fair objection.

    I suspect that people who say that evidence is factive are really offering a theory of good evidence.

  6. Brian Weatherson says:

    Barry,

    That’s pretty much the kind of example I had in mind, though it’s not exactly the same, and maybe the differences matter. I wanted a case where both of them had enough evidence to make the inference, but in one case the inference would have been not warranted because defeated by other, idiosyncratic, properties they have. In that case I don’t want to say that the extra entitlement of the first person comes from extra evidence, its simply from the absence of a defeater.

    The Carrollian regress I have in mind goes like this.

    Ringo: I’ve got evidence E, and anyone with evidence E can infer p, so I infer p.
    Me: That second conjunct looks dubious, are you sure it’s true.
    Ringo: Yep, I am, and you would be too if you thought about it.
    Me: Oh, you’re right. That’s very good. I agree with your claim about inference. Why not add it to your evidence after all, so your evidence is E + Anyone with E can infer p?
    Ringo: Good idea.
    Me: But I’m still not sure about the jump you’re making to p.
    Ringo: But if someone has E (as I do), and if anyone with E can infer p, then I can infer p.
    Me: Excellent. That’s true. Let’s add that to your evidence too. But something is still nagging me…

    You see how this goes. Eventually you have to move from evidence, and I treating the move as yet more evidence will invite a regress.

  7. Brian Weatherson says:

    John,

    But couldn’t that person also say that you have no evidence to believe that p?

    At some level I think we’re going to have to draw a bad evidence/no evidence line, just like we draw a bad joke/no joke line. “The cat is on the mat” isn’t a bad joke, it’s simply not a joke. Similarly, if I’m in my office with no mat and no cat, the proposition that the cat is on the mat isn’t bad evidence that there’s a cat in my office, it’s no evidence that there’s a cat in my office.

  8. John Turri says:

    Brian,

    Let’s take your case and extend it a bit. You believe that there’s an animal on an artifact because you (falsely) believe there’s a cat on a mat. Isn’t your evidence for believing “an animal is on an artifact” the belief or claim “a cat is on a mat”? If your evidence were true, then it would be good evidence. But it isn’t true, and in fact it’s bad evidence.

    I agree that it’s sometimes hard to draw the line between bad evidence and no evidence. But it seems unlikely, I think, that the correct way to draw the line will, in every case, leave all the false claims on the “no evidence at all” side of that divide. I also wonder, what theoretical advantage would we gain by calling it “no evidence” rather than “bad evidence”?

  9. clayton says:

    Hey John,

    “Doesn’t it seem that one way for evidence to be bad is for it to be false?”

    I don’t feel the pull. That someone smokes cigarettes is bad evidence for the hypothesis that they are a crook because it raises the probability of that hypothesis not at all, a very little bit, or lowers it. That a loving and all powerful God exists is not the sort of thing to be bad evidence for the hypothesis that all dogs go to heaven, not if God doesn’t exist. If I imaginatively project myself into the mindset of an atheist, I can’t bring myself to call it bad evidence. If I imagine myself into the mindset of someone who abandons theism, I wouldn’t redescribe what was once good evidence as bad. I could say ‘bad reason’, but that (I think) just means ‘bad thing to be moved by’, bad because it’s not evidence.

    Suppose someone believed that all contingent propositions they knew non-inferentially could be explained. Don’t worry why. They come to know p non-inferentially. They infer that there’s got to be some q such that ‘p because q’ is true. They know that ‘p because q’ is true only if p is true and q is true.

    I could see someone saying that not all contingent propositions known non-inferentially could be explained, but I can’t see any other mistake in that argument. So, I think that one reason to think that evidence consists of facts is that when you know p is part of your evidence, you know that p is constitutionally capable of figuring in infernece to best explanation and explanations are factive.

    Hey Brian,
    I have to respectfully disagree with this, ““The cat is on the mat” isn’t a bad joke, it’s simply not a joke.” People are looking at me right now trying to figure out why I’m laughing.

    I find myself agreeing with pretty much everything that you’re saying, but I’m still missing something key. (I might say something thick, so everyone but BW divert your eyes.) I don’t disagree with your claim that, “he fact of Paul’s belief is enough” to justify the relevant belief in your example. But, I thought the idea behind the example was not (a) the thing I’d classify as evidence isn’t evidence that’s needed to justify the belief but (b) the thing I’d classify as evidence isn’t evidence at all. That’s why I thought that the right response was to say that there are some beliefs that someone could use something they knew about logic as evidence for some belief even if they needn’t use that knowledge/evidence in a wide range of cases where the rule of inference they know to be correct is used. That’s why I didn’t think the regress got started. I thought that the regress got started only if you insisted that knowledge that the rule is a rule is treated as evidence or a premise in reasoning that also uses that rule. But, it seems that someone who identifies non-inf K with evidence only has to say that there’s some inference or other where the relevant proposition could be used to justify. I’ll have to think about whether I’d have to deny, “that the extra entitlement of the first person comes from extra evidence, its simply from the absence of a defeater.” I’d rather not deny it.

  10. John Turri says:

    Clayton,

    Suppose it’s true that not-P. Now suppose we have two people, Art and Betty. Art reasons like so:

    A1) (If P, then Q) & not-P.
    AC) So, not-Q.

    Betty reasons like so:

    B1) (If P, then Q) & P.
    BC) So, Q.

    Art’s premise is evidence, albeit bad evidence, for his conclusion. But if evidence is factive, then Betty’s premise isn’t bad evidence for her conclusion, because it isn’t evidence to begin with.

    I don’t understand why we’d want to say that. What do we gain? If we want to, for instance, prevent false claims from justifying beliefs, why don’t we just say that only true claims can be good evidence (and that only good evidence justifies)?

    It seems that Art and Betty both believe based on evidence, and their respective evidence is bad for different reasons. Betty’s failure is not that she simply failed to believe based on evidence.

  11. clayton says:

    Why are we assuming that Betty’s premise is evidence? Because she treats it as if it’s evidence? If that’s why, then I wonder why can’t someone treat something as if it is a reason when it’s not a reason? You can treat decoy ducks as if they are ducks but that doesn’t make them ducks. I think pieces of evidence are like ducks.

    “What do we gain?”

    Here are a few things. Maybe these answers aren’t perfect, but I think they are a start.

    (1) There’s the connection between evidence, truth, and inference to best explanation that you lose if you deny the factivity of evidence. If someone knows that p is part of her evidence, it seems that the question ‘Why is it that p’ is appropriate/in place/proper whenever p is part of the speaker’s evidence but that seems to assume that we’ll either say ‘No reason, it’s just a brute fact that p’ or ‘p because q’ but both entail p. How does someone who denies that evidence is factive account for the apparent fact that this question is always in place/proper/appropriate when all answers to that question entail, inter alia, p. To say ~p is to say that the question rests on a mistake.

    (2) It seems that if S knows that p is part of her evidence, she knows that p is true.

    (3) It seems that if A asserts that p is part of A’s evidence and then B asserts ~p, it seems that A and B disagree.

    (4) If p is part of my evidence and I know it, I think I’m in a position to A for the reason that p (when I know that my choice to A is a p-dependent choice) and I think you cannot A for the reason that p if ~p.

    (5) It just sounds weird to say, ‘His reason for thinking that was that p, but of course ~p’ I don’t think all (normative) reasons for belief are pieces of evidence, but I think all pieces of evidence are reasons for belief.

    My question is what do we lose by saying that evidence is factive (apart from the sympathy of many referees)?

  12. John Turri says:

    Clayton,

    If you want an argument, I’d suggest something like this: It’s her evidence, so it’s evidence.

    I think your answers constitute a fine start. Here’s what I have in response.

    I don’t think 2 sounds true.

    3 is interesting, but I don’t see how it supports the factive view of evidence. What do you mean by ‘disagree’?

    4 is ambiguous. When you say, “and I know it,” it could be understood as knowing p, or knowing that p is part of your evidence for believing something. Understood the former way, I think you get something in the neighborhood of a license to act on p. Understood the latter way, I don’t think you get anything of the sort.

    As for 5, consider: “His reason for accepting creationism is the claim that everything the Bible says is true. But of course that’s a bad reason [or: But of course he’s wrong about that] — not everything the Bible says is true.” Weird? Seems perfectly fine to me.

    I don’t fully understand 1.

  13. clayton says:

    Hey John,

    By ‘disagree’, I’d say that if I overheard A say that his evidence included that p, q, and r and I overheard B assert ~q, I’d say that they disagree about q. Maybe not face to face, but I don’t think that what they both said could be correct.

    On 4, make that p is part of my evidence and I know that p is part of my evidence.

    WRT to (5), you said, “His reason for accepting creationism is the claim that everything the Bible says is true.” To my ear, that’s really odd/off. “His reason for accepting creationism is that so and so claimed such and such” sounds fine, but “His reason for accepting creationism is that the Bible said that God made dinosaurs but the Bible never said that” sounds bad. Have to run, talk to you about this stuff later. I know that given whatever our evidence is.

  14. fitelson says:

    I tend to side with John (and other sometimes-evidence-is-non-factive folks) on this question. Here’s one oddity I see with the always-factive view. One way that evidence can be bad is if it’s misleading. And, one way (or so it seems to me) that evidence (E) can be misleading is by supporting (e.g., raising the probability/credibility of) of p, in a case where p happens to be false. Those who insist that evidence is always factive have to explain why this kind of misleading evidence ceases to be misleading EVIDENCE (because it ceases to be EVIDENCE at all) when E ENTAILS such a p (as opposed, e.g., to merely raising the probability/credibility of such a p). Compare the following two kinds of cases:

    (1) cases in which (a) the probability/credibility of p is raised from 0.001 to 0.999 by E (or by supposing that E), and (b) p is false.

    (2) cases in which (a) the probability/credibility of p is raised from 0.001 to 1 by E (or by supposing that E) — because (or, at least, in part, because) E ENTAILS p, and (b) p is false.

    In (1), always-factive people (I presume) have no problem with classifying E as evidence (per se). But, in (2), the always-factive folks must say that E cannot be evidence.

    That strikes me as an odd discontinuity. But, maybe I’m missing something here.

  15. Brian Weatherson says:

    Clayton,

    I think I see. I might have been sliding between the following two claims:

    A) When you use an inferential pattern you know to be sound, that’s always part of your evidence, so it must be included in a proper accounting of your thinking.
    B) When you use an inferential pattern you know to be sound, you can include that as part of your evidence, so it may be included in a proper accounting of your thinking.

    And it isn’t clear that (B) leads to a a regress.

    I think at that point I need to fall back to some intuitions about what is and isn’t evidence. I think we still get a neater accounting by not including facts about inferrability, but I might have been overstating the importance of regress arguments for that conclusion.

    Branden,

    I think we could live with the discontinuity, but I’m not sure we’re committed to it. Imagine that we’ve previously given credence 1 to the false claim ¬E ∨ H. Then we conditionalise on (true) E, and we gain credence 1 in the false claim H. In that case E is quite misleading evidence for H, but it is still true. So evidence can be extremely misleading, if you have the wrong priors.

  16. John Turri says:

    I think the factive crowd also needs to say something about cases involving knowledge from falsehood, which Warfield and Klein have discussed.

    One of Warfield’s cases: “CNN breaks in with a live report. The headline is ‘The President is speaking now to supporters in Utah’. I reason: ‘The President is in Utah; therefore he is not attending today’s NATO talks in Brussels’. I know my conclusion but my premise is false: the President is in Nevada — he is speaking at a ‘border rally’ at the border of those two states and the speaking platform on which he is standing is in Nevada. The crowd listening to the speech is in Utah.” (“Knowledge From Falsehood,” Phil Perspectives 19, 2005, p. 408).

    In cases where we know based on falsehood, do we know despite failing to base our belief on evidence? “The President isn’t in Brussels” doesn’t seem like the sort of thing you could know without basing it on evidence.

    (Knowledge-from-falsehood also poses a challenge for the view that good evidence must be true.)

  17. clayton says:

    Hey John,

    That’s an interesting challenge. In some ways, it is similar to some of Comesana and Kantin’s objections to E=K. They say that if E=K is true, so is:

    E=K1: The proposition that p justifies S in believing q only if S knows p.

    They then argue that there are two problems with E=K1. First, that it is incompatible with a plausible closure principle of justification. Second that (some) Gettier cases wouldn’t be Gettier cases if E=K1 were true.

    So far as I can tell, their objections to E=K assume:

    JE: The belief that p is part of what justifies S in believing that q only if p is included in S’s evidence.

    But, if JE is false we can accept E=K and say that the Gettier cases they like are Gettier cases. We can accept J-closure. I think their objections fail and so we need to distinguish between (i) p’s that are justifiably believed and play a deliberative role in deliberations that yield new justified beliefs and (ii) p’s that are included in the subject’s body of evidence and play the evidence role in justifying belief. Here’s why I think everyone should recognize that JE is false. In cases of justifiably believing p on the basis of inductive generalization are not cases where the hypothesis that there are unobserved counterexamples to the generalization is incompatible with the subject’s evidence prior to or subsequent to inferring p. When I know p on the basis of inductive grounds and deduce ~p — > q, I don’t think that this new belief could only constitute knowledge if p is included in my evidence. So, I think everyone needs to draw this distinction. The question is whether this is any use in addressing your example as well.

    We have cases of justified belief based on falsehood (e.g., false lemma involving Gettier cases), so someone who holds that all evidence is true could probably tell a story that is very similar to the story that others would tell. If JE is false, we all say that the subject’s belief that the President is not in Brussells is based on a further justified belief, the belief that the President is in Utah even if the proposition that the President is in Utah is not in the subject’s evidence. We can say that this false belief about Utah plays an essential role in justifying the true belief (e.g., it allows the agent to connect the evidence that supports the false Utah belief to the true Brussell’s belief) and does this without adding additional evidence into the mix beyond the evidence that supported the Utah belief.

    What turns that justified true belief about Brussells based on a justified false belief about Utah which is based on further justified true beliefs (where the contents of those beliefs are pieces of evidence that justify believing that the President is in Utah) into knowledge? Beats me. I just don’t see why I can’t steal whatever answer others would give.

  18. fitelson says:

    Thanks, Brian. I agree with your conclusion about your latest example, but I’m not seeing how it directly addresses the discontinuity I had in mind. The discontinuity I had in mind was that (from an always-factive perspective) some evidential E’s can very strongly support (or “nearly entail”) some false p’s, but no evidential E’s can entail any false p’s. If one is not already committed to factivity of evidence, then (it seems to me) one could find this puzzling. I guess I’m looking for a different explanation of this discontinuity — one that appeals to the nature of evidential support (perhaps a difference between inconclusive vs conclusive evidential support), rather than a direct appeal to factivity.

  19. Gabriele Contessa says:

    Hi Branden,

    Could you suggest a concrete example of a case in which we know that E and E (by itself) entails some false p? I can’t seem to think of any such case. As far as I can see, all cases in which evidence is misleading are case that are compatible with its factivity. (I think that we would not be inclined to say ‘The blood stains on the butler’s gloves misled the detective in believing he was the murderer’ if it turned out to be the cases the stains were ketchup stains…)

    (btw, congrats on your new job!)

  20. Gabriele Contessa says:

    (oops sorry for all the typos!)

  21. Gabriele Contessa says:

    Hi Brian,

    I think that the story is quite different in its original version and in its Harmanian variant. In the original you say that ‘in the world of the story classical logic is correct’, but that seems to suggest that one can have evidence that a certain logical system is correct at one’s world (what that evidence might be it’s a completely different question) and insofar as G’s, C’s, and R’s beliefs as to what the correct logic at their world are all epistemically justified an evidentialist will think their total evidence differs.

    If on the other hand you stick to the “Harmanian” version, it would seem that, evidentialists like Rich and Earl would maintain that, insofar as they have the same evidence, G, C and R are all in a position to believe that p ∧ q (that belief fits their total evidence) but only Ringo does believe so and so only he is epistemically justified in his epistemic attitude towards that p ∧ q.

    Am I missing something?

  22. fitelson says:

    Hi Gabriele,

    Sorry if I’m being unclear here. I think knowledge is factive. But, I don’t think all evidence must be known (neither does Brian, I take it). To try to put it more simply, the worry I have in mind is that the following two claims might seem to be in tension to someone who doesn’t already accept that evidence must be factive:

    (1) Some evidence can very strongly (in fact, arbitrarily strongly, so long as there isn’t a logical entailment) support some false claims.

    (2) No evidence can entail any false claims.

    The person I have in mind is someone (on some days, myself!) who doesn’t see why they should be committed to (2), in light of the fact that everyone accepts (1).

    Does that help to clarify the worry?

  23. fitelson says:

    Correction: I guess Brian does think all evidence must be known. He just thinks some things that are known are not evidence. Be that as it may, the worry I have in mind is orthogonal to this issue, and so can be stated in terms that don’t explicitly involve knowledge (as in my (1) and (2) above).

  24. Gabriele Contessa says:

    Hi Branden,

    Thanks, but I think I didn’t express my worry clearly. I was trying to suggest that even misleading evidence must be factive (the butler example) and I was curious to hear from you of a concrete example in which someone has a genuine piece of evidence that entails a false claim (without the aid of false auxiliary assumptions).

  25. fitelson says:

    Thanks, Gabriele. I’m only thinking about the tension between (1) and (2). I’m not saying that I think people should (necessarily) reject (2). To put it another way: if you think (2) must be true, then why not reject (1) as well? That is: in light of the fact that you accept (2), why not say that false claims can’t even constitute very strong evidence for false claims? This is the tension I have in mind (just put another way ‘round). Does that help?

  26. fitelson says:

    Sorry: I meant to say “why not say that we can’t have very strong evidence for false claims”, not just that we can’t have evidence that entails false claims?

  27. fitelson says:

    Another possible correction: in light of the fact that Clayton has a paper called “The Myth of the False, Justified Belief” (sorry — I just came across that title today), maybe NOT everyone would accept the claim that we can have very strong evidence (arbitrarily strong, so long as no entailment holds) for false claims. Is that right, Clayton? Would you reject (1) Clayton? And, could you send me a copy of that paper (“The Myth of the False, Justified Belief”)?

  28. Gabriele Contessa says:

    Hi Branden,

    I would say that any time evidence very strongly supports false beliefs it does so in conjunction with some false belief and cannot do so by itself. So, I can’t really see any tension between (1) and (2). In both cases, I would deny that E can very strongly support or entail a false belief by itself although it can do so in conjunction with some false beliefs.

  29. fitelson says:

    Thanks Gabriele. So, to be clear then, you agree with me that there is a tension between (1) and (2). You accept (2), and you reject (1) — at least, if both are interpreted as involving what E ALONE supports/entails. Right?

  30. fitelson says:

    One more thing, Gabriele. Let’s just think about cases in which E is one’s TOTAL evidence. That makes things clearer (for the tension I have in mind, at least).

  31. fitelson says:

    One final point. It seems to me that if one has (a) evidentialist leanings and (b) fallibilist leanings, then one should accept (1), when E is one’s total evidence. If this isn’t right, then what do we say about (e.g.) Newton. Did his total evidence not strongly support his theory (which is false)?

  32. Gabriele Contessa says:

    Hi Branden,

    Sorry if I my last comment was unclear, but that’s not what I meant. I meant that if we are talking about evidence alone I deny both (1) and (2).

    And I would be inclined to say the same even if the E represents my total evidence (at least if I have gathered enough evidence). I think it would be quite troublesome to assume that our total evidence can very strongly support a falsehood. Of course, this does not mean that evidence cannot support falsehoods to some degree. Only that there is no discontinuity from the entailment case to the very strong support case.

    Sorry, I gotta go now—I’ll comment on your point about Newton later, if you don’t mind.

  33. fitelson says:

    Hi Gabriele,

    I’m still not clear on your beliefs regarding (1)/(2). You say that you reject both of them, but that I don’t see how that can be right. If you reject (2), then (I presume) you accept its negation, which is:

    (not-2) Some evidence E can entail some false claims.

    But, you don’t accept that do you? I’m still puzzled. Sorry if I’m being thick here.

    Putting that to one side, I’m curious to hear more about your general view of evidence (and evidential support). I find it interesting to hold that evidence can never strongly support anything false. How “strong” is “strong”? Is there a threshold t at which support stronger than t for false claims is impossible? Can you say more about that? Also, I look forward to your remarks about Newton (which, I take it, will be related to this issue).

  34. Gabriele Contessa says:

    Branden,

    Coming to your Newton comment (#31). Did Newton’s total evidence provided him with very strong (absolute) support for his (false) theory? As far as I can see, all we can say is that his degree of belief in his theory given his total evidence should have been higher than .5 (which I’m sure it was) but I don’t see any reason to think that it should have been even remotely close to, say, .9.

    So, no, I don’t think his total evidence supported his theory very strongly in absolute terms. What I do believe is that it supported his theory very strongly in comparative terms (i.e. vis-a-vis the rival theories available to him).

  35. Gabriele Contessa says:

    Sorry, Branden – you were right! Of course, I do reject (1) and accept (2)! (I thought (2) both said ‘some evidence…’). And yes I do agree that there is a tension between them. Sorry for the confusion! But rejecting (1) does not seem to be a problem. One would hope that evidence (especially total evidence) would not support falsehoods very strongly. Isn’t that what’s good about evidence after all?

  36. Brian Weatherson says:

    This has spiralled on a bit while I’ve been away, but I wanted to respond to some questions Gabriele asks at #21.

    I was being a little flippant about classical logic being correct in the story. I didn’t mean to suggest that it could be true in some worlds false in others, or that it’s the kind of thing susceptible to being supported by evidence.

    In general, I think there’s a problem here for evidentialists. Let T be the correct theory of what can be inferred from what. You don’t need to know T in order to use it; at least you don’t except in cases where T says that a part of itself must be known to be used. If T says all of itself must be known to be used, then we get into Carrollian regresses, so that can’t be right.

    Assume it says that H can be inferred from E, where this is one of the inferences where you don’t have to know it is correct to properly make it. Assume that S believes that H cannot be properly inferred from E, but makes this inference anyway. I want to say that their claim to knowledge is defeated by this extra belief. (I’m neutral on whether their claim to justified belief is so defeated; it might not be.) If evidentialists deny this, so much the worse I say for evidentialists.

    The little example in the paper was meant to be using double negation elimination as an instance of something you can do without justification. But the example is less important than the principle.

  37. Brian Weatherson says:

    This is even further out of sequence, but it’s a response to John Turri’s question at #16.

    I think in all of those cases it is clear that there is some true evidence that’s the foundation. What happens in every case is that there is some evidence E (say a news report) which is excellent evidence for false p. Then p obviously entails true q. The speaker believes p on the basis of E, then infers q.

    I agree that these are cases of knowledge, and this is bad news for the no false lemmas crowd. But I don’t believe they are cases of false evidence. The evidence is E, not p. If there were a case where there was no true E to start with, then I’d be more impressed, but none of the cases I’ve seen fit that profile.

  38. Gabriele Contessa says:

    Hi Brian,

    Thanks for answering my question. Now I see more clearly what you meant, but I still don’t see why you think an evidentialist would have to deny that S’s claim to know that p can be defeated by some other unjustified belief of S, q, insofar as S’s belief that p fits S’s total evidence (and q doesn’t). (And, in any case, Earl and Rich, for example, distinguish between epistemic justification and well-foundedness and, I think, they would claim that the belief that p is justified though not well-founded).

    Moreover, I suspect there is still a tension between your story and evidentialism. You say ‘I didn’t mean to suggest that [classical logic is] the kind of thing susceptible to being supported by evidence’, but you also say ‘Assume that S believes that H cannot be properly inferred from E’. I have trouble seeing how there can be beliefs that are not susceptible to being supported by evidence (at least on an evidentialist framework). The evidence for some beliefs may be a priori evidence (such as “intuitings” or “seemings”) but insofar as you can have a belief about something it also has to be the case that belief either fits your total evidence or it does not, for there are no beliefs that are neither justified nor unjustified.

  39. clayton says:

    I wish I had said what Brian did at 37 rather than what I managed to say when I wrote 17, which wasn’t very clear.

    I’ve been thinking about Branden’s challenge and the evidence is piling up fast that there’s not much interesting I can say in response.

    A couple of points. Branden knows that I think there cannot be false, justified beliefs. (So does John, and it’s a testament to his character that we’re still philosophy pals and he doesn’t shun me in public given what he knows.) I wouldn’t want to argue from the fact that that there cannot be false, justified beliefs to the denial of Branden’s (1):

    (1) Some evidence can very strongly (in fact, arbitrarily strongly, so long as there isn’t a logical entailment) support some false claims.

    An example of (1). I know I’m a philosopher with a philosopher’s bank account and I know that I have a ticket for tomorrow’s lottery. I have evidence that provides exceptionally strong support for the proposition that I won’t be rich by Monday as my only chance for getting rich is winning the lottery. As the lottery gets bigger and bigger, the degree to which my evidence supports the hypothesis that I’ll still be poor on Monday will increase accordingly.

    Someone who says that there cannot be false, justified beliefs should say that the permission to believe p depends upon whether p (obviously) but doesn’t depend upon whether the believer has evidence that entails that the subject has a permission to believe p. (They should be fallibilist and anti-evidentialist.) In other words, they should say that the deontic properties of a belief does not supervene upon the believer’s evidence. [Suppose someone says, as Sutton does, that justified beliefs just are items of knowledge. It follows that there would be no false, justified beliefs. It does not follow, however, that if you justifiably believe p and you have evidence E that anyone who has evidence E and believes p justifiably believes p because it’s possible for someone to have E and not know p because someone could have E and yet ~p. Now, someone could object and say that since E=K this kind of case isn’t possible. But, I don’t accept E=K. Gettier cases and cases of inferential knowledge seem like counterexamples to E=K. Also, I don’t accept JB = K because of Gettier cases. So, if we retain the factivity of justified belief and weaken the requirements for justified belief, we can introduce a new range of externalist accounts of justified belief on which two subjects can have precisely the same evidence but only one subject has the permission to believe some proposition. (For the record, I think S’s belief is justified iff p and the subject reasonably take it to be the case that p or is personally justified in believing p.)]

    I take it that Branden’s challenge is so vexing because you can tell a (convoluted?) story about how (1) could be true even if evidence is factive (which is a separate issue from the factivity of justified belief) but then your views seem to be in tension because you accept (1) but have to reject:

    (2) No evidence can entail any false claims.

    Maybe the weirdness is this:

    If evidence is factive, it seems that the degree to which some arbitrary proposition can be supported by some body of evidence is going to be a function of the way the world happens to be in terms of its contingent features rather than by relations between the proposition and the evidence

    That can’t really be the weirdness, however, because if evidence is factive, I think you can’t have cases where there are two words, w1 and w2, where some subject in w1 has evidence E and P(p/E) = 1 and some subject in w2 has evidence E and P(p/E) < 1.

    Maybe the oddity is supposed to be this instead:

    If evidence is factive, it will be an empirical matter rather than an apriori matter whether some arbitrary proposition can be supported by entailing evidence or not.

    That can’t be the oddity, however, because someone who defends the factivity of evidence can place epistemic restrictions on what it would take to get something into your evidence such that you could know from the armchair whether some arbitrary proposition could be supported by entailing evidence or not. For example, if p could only be part of your evidence if p is knowable from the armchair or apriori, you could know whether some arbitrary proposition could be supported by entailing evidence or not without having empirical knowledge of the way things happen to be.

    That’s a tad disingenuous, you might say, because I think that (i) p can be part of your evidence only if p, (ii) whenever you know p non-inferentially p is part of your evidence, and (iii) the scope of non-inferential knowledge includes more than just that which we can know from the armchair. I think this means that if you accept the factivity of evidence and don’t limit evidence to that which can be known apriori or from the armchair you’ll have to just bite this bullet:

    If evidence is factive, you cannot know from the armchair or know apriori whether some arbitrary proposition could receive support from entailing evidence or not until you know what your evidence is our could be (in your world, depending upon what you come to know later) and that kind of knowledge will be empirical knowledge.

    This might be a bullet that you shouldn’t look forward to biting, but you could say that it’s not the worst thing in the world. In conceding this, you can still say that it isn’t a contingent matter whether some p could receive support from entailing evidence or not from some particular body of evidence and it is still an apriori matter whether some particular body of evidence could provide entailing support for the propositions it provides that support for.

    I think that’s the best that someone can do if they think that evidence is factive. Try to limit the extent to which empirical knowledge infects knowledge concerning evidential relationships and show that you can accommodate some of the intuitions that might (this is a guess) lead someone to think that accepting (1) and rejecting (2) creates a tension we shouldn’t.

  40. Gabriele Contessa says:

    Sorry I meant to write:

    ‘I still don’t see why you think an evidentialist would have to claim that S’s claim to know that p can be defeated by some other unjustified belief of S’

  41. Gabriele Contessa says:

    Even better:

    ‘I still don’t see why you think an evidentialist would have to affirm (or deny) that S’s claim to know that p can be defeated by some other unjustified belief of S’.

    After all evidentialism is primarily an account of epistemic justification not knowledge and an evidentialist would seem to be able to accept that something can defeat a justified true belief’s claim to constitute knowledge.

  42. Brian Weatherson says:

    Maybe that’s right, and this isn’t a problem for evidentialism as such. One could hold that

    (a) When you have false beliefs about what can be inferred from what; and
    (b) You make inferences that accord with the true theory of inference, but not with your beliefs, then

    you end up with justified true beliefs that are not knowledge.

    This doesn’t seem like a particularly happy result to me, but it’s certainly consistent if one wants to hold on to evidentialism about justification.

  43. fitelson says:

    Thanks for the clarification, Gabriele.

    And, thanks Clayton for the very helpful and informative reply. That’s just the sort of thing I was yearning for! I’ll have to think a lot more about all of this — but this thread has been quite useful for me, so thanks all!

  44. Gabriele Contessa says:

    Brian,

    What I think evidentialists should find really problematic by their own light is that the distinction between the EJ and the well-foundedness of a belief on which this strategy relies seems to undermine the internalist credentials of the theory. Isn’t it?

    Also, throughout this thread, you seem to suggest that there can/might be no evidence for or against the correctness of a certain logical system (or at least that you doubt there can be any such evidence). Is that right? If so, when and if you have a chance, I’d be interested to hear more about why you think so. Your Crispin, Graham, and Ringo seem to have different beliefs about what the correct logical system is and, presumably, each has his reasons for believing what he believes (or at least their well-known real world counterpart do). And, also presumably, these reasons can be good or bad reasons, right? But then why think that they cannot have evidence for or against the correctness of a certain logical system? Is it because you think that evidence is always empirical evidence and that there can’t be empirical evidence for or against the correctness of a logical system?

  45. Gabriele Contessa says:

    In light of Clayton’s lottery argument, I would have to restate my position wrt (1) and (2) more carefully. I guess that what I would want to say is that both (1) and (2) are true but the higher the degree of support in question the less likely it is that the evidence supports a false proposition to that degree.

    So, for example, the more tickets there are in Clayton’s lottery (and he knows that) the more his evidence supports the proposition that he won’t win. However, the more tickets there are in his lottery, the more unlikely it is that he will in fact win. Now keep the total number of tickets fixed and increase the number of tickets he bought to 1/10 of the total tickets. His evidence supports the assumption that he won’t win somewhat less strongly than it used to when he only had one ticket. Decrease the number of tickets. The support increases. When you get to zero tickets his evidence entails that he won’t win. So, Branden’s discontinuity disappears despite the seeming tension between (1) and (2). Does this make sense?

  46. Fritz Warfield says:

    “Brian Weatherson says:

    This is even further out of sequence, but it’s a response to John Turri’s question at #16.

    I think in all of those cases it is clear that there is some true evidence that’s the foundation. What happens in every case is that there is some evidence E (say a news report) which is excellent evidence for false p. Then p obviously entails true q. The speaker believes p on the basis of E, then infers q.

    I agree that these are cases of knowledge, and this is bad news for the no false lemmas crowd. But I don’t believe they are cases of false evidence. The evidence is E, not p. If there were a case where there was no true E to start with, then I’d be more impressed, but none of the cases I’ve seen fit that profile.”

    In some of the knowlege from falsehood cases, as I recall anyway, the agent doesn’t even believe the nearby truth that would be a good epistemizer if she believed it. So I don’t know why one would be so confident that in the cases the agent’s knowledge is based on the unbelieved truth rather than the believed falsehood.

    And as I and others (Klein I believe) have argued. Optimistic declarations about the basing relation in such cases will overascribe knowledge in clear no-knowledge Gettier cases. Hmm, Brian may not classify the Gettier cases in the same way I do…but I’ll leave that aside.

Leave a Reply

You must be logged in to post a comment.