PHIL2606: Knowledge, Reason and Action
Week 4: Two alternative approaches to knowing

    A Causal approach to knowing

  1. Here is an idea that some people like (e.g. Goldman, A. I. (1967), ‘A Causal Theory of Knowing’, Journal of Philosophy 64, pp. 357-72). In the ten coins case, the reason why Smith does not know that the man who will get the job has ten coins in his pocket is this:

    The fact that Smith believes that the man who will get the job has ten coins in his pocket was not caused by the fact that it is true that the man who will get the job has ten coins in his pocket.

  2. The idea is that there is an additional condition on knowing:

    Necessarily: for all s and p: if s knows p then the fact that s believes p was caused by the fact that p is true.

  3. We might then formulate the following causal account of knowing:

    CTB: Necessarily: for all s and p: s knows p iff (a) p is true, (b) s believes p, and (c) the fact that s believes p was caused by the fact that p is true.

  4. Some observations

  5. CTB assumes that facts can be the relata of causation (i.e. can cause and be caused). This is somewhat controversial. Some argue that only events can cause and be caused. If so, then perhaps we can formulate CTB in terms of event causation instead:

    Necessarily: for all s and p: s knows p iff (a) p is true, (b) s believes p, and (c) the event of s coming to believe p was caused by the event of p becoming true.

  6. CTB can be understood as allowing that a fact might have more than one cause: it requires that the fact that s believes p was caused by the fact that p is true; this allows that it might have been caused by other facts as well (e.g. every contributing fact).
  7. CTB seems to be a departure away from the idea that justification is necessary for knowing: there is no explicit mention of justification in the account.

    Or maybe not: maybe if the fact that s believes p was caused by the fact that p is true then ipso facto (i.e. thereby) s counts as having a certain sort of justification for believing p. We might call it 'causal justification' (or something like that).

    (This would be an externalist kind of justification - justification that s can have without knowing that she has it.)

  8. Testing the account

  9. We have seen how CTB handles the ten coins case. How does it handle the other cases we have seen so far?
  10. A problem case

  11. We seem to have some mathematical knowledge: I know that 2 + 2 = 4. But does CTB allow this?
  12. The concern is that the fact that it is true that 2 + 2 = 4 cannot have caused the fact that I believe that 2 + 2 = 4. Why not? Perhaps it is easier to see why not in the event-based version of CTB: is there such an event as the event of it becoming true that 2 + 2 = 4?
  13. In response to this problem, Goldman proposes CTB only as an account of what it is to know certain things - empirical truths, for example. He proposies that JTB is fine for non-empirical truths.
  14. Is he thus committed to saying that 'know' is ambiguous? If so, is that a problem? Or can he say that knowing is a disjunctive relation? If he does, is that a problem?
  15. Another problem case

  16. I know that smoke is coming out of my chimney, because I can see that my fire is burning. But was the fact that I believe that there is smoke coming out of my chimney caused by the fact that it is true that there is smoke coming out of my chimney? If not, then we have a problem for CTB.
  17. One way to avoid this problem might be to modify CTB slightly (Goldman does this):

    CTB2: Necessarily: for all s and p: s knows p iff (a) p is true, (b) s believes p, and (c) the fact that s believes p is causally connected to the fact that p is true.

    In the chimney case, the fact that I believe that smoke is coming out of my chimney is, if not caused by it, at least causally connected to the fact that it is true that smoke is coming out of my chimney - they are connected by a common cause, the fact that my fire is burning.

  18. Testing the account

  19. But wait, now that we have modified CTB we should check that it still works in the various cases:
  20. Another problem case

  21. Here is a problem case for CTB2 (a variation of an example from Goldman):

    The newspaper case
    Scoop observes that Speedy won Race, and writes an article for the Daily Advertiser. But there is a typo, and it is printed that Speedy did not win Race. But Blurry misreads the Daily Advertiser as saying that Speedy won Race, and comes to believe that he did. So (a) it is true that Speedy won Race, (b) Blurry believes that Speedy won Race, and (c) the fact that Blurry believes that Speedy won Race is causally connected to the fact that it is true that Speedy won Race, but (d) Blurry does not know that Speedy won Race.

  22. We could get a similar case by modifying the original ten coins case. Let's add to the story that the reason why Smith's boss told Smith that Jones will be getting the job is that he knew that the man who will get the job has ten coins in his pocket, and in such cases (for some strange reason!) he always tells Smith that he will get the job. Then the fact that Smith's belief that the man who will get the job has ten coins in his pocket is causally connected to the fact that it is true that the man who will get the job has ten coins in his pocket. But Smith still does not know.
  23. Goldman accepts that this kind of case is a counterexample to CTB2, and he makes a modification:

    CTB3: Necessarily: for all s and p: s knows p iff (a) p is true, (b) s believes p, and (c) the fact that s believes p is causally connected in an appropriate way to the fact that p is true.

    Not any old causal connection is enough to yield knowledge - it has to be one of an appropriate kind. The kind in the newspaper case is not appropriate.

  24. How does one get causally connected in an appropriate way? By processes, Goldman claims, that include the following:
  25. But is CTB3 now immune from counterexamples? Suppose we propose a case in which (a) some proposition, p, is true, (b) someone, s, believes p, (c) the fact that s believes p is causally connected in an appropriate way to the fact that p is true, but (d) s does not know p. Can't Goldman just reply that all this case shows is that the causal connection is not, after all, appropriate? Recall: it's up to him to tell us more about what he means by 'appropriate' - it is not something that we have any pre-theoretic intuitions about. But then it will be impossible for anyone to come up with a counterexample.
  26. This is a vice and not a virtue of CTB3. It raises the concern that what the 'if' part of CTB3 is saying is this: such-and-such is sufficient for knowing, unless it's not. That's true! But it tells us nothing about knowing.
  27. Yet more problem cases

  28. That concern aside, there are problem cases for CTB3:

    The ten coins with discovery case
    Smith later discovers that it is he who will get the job and that he also has ten coins in his pocket, and so comes to know that the man who will get the job has ten coints in his pocket. But this all makes no difference to the causal history of his forming the belief that the man who will get the job has ten coins in his pocket, so if the fact that he believes that the man who will get the job has ten coins in his pocket was not causally connected in an appropriate way with the fact that it is true that the man who will get the job has ten coins in his pocket, then it still is not.

    The eruption case (modified from Dretske 1971):
    Tom believes, truly, that mountain A erupted, his justification being that there is solidified lava all around it. But not far from mountain A there is mountain B, such that if mountain A had not erupted then mountain B would have erupted instead, producing the same pattern of solidified lava. According to CTB3, Tom knows that mountain A erupted, because the fact that he believes that mountain A erupted is causally connected in an appropriate way with the fact that mountain A erupted. But Tom does not know.

    The BIV case (Nozick):
    Mary is a brain in a vat, being stimulated by scientists to believe that she is a BIV being stimulated by scientists. According to CTB3, Mary knows that she is a BIV being stimulated by scientists, because the fact that she believes that she is causally connected in an appropriate way with the fact that she is. But Mary does not know.

  29. Undefeated justification

  30. Now for another approach to knowing
  31. According to the no false lemmas account of knowing, the reason, in the ten coins case, why Smith does not know that the man who will get the job has ten coins in his pocket is that his justification includes something false (that Jones will get the Job).
  32. There is another line of thought, according to which the problem is with what Smith does not include in his justification: there are facts not included in his justification which defeat his justification.

    (See Lehrer, K. and Paxson, T. (1969), ‘Knowledge: Undefeated Justified True Belief’, Journal of Philosophy 66, pp. 225-37. See also an earlier version of their account: Lehrer, K. (1965), ‘Knowledge, Truth, and Evidence’, Analysis 35, pp. 168-75. Also Chisholm, R. (1966), Theory of Knowledge (Englewood Cliffs, NJ: Prentice-Hall), p. 48.)

  33. Consider the ten coins case. Simth's justification for believing that the man who will get the job has ten coins in his pocket is that Jones will get the job and Jones has ten coins in his pocket.

    The are plenty of facts not included in this justification: that Baton Rouge is the capital of Louisiana, that every water molecule contains an Oxygen atom, and so on. Most of these do not defeat his justification: if we add it to his justification then he will still be justified.

    But there is at least one of them that does defeat his justification: the fact that Jones will not get the job. If we add this to his justification then he will no longer be justified. This is a defeater of his justification, and his justification is defeated. That's why Smith's justification is not enough to yield knowledge.

  34. The idea is that there is an additional condition on knowing, the no defeat condition:

    Necessarily: for all s and p: if s knows p then s's justification for believing p is not defeated.

  35. What is it for a justification to be defeated?

    Suppose that s is justified in believing p. Her justification is defeated iff there is a true proposition d such that if d is added to s's justification then s is no longer justified in believing p.

  36. We might then formulate the following undefeated true belief account of knowing:

    UTB: Necessarily: for all s and p: s knows p iff (a) p is true, (b) s believes p, (c) s is justified in believing p, and (d) s's justification for believing p is not defeated.

  37. Testing the account

  38. How does UTB handle the various other cases that we have considered?
  39. Some observations

  40. Do we need the truth condition in UTB? It seems not: one cannot have undefeated justification for any false proposition. For suppose that s is justified in believing p, where p is false. Then not-p is a defeater of s's justification, since adding not-p to s's justification for beleiving p takes away her justification for believing p.

    So the truth condition on knowing follows from the no defeat condition on knowing, and we can remove it from the formulation of UTB.

    Note: UTB thus seems to give an explanation of why knowledge is factive, which is a mark in its favour.

  41. Note that any justification which includes a falsehood (a false lemma) is defeated. For suppose that s is justified in believing p, where her justification includes some falsehood, q. Then not-q is a defeater of s's justification, since not-q is true and adding not-q to s's justification for beleiving p takes away her justification for believing p.

    So the no defeat condition on knowing entails the no false lemmas condition on knowing.

    But the no defeat condition is a stronger condition than the no false lemmas condition - one's justification can be defeated even when it contains no false lemmas. For this reason, UTB avoids some of the problems that beset the no false lemmas account.

  42. A counterexample?

  43. Can we come up with a counterexample to UTB?

Exercise

Read Lehrer, K. and Paxson, T. (1969), 'Knowledge: Undefeated Justified True Belief', Journal of Philosophy 66, pp. 225-37. Their final account differs in three ways from the account we have looked at here:

Pick one of these modifications, explain in more detail what it is, and explain why L & P make the modification.