PHIL2606: Knowledge, Reason and Action
Week 9: Semantic Externalism; Relevant Alternatives

    Rejecting the first premise: semantic externalism

  1. Hilary Putnam has suggested a different way to reject the first premise. He argues in a more sophisticated way that we do know that we are not BIVs (or at least can come to know via the argument that he gives). (see Putnam, H. (1981), ‘Brains in a Vat’, in his Reason, Truth and History (Cambridge: CUP)).
  2. Putnam argues that whether or not we are BIVs, the sentence ‘I am a BIV’ expresses a proposition that is false.
  3. Semantic externalism

  4. His argument relies on a view called semantic externalism.
  5. According to this view, what our words refer to depends upon what it is in the world that they are causally connected to.
  6. Consider twin-Earth, a planet exactly like ours in every respect, except that the watery stuff on twin-Earth is XYZ rather than H20. When twin-Earthlings use the word 'water' they are referring to XYZ rather than H20 ('water' in their mouths refers to XYZ, 'water' in our mouths refers to H20). This is so, even though there is absolutely no internal difference between them and us (hence the term 'semantic externalism').
  7. Putnam's argument

  8. Consider a BIV using the word 'brain'. The word ‘brain’ in the mouth of a BIV does not refer to brains, because a BIV is not appropriately causally connected to brains. This is so, even if from the inside things seem the same to BIVs as they do to non-BIVs.
  9. To what does ‘brain’ refer in the mouth of a BIV? Whatever stands in the appropriate causal connection to the BIV’s use of the word, perhaps:

    Let’s say that it refers to computer subroutines - call them 'brain-subroutines'.

  10. Similarly:
  11. Since a BIV is not a brain-subroutine in a vat-subroutine, the sentence ‘I am a BIV’ in her mouth expresses a proposition that is false.
  12. Now consider a non-BIV (let's say someone who is in the situation that we think we are in). The word 'brain' in her mouth refers to brains, and the word 'vat' in her mouth refers to vats, and the sentence ‘I am a BIV’ in her mouth expresses the proposition that she is a BIV. Since she is not a BIV, the sentence ‘I am a BIV’ in her mouth expresses a proposition that is false.
  13. So, no matter whether a speaker is a BIV or a non-BIV, when she utters ‘I am a BIV’ she expresses a proposition that is false.
  14. In particular, when I utter ‘I am a BIV’ I express a proposition that is false. So when I utter ‘I am not a BIV’ I express a proposition that is true.
  15. Moreover, I know that I express a proposition that is true (having understood and accepted Putnam’s argument), so I can confidently assert that I am not a BIV!
  16. We can summarise the argument as follows:
    1. Either I am a BIV or I am a non-BIV.
    2. If I am a BIV then:
      1. My utterances of ‘I am a BIV’ are true iff I am a brain-subroutine in a vat-subroutine.
      2. I am not a brain-subroutine in a vat-subroutine.
      3. So my utterances of ‘I am a BIV’ are false.
    3. If I am a non-BIV then:
      1. My utterances of ‘I am a BIV’ are true iff I am a brain in a vat.
      2. I am not a brain in a vat.
      3. So my utterances of ‘I am a BIV’ are false.
    4. Either way, my utterances of ‘I am a BIV’ are false.
    5. So my utterances of ‘I am not a BIV’ are true.
    6. So I am not a BIV.
  17. Problems for Putnam

  18. Suppose Putnam is right, that my utterances of 'I am not a BIV' are true, and that I know this on the basis of his argument. Is this any consolation? Here's a problem: although I know that what I am saying is true, I don't know what I am saying (because I don't know whether I am a BIV or a non-BIV). It is no more consolation than confidently asserting, when lost in the bush, I am here! So we are still left wondering whether or not we know that we have hands.
  19. Note also that Putnam's argument only applies to a particular BIV scenario - one in which the BIVs have come to acquire their words while in the vat. What about a scenario in which I am a recently envatted brain, having acquired my words and their referents in the usual way while outside the vat. Then my utterances of 'I am a BIV' turn out to be true, because I am referring to brains and vats, rather than brain-subroutines and vat-subroutines.
  20. Rejecting the second premise: relevant alternatives

  21. What about rejecting the second premise of the BIV argument, that if I do not know that I am not a BIV then I do not know that I have hands?
  22. Some people have argued that we should. They argue that I can know that I have hands, without knowing that I am not a BIV (even if I know that my having hands entails my not being a BIV)(recall: BIVs are handless, by definition).
  23. Most notable are Fred Dretske (Dretske, F. (1970), ‘Epistemic Operators’, Journal of Philosophy 67, pp. 1007-23; Dretske, F. (1971), ‘Conclusive Reasons’, Australasian Journal of Philosophy 49, pp. 1-22) and Robert Nozick (Nozick, R., Philosophical Explanations). Also see Goldman, A. (1976), ‘Discrimination and Perceptual Knowledge’, Journal of Philosophy 73, pp. 771-91.
  24. Closure

  25. If we reject the second premise then we have to reject the epistemic closure principle, the principle that knowledge is closed under known entailment:

    Closure: Necessarily: for all s, p and q: if s knows p, and s knows that p entails q, then s knows q.

  26. Why? Because if Closure is true then the second premise is true; so if the second premise is false then Closure is false (by contraposition).

    Here's a proof that if Closure is true then the second premise is true. Suppose that Closure is true. Suppose I know that I have hands. In such a case I also know that if I have hands then I am not a BIV (BIVs are handless, by definition). But then, by Closure, I know that I am not a BIV. So if I know that I have hands then I know that I am not a BIV; or, by contraposition, if I do not know that I am not a BIV then I do not know that I have hands. That is, the second premise is true. So if Closure is true then the second premise is true.

  27. Closure is a very plausible principle (that's why we find the second premise of the BIV argument so compelling). We seem to appeal to it every time we expand our knowledge by inference. I weigh myself and come to know that I am 70kg; since I know that being 70kg entails being less than 76kg, I thereby come to know that I weigh less than 76kg. Who would want to reject this principle?
  28. Note that Closure is similar to, but importantly different from, the following principle:

    Closure+: Necessarily: for all s, p and q: if s knows p, and s knows that p entails q, then s knows q.

  29. Because it is so plausible, it is no good to simply reject Closure - reasons must be given. And we need a better reason than this: by rejecting Closure we can avoid accepting the conclusion of the BIV argument - such a response to the BIV argument is no more satisfying than the Moorean response. What we need is an explanation for why Closure is false, and why we all (or most of us) mistakingly think that it is true. Dretske attempts to do this.
  30. Dretske accepts that something like Closure is true for other things. He accepts the following principles:
  31. He even accepts that something like the stronger Closure+ is true for these things. He accepts the following principles:
  32. But he rejects Closure. Why the difference between being true, being necessary, being possible and knowing?
  33. Dretske's approach: relevant alternatives

  34. Dretske's view is something like this. The following two sentences each express the proposition that Lefty killed Otto, but each also determines a set of relevant alternatives:

    (An alternative to the proposition that Lefty killed Otto is any proposition that is incompatible with it.)

  35. They differ in which alternatives are relevant in each case:
  36. Note that all of these propositions are alternatives to the proposition that Lefty killed Otto: that Alan killed Otto, that Brian killed Otto, that Charlie killed Otto, that Lefty killed Alan, that Lefty killed Brian, that Lefty killed Charlie, and so on. But only some of these are relevant to each sentence.
  37. Some operators are not sensitive to this difference:

    (An operator is an expression that operates on a sentence to produce a new sentence. E.g. 'It is true that ...': this operates on the sentence 'grass is green' to produce the sentence 'It is true that grass is green'.)

    Dretske calls these operators fully-penetrating.

  38. Some operators are sensitive to this difference:

    These operators are not fully-penetrating.

  39. In particular, the operator 'so-and-so knows that ...' is sensitive to this difference:
  40. And in the case of most interest to us:
  41. Why is knowing sensitive to this difference? Because to know p you have to know that the relevant alternatives are not true, but you do not have to know that any irrelevant alternatives are not true. So:

    (Maybe rather than talking about knowing that the relevant alternatives are false we should talk about being able to rule them out or about having evidence against them. Why? Concerns about circularity. But maybe this does not help - maybe there is a problem here for Dretske.)

  42. In the case of interest to us:
  43. Other examples:
  44. Note that being a BIV is an alternative to having hands (they cannot both be true), so having hands entails not being a BIV. But I can know that I have hands without knowing that I am not a BIV, even if I know that having hands entails not being a BIV. Why? Because being a BIV is not a relevant alternative.
  45. A different closure principle

  46. If Closure is true then we have the following:

    You cannot know p unless you know that all known alternatives to p are false.

  47. Dretske rejects this result, but endorses the following one:

    You cannot know p unless you know that all relevant known alternatives to p are false.

  48. So he proposes that we reject Closure but endorse the following weaker principle:

    Closure-: Necessarily: for all s and p: if s knows p and s knows that p entails q and q is relevant then s knows q

  49. A challenge

  50. Challenge for Dretske: what makes some alternatives relevant and others not? Dretske: “A relevant alternative is an alternative that might have been realized in the existing circumstances if the actual state of affairs had not materialized.” Goldman: a relevant alternative is one that we have some reason to think is true.
  51. Appealing to an account of knowing

  52. Dretske backs up his idea that Closure is false with an account of knowing according to which it is false (see his 'Conclusive Reasons'). Nozick's account is also one on which Closure is false.
  53. Nozick:

    Necessarily: for all s and p: s knows p iff (a) p is true, (b) s believes p, (c) if p were not true then s would not believe p, and (d) if p were true then s would believe p.

  54. This account allows that I can know that I have hands without knowing that I am not a BIV, even while knowing that if I have hands then I am not a BIV.
  55. How? Suppose I am not a BIV, and am in fact in the situation that I think I am in. Let p be the proposition that I have hands. Then p is true, I believe p, if p were not true then I would not believe p, and if p were true then I would believe p. So I know p - I know that I have hands. Let q be the proposition that I am not a BIV. Then q is true, I believe q, if q were not true then I would still believe q. So I do not know q - I do not know that I am not a BIV.
  56. Note that Nozick's account does not declare that we know that we have hands. It just declares that if we are not BIVs then we know that we have hands.
  57. Abominable conjunctions

  58. Here is what many think is a decisive objection to Dretske and Nozick: their accounts allow there to be true abominable conjunctions: "I know that I have hands, but I do not know that I am not a (handless) BIV."
  59. For other criticisms of the relevant alternatives approach, see Sosa, E. (2003), ‘Relevant Alternatives, Contextualism Included’, Philosophical Studies 119, pp. 3-15; Vogel, J. (1999), ‘The New Relevant Alternatives Theory’, Philosophical Perspectives 13, pp. 155-80; De Rose, K. (1995), ‘Solving the Skeptical Problem’, Philosophical Review 104, pp. 1-52.

Further reading

Exercise

Read Nozick, R. (1981), Philosophical Explanations (Cambridge, Mass.: Harvard University Press), 'Nonclosure'. Why does Nozick think that knowledge is not closed under known implication?