Wednesday, June 18, 2014

Two Interesting Comments
In 1976, when I delivered the John Locke Lectures at Oxford, I often spent time with Peter Strawson, and one day at lunch he made a remark I have never been able to forget. He said, "Surely half the pleasure of life is sardonic comment on the passing show".  This blog is devoted to comments, not all of them sardonic, on the passing philosophical show.
Hilary Putnam

A reader of this blog (“undergrad student”) writes “…it is unfair to attribute to Carnap the view that analytic sentences are unrevisable.”  That is true. But I did not talk about sentences but about statements, including ones regarded as “definitions”. Yes, Carnap would allow us to negate or otherwise revise any sentences we might write down, provided we say that, in the case of sentences he regarded as analytic, we have changed the meaning of the sentences in question (so they no longer express the same statements). My objection (in The Analytic and the Synthetic”) was different from Quine’s.  Quine regarded the notion of “meaning change” as unintelligible (although, when he wrote "Two Dogmas", he did not yet regard reference as indeterminate). I do not. But I argued (see the section titled “Analytic and Nonanalytic Statements”) that when Einstein replaced the Newtonian (actually post-Newtonian) definition of kinetic energy, e = ½ mv2, with  e = m + ½ mv2 + 3/8 mv4 +  . [choosing units so that the speed of light =1], the meaning has not changed enough to effect what we are talking about.(see pp. 52-53 of my Mathematics, Matter and Method.)  Even if one regards only the terms in this power series after "m" as corresponding to what is properly regarded as kinetic energy, the remainder—½ mv2 + 3/8 mv4 +  — is only approximately equal to ½ mv2, and then only at non-relativistic velocities. If we say that we changed the meaning of kinetic energy, we had better also claim that the meaning change was great enough to affect the reference (the Bedeutung) of the term "kinetic energy" (otherwise the meaning change would be irrelevant to the truth-value of the statement). But if we did claim that, we would be wrong: the physical magnitude we referred to as "kinetic energy" never equaled one half m times velocity squared; it always equaled what Einstein said, and we discovered that the very statement we thought was analytic was false, and not just that we wish to write down a different sentence.
 Charles Pigden, who is a master of this subject, points out that Quine uses the expression "unit of empirical significance" in "Two Dogmas".  But the use of the term "empirical significance" (or "empirical meaning", or "empirical import") of a theory to refer to the theory's observational consequences is widespread in philosophy of science, and its use does not indicate that the writer thinks that empirical meaning = linguistic meaning. In "The Problem of Meaning in Linguistics", Quine equates speaking meaningfully with using words in ways that don't evoke  responses "suggestive of bizarreness" and equates understanding with possession of "speech dispositions", but does not talk about theories and their empirical meaning.


  1. Post 1
    Dear Professor Putnam,
    You are too kind. I am no expert in these matters but really just a dabbler since metaethics is my principal AOS. However your posts open up a way of understanding Quine’s overall position that is at least new to me. And since I WILL be teaching ‘Two Dogmas’ towards the end of next semester, I would like to get things straight.

    Here goes:

    To prove: ‘Meaning’ is not a scientifically respectable concept. There are no meanings in any scientifically respectable sense. This does not mean that nothing ever means anything. Rather meanings are like sakes. We do things for the sake of things, but sakes are not the kinds of things that will figure in a scientifically respectable theory or in a scientifically respectable philosophy. Meanings, like sakes, cannot bear any intellectual weight.

    1) For the meaning of an expression or set of expressions to be an empirically (and hence scientifically) respectable notion, meaning would have to be equivalent to empirical significance (or a determinate contribution to empirical significance).
    [Quine thinks that the meanings (if any) of subsentences are derivative from the sentences in which they are semantically equipped to appear.]
    2) For an individual sentence to be empirically significant it would have to have determinate verification-conditions.
    3) No individual sentence has determinate verification-conditions. [Quine/Duhem: Whether a given set of observations confirms or disconfirms a sentence depends upon what else we assume.]
    4) So the ‘unit of empirical significance’ (if any) must be an entire theory or maybe the whole of science.
    5) But entire theories do not have verification-conditions, determinate or otherwise [Popper].
    6) So entire theories do not have empirical significance [From 5) & 2)].
    7) So entire theories do not have meanings in any (scientifically respectable) sense (and the same goes for the whole of science). [From 6) & 1).]
    8) Since subsentential expressions only have (scientifically respectable) meanings if the sentences in which they appear have (scientifically respectable) meanings, and since sentences only have (scientifically respectable) meanings if the larger theories in which they appear have (scientifically respectable) meanings, NO expressions have (scientifically respectable) meanings.

    Corollary 1. Analyticity, conceived as truth in virtue of meaning, is not a scientifically respectable concept, since we have no scientifically respectable conception of meaning.

    Corollary 2. If Necessity has to be understood in terms of analyticity, then Necessity is not a scientifically respectable notion either.

    Corollary 3. Carnap’s idea that ontological claims only make (scientifically respectable) sense in the context of a framework relies on a concept of meaning that is analogous to a sake. Either NOTHING makes scientifically respectable sense, including Carnap’s context-relative ontological claims, or we adopt a loose and colloquial conception of meaning in which context-free ontological claims such a s Quine’s ‘There are Numbers’ make just as much sense as Carnap’s context-bound claims.

  2. Post 2

    Objection: Couldn’t the empirical significance of any linguistic item from a sentence on up be defined in terms of its FALSIFICATION conditions? In that case we could derail Quine’s argument at step 2).

    This wont’ work

    A) Because individual sentences don’t have determinate empirical falsification conditions. [Quine/Duhem strikes again.] Thus instead of 3) we would get

    3#) No individual sentence has determinate falsification-conditions. [Quine/Duhem: Whether a given set of observations confirms or disconfirms a sentence depends upon what else we assume.]

    B) Because entire theories don’t have determinate empirical falsification-conditions either – auxiliary conditions, if only of the form ‘God isn’t messing with the system’, are always necessary. (A point you make in your critique of Popper and to which he responds with grumpy incomprehension.) So we would still get a version of step 5) namely:

    5#) Not even entire theories have determinate falsification-conditions [Putnam versus Popper].

    C) Because even if the totality of science (conceived of as great big heap of possible sentences) DOES have determinate falsification conditions and hence empirical significance, it can’t pass its (scientifically respectable) meaning back down to the individual sentences of which it is composed. On this scenario, the totality of science would be the ONLY thing with a scientifically respectable meaning. So we would get a version of conclusion 8) namely

    8#) NO expression except the set of propositions expressing the totality of science has a (scientifically respectable) meaning.

    This would be quite enough for Quine to draw his corollaries 1, 2 & 3.

    The underlying assumption is that meaning is no good unless it can be given an epistemic analysis. But it can’t be given an adequate epistemic analysis. Hence meaning is no good.

    Does this seem right to you as a capsule summary of Quine’s position?

    The way out, I suggest, is to reject the underlying assumption. Meaning cannot and should not be given an epistemic analysis.

    Charles Pigden

  3. Dear Professor Putnam,

    I am very pleased and humbled by your reply. I have heard the sentence/statement distinction made before in regards to this very topic, but, to be honest, I didn't understand the point fully. I will study the works of yours that you mention when I get a chance, and hopefully, inter alia, I will come to a better understanding of that.