Tuesday, August 5, 2014

Quine's "Underdetermination Doctrine"
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

What follows this introductory paragraph are the opening paragraphs of chapter 16 [A Proof of the Underdetermination "Doctrine"]  of my Philosophy in the Age of Science. Because the confusion they address, between Quine’s “underdetermination doctrine” and “the Quine-Duhem Thesis” is still widespread, I have decided to post just those paragraphs. They point out that in  “On Empirically Equivalent Systems of the World,”  a paper I heard him read at a symposium in Stuttgart a year or two before it was published, Quine admitted that the underdetermination doctrine admitted of a trivializing interpretation, and proposed a way to rule it out, but confessed that, when so modified, it was something he could not prove. In homage to his memory, I decided to look for a proof. The reader who is interested in these matters can find my proof (which uses heavy-duty mathematical logic)  in my chapter, and can find further information in the book by Yemima Ben-Menahem I cite below.

Prior to the publication of “On What There Is” in 1948, W.V.O. Quine was known to most philosophers (if they had heard of him at all) as a “logician.” But now it is universally acknowledged that the series of Quine’s philosophical papers and books that began with that famous essay are among the most important to have been published in (approximately) the second half of the twentieth century. One of the most famous theses in that series of works is “the doctrine that natural science is empirically underdetermined.”[i] Yet I have observed that Quine’s “doctrine” is widely misunderstood, and its logical status is, in fact, murkier than most readers imagine. The purpose of the present essay is to remove the murk and to prove the doctrine.
The principal misunderstanding I have observed is the idea that underdetermination follows immediately from “the Quine-Duhem thesis,” by which is meant the thesis that theoretical claims cannot be either conclusively falsified or conclusively confirmed by isolated experiences. If those experiences appear to falsify a theory, it is always possible to revise parts of the theory (including what the logical empiricists called the “coordinating definitions”), or, if worse comes to worse, to reject the observation sentences that we use to report those experiences. However this claim has literally nothing to do with the “underdetermination” doctrine. That doctrine, to which Quine gave considerable weight in his writings,[ii] says that even if we do not reject any observation sentences that we take to have been verified, and even if we do not allow any changes to the theory, still, if there is at least one theory that has a given set of observational consequences, then there will always be more than one.
Another misunderstanding that I have observed, even among those who do not confuse “the underdetermination doctrine” with “the Quine- Duhem thesis,” is that the underdetermination doctrine is, somehow, logically obvious. But it is not obvious, and in her brilliant book on conventionalism, Yemima Ben-Menahem reminds us that in a little-known lecture Quine surprisingly confessed that the doctrine risks being either trivial or, at best, unprovable! As Yemima Ben-Menahem has pointed out, in the original version of “On Empirically Equivalent Systems of the World,” Quine suggested a way of stating the doctrine that would avoid triviality, but concluded that so restated, the truth of the doctrine “is an open question.”[iii] However, all of this rethinking was excised from “On Empirically Equivalent Systems of the World” before that essay was collected in Pursuit of Truth![iv]
As Ben-Menahem explains the problem:
Underdetermination crumbles under a problem that seems utterly trivial at first. Quine invites us to consider two theories that are identical except for a permutation of terms, for example “electron” and “proton.” What is the relation between these theories? Taken at face value, they are clearly incompatible, for each affirms sentences the other denies, for instance, "The negative charge of the electron is..." It is likewise clear that these incompatible theories are empirically equivalent—they have exactly the same empirical import. The question is whether this would count as an example of underdetermination, that is, whether such a minor permutation of terms suffices to render the two theories empirically equivalent but incompatible alternatives. Rather than taking them at face value is it not more reasonable to regard them as slightly different, though perfectly compatible, formulations of the same theory?[v]
She goes on to explain that Quine endorsed the latter alternative. He wrote, “So I propose to individuate theories thus: two formulations express the same theory if they are empirically equivalent, and there is a construal of predicates that transforms one theory into a logical equivalent of the other.”[vi]


[i] W.V. Quine, “On Empirically Equivalent Systems of the World,” Erkenntnis, 9 (1975): 313-328; quotation from p. 313.
[ii] For example, in W.V. Quine, “On the Reasons for the Indeterminacy of Translation,” Journal of Philosophy, 67 (1970): 178-183, Quine gives the underdetermination of theory by data as one of the two principal reasons for his famous (and controversial) thesis of the indeterminacy of translation.            
[iii] W.V. Quine, “On Empirically Equivalent Systems of the World,” Erkenntnis, 9 (1975): 313-328; quotation from p. p. 327.
[iv] W.V. Quine, Pursuit of Truth (Cambridge, Mass.: Harvard University Press, 1990/2004).{ADDED TODAY: This was a mistake. Quine did not "collect" this paper in Pursuit of Truth; but he did write (p.96) "Effort and paper have been wasted, by me among others, over what to count as sameness of theory and what to count as mere equivalence. It is a question of words; we can stop speaking of theories and just speak of theory formulations. I shall still write simply 'theory', but you may understand it as 'theory formulation' if you will." He does not mention that this makes the underdetermination doctrine trivial.]
[v] Yemima Ben-Menahem, Conventionalism: From Poincaré to Quine (Cambridge. Cambridge University Press, 2006), 246.
[vi] Quine, “On Empirically Equivalent Systems of the World,” 320.


  1. Very interesting. There are several points that could be elaborated and specified further. Our library doesn't have your book that you mentioned, but I requested that they purchase it. The core of the "underdetermination doctrine" as you state it seems to be that, "... if there is at least one theory that has a given set of observational consequences, then there will always be more than one." Can I restate it like this? "If there is at least one theory T1 that has the given set of observational consequences O1, then there will always be another theory T2 that has the set of observational consequences O1." I take it that T1 and T2 are different somehow, but that they are equivalent with respect to the fact that they have as observational consequences the set O1, although not equivalent in other respects. (I am not here talking about alternative (equivalent) formulations F1, F2 of a single theoryT1.) However the set of relevant experiences is never closed and there could be a verified observation sentence not included in set O1, which is implied by T2 but not T1, because the categories of T2 have differentiations not contained in T1. When one claims that objects, such as T1 and T2, are equivalent, one must specify in which respects they are equivalent, and also, if there are two of them, the respects in which they differ. There indeed is a parallel, I think, between the case of theories and

  2. (continued from above comment)
    that of translation, in that there is a specifiable equivalence in general logical form of the relation between a theory (as a structural instrument), or two of them, and their observational consequences (expressed in observation sentences, to use your terminology) on the one hand, and on the other the relation between a linguistic system (or two of them) and true descriptions of a given observed situation e1. What are usually called translation equivalents are never equivalent in every possible respect. The respects in which they differ may be the source of insights into the understanding of the situation e1, that were not available from the one language alone. Is this why Carnap held his principle of charity?

  3. It looks like I didn't say, but maybe I didn't need to say, that, like linguistic systems and their categories, theories are open- ended with regard to differentiation of their categorial structures, and thus it will always be possible to integrate new experiences into a series of theories related as T1 and T2 above. (Or maybe I shouldn't have said this, but that's what it looks like..) Anyway, as I said above, if we give these ideas and the ones in Prof. Putnam's post further elaboration and specification, I think they can help us to understand how theories and languages enable us to make sense of the world.