Saturday, February 25, 2012

Philosophy of science as it was taught to John Rawls

[crossposted at Footnotes on Epicycles]

My colleague Jon Mandle has been looking at John Rawls 1950 doctoral dissertation, A Study in The Grounds of Ethical Knowledge. Jon asked me about a section in which Rawls contrasts ethical theory and scientific theory. The philosophy of science that he presumes is really just background. Yet he discusses what is now often called the Duhem-Quine Problem, a couple of years before Quine's "Two Dogmas of Empiricism". So where did Rawls get it from?

I did not have a good answer to this, beyond the obvious suggestion. So I decided to share the interesting bit here.

Some historical context: After coming back from service in World War II in the mid 1940s, Rawls began graduate work at Princeton. He spent a year at Cornell, where he interacted with Norman Malcolm and Max Black. Although he did not defend his disseration until February 1950, it was completed by about Fall 1948. (He had funding which was contingent on him still being a student.) So the bit here reflects philosophy of science as he was taught it at Princeton and Cornell in the mid to late 1940s.

The obvious answer, suggested by the footnote, is that he got his philosophy of science primarily from Max Black at Cornell. If you have other ideas as to who he might have learned philosophy of science from, please mention them in the comments.

One crucial distinction between the use of a theory in natural science, as opposed to its use in ethics, is that in the former the subject matter is the empirical laws expressed by different causal relations, and whenever the theory does not explain these the theory must be modified; whereas in the latter, the subject matter is the rational judgements of reasonable men, and while we have defined them to be coercive over a theory in the preliminary stage of inquiry, they can be altered, if reasonable men wish to do so; and they may want to change them should they discover that a few recognized judgments are not in harmony with some general principles which explicate most of their other judgments, and which seem to be justifiable. While an explication could hardly cause us to change all our judgments, it may, after we have reflected upon it, cause us to change some of our opinions. Therefore, not only may an ethical theory provide an answer where there is a genuine conflict, and so where there is no opinion at all; but it may actually change some accepted appraisal which was originally considered a part of the subject matter.

Thus an adequate and comprehensive ethical theory may have a control over its data which we generally do not allow to a theory in a natural scienoe. We cannot think that physical processes, having found that Newton's theory explained much of their behavior, would voluntarily agree to act in a manner conformant to its predictions. But in ethical theory this is just what may, and does, happen. Consider, for exanple, the argument of a reformer. He points out that an accepted moral judgment, or an accepted pattern of moral behavior, actually conflicts with a principle which explicates most of our best and soundest opinions. He appeals to us to recognise that such and such a judgment or pattern of conduct violates the principles which underly our common morality. He urges us to bring those discordant judgments and modes of conduct into line. This we can do; and this is a point at which the final use of an ethical theory may be so different from the final use of a theory in the natural sciences.

It may be objected to this difference that it is not so great as I have stated it. Scientific theories control their data, and exercise a coercive power over observations. For example, if an observer were to report that he had seen a body grow hotter in surroundings of lower temperature; or that he had seen all the molecules of a gas collect in a small volume at one end of a container; or that he had watched a heavy body float up in the air; - we should, on the basis of well-confirmed theories, strongly doubt his observations. We should use the evidenced strength of certain physical theories to argue that it is more probable that the observer is mistaken than that the recorded events have happened as he describes. A theory may be so generally accepted that it will throw out numerous reports on similar grounds. This is why, for example, miracles of one kind or another are not scientifically acceptable. It is more likely that the report of a miracle is false than that the theory it contradicts is mistaken. Naturally there is a limit as to how far a theory can discard observations relevant to it. Otherwise, it would not be a theory at all, but an opinion stubbornly maintained in the face of contradicting reports. But, by and large, the general rule obtains that a widely successful theory will serve to discard the few and scattered observations against it; and this is because we take it as more probable that the reports are mistaken than that the theory is incorrect. A theory is refuted by showing that there is a general law which is directly contrary to it; and a few random reports are not sufficient to show this.[footnote; see below]

Thus, while it is true that physical theory may control its data, the relation is entirely different from that which may exist between an ethical theory and its data. In the physical case, it is a question of weighing probabilities, and, in view of a satisfactory theory, there must be strong doubt as to whether a reported event contradictory to the theory eer occurred. But not so with an ethical theory: No one doubts that the common sense jugments contradicting an explication happen every day. We egrant, of course, their existence, but demand, in the light of the explication, that they be changed. This can be done, men being what they are. Ethical theory can have a distinctive control over its data; and it is part of its value that it can have this control, since it then can serve as a means for the reform and improvement of common morality.
[footnote:] The subject matter of a natural science like physics is, primarily, laws, which are stated, when the science is developed, in mathematical terms. If not, they may be called routines. See the discussion in Campell, Physics; the elements, Ch. 4. Or, as Feigl says, the 'Erkenntnisziel' of physics in the 'gesetzliche Gerust dar Welt', of Theorie und Erfahrung in die Physik, 16-18. Thus to refute a theory we must establish a law contradicting it. Popper, in Logik der Forschung, made an attempt to avoid the overstrict criterion of meaning then held by the Vienna Circle by urging that a theory be considered meaningful if it could be conclusively falsified. This test, he thought could be carried. out strictly, since he believed that a finite number of observations could refute a theory. But Black, in a review, exposed the fallacy: " scientific law is rejected on the basis of a finite number of contrary observations unless it is believed that the number of such observations could be indefinitely extended by any competent observer under similar conditions; strictly unique experiments, however discordant with theory, are neglected because their uniqueness guarantees their irrelevance: their importance is merely that of the inexplicable." 45 Mind 105 (1936).


  1. Well, I've seen it written that the Duhem-Quine Thesis really ought to have been called the Duhem-Neurath-Quine Thesis, and I believe that these ideas would have been "in the air" amongst philosophers of science at the time due to Neurath's influence.

    A quick look at Wikipedia shows that another philosopher of science at Cornell would have been E.A. Burtt (who is supposed to have indirectly influenced Kuhn). The obvious thought at Princeton would be Hempel, but he didn't arrive until '55. It looks like Ledger Wood, Andrew P. Ushenko, and George Berry may have taught philosophy of science there, but that doesn't tell me anything useful (though the latter two both published essays about / responding to Quine in the '40s).

  2. Yeah, I agree with Matt. The Duhem problem was well known by the time Quine wrote about it. If you follow the footnote, you can see that its a review of Popper's _The logic of discovery_. And black notes that Popper is aware of the Duhem problem (not by that name, of course) but refuses to admit that it "removes the chief reason for emphasizing falsification", and cites pages 17, 19, 46, and 48. So, even the greatest proponent of falsification of the era was well aware of the Duhem problem.

    Anyway, why think it is surprising that Rawles would have read a review, in Mind, of one of the most important philosophy of science books of the day before he wrote something about philosophy of science. It doesnt seem like much more explanation is called for.

  3. Not only, as Eric suggests, Popper was already aware of the Duhem-Quine problem but, if I remember correctly, he draws a distinction between the logical and methodological units of falsification, which largely anticipates Quine's holism.

  4. I'm sure that Darrell (our in-house Popper expert) can help us here.

  5. Eric, as you say, there really is not further explanation required. This passage might just reflect the bit of Popper that Rawls had learned from Max Black.

    I was just wondering if Rawls would have been steeped in other sources for this kind of view. As Matt notes, some related characters were at Princeton - but in the early 1950s.

  6. I'd like to know more about what was being taught at Princeton in this area in the mid-40's, but I've not had a lot of luck finding information on those three figures. I looked up the Campbell cited in the footnote, who is himself pretty interesting. Here's an article on him:

  7. I just finished my B.A. thesis (in history) on Rawls's early work, for which I did some archival research. I don't have more information than you guys about the teaching Rawls received at Princeton or Cornell, but based on my reading of his early unpublished work (and, of course, the dissertation) you're all quite right that Rawls was reading a lot of Popper, Neurath, and other Germanophone authors (he was reading in the original), probably through Black and Malcolm. Hempel and Nagel were other important sources, and he was citing a lot of articles not only from Mind but also Erkenntnis. P.D., you're definitely right that the philosophy of induction was "in the air" at the time, and Rawls understood himself to be applying findings from the philosophy of science as a field to ethical theory. In that respect, he drew a great deal on this article by C.J. Ducasse:

  8. I took 2 philosophy courses with Rawls at MIT (I was an undergrad there) in 1961-1962. I think that he mentioned Hanson, Polanyi and Toulmin on the coercive power of theory over observations.

  9. Great post! This is such an interesting passage from Rawls.

    Does anyone know why the thesis isn't just simply called "the Duhem thesis"? As far as I know, there never was a clearer explanation of the thesis than in Duhem's "Physical Theory and Experiment". I suspect that the answer has something to do with the specific *use* that Quine put the thesis to, but that seems a poor reason to name it after him.

  10. Interesting to see the strong contrast between causal relations and rational judgments, obviously before the development and influence of Davidson...

  11. Hempel talks about holism in his "Problems and Changes in the Empiricist Criterion of Meaning", I think, which is from 1950, but was at Yale at this time. But as others have pointed out Holism was "in the air" for quite a while.

    I mention Hempel since he was the one who apparently pointed out to Quine that Duhem had said something similar to Quine's holism in the Two Dogmas. In the original, Phil Review version of the Two Dogmas, there is no footnote to Duhem - Quine added it when it was reprinted.

    Re: why it isn't just called "The Duhem Thesis" - I think its mostly because Duhem's thesis was really just about physics (Duhem explicitly says other areas of science are unlike physics in this way), whereas Quine claimed it was true of all science (and, indeed, all one's beliefs).

  12. Carnap was on about this in the 1920s. Quine credits him, in 2 Dogmas, with the confirmational holism he uses against Carnap: "My countersuggestion, issuing essentially from Carnap's doctrine of the physical world in the Aufbau, is that our statements about the external world face the tribunal of sense experience not individually but only as a corporate body."

  13. Interesting. On Duhem's thesis, and some of the related discussion, I have a few comments:

    (i) Gabriele is right that Popper knew about the thesis early on (although he didn't call it that) and drew a distinction between logical and methodological falsification. One of my favourite passages for showing this, which comes from _Logik der Forschung_, appears at the beginning of section two of my recent 'Kuhn versus Popper on Criticism and Dogmatism' paper, which you can find, if you really want to, on my webpage! I don't quote it here because I want - shock, horror! - to make space for Ayer...

    (ii) Ayer not only knew about it too, but also recognised its significance for verificationism. This is from chapter one (pp. 20-1 of my copy) of _Language, Truth, and Logic_: 'A hypothesis cannot be conclusively confuted any more than it can be conclusively verified. For when we take the occurrence of certain observations as proof that a given hypothesis is false, we presuppose the existence of certain conditions.... Accordingly, we fall back on the weaker sense of verification. We say that the question that must be asked about any putative statement of fact is not, Would any observations make its truth or falsehood logically certain?... [W]e may say it is a mark of a genuine factual proposition... simply that some experiential propositions can be deduced from it in conjunction with certain other premises without being deducible from those other premises alone.' Someone may be able to tell us if something similar appears in the published writings of any members of the Circle beforehand...? (_Logische Syntax der Sprache_ would be the first place I'd want to look. I think the translation appeared too late, and have no idea whether it is entirely faithful to the original.)

    (iii) It is fair, as Chris says, to emphasize the restricted scope of the thesis as discussed by Duhem. But one when relaxes that scope, one doesn't arrive at Quine's thesis (as such). Donald Gillies's discussion of the two theses in his textbook _Philosophy of Science in the Twentieth Century_, which reappears in Curd and Covers' _Philosophy of Science: The Central Issues_, covers this pretty well. It's a great discussion for undergrads, I think.

    (iv) Actually, I find it plausible that Duhem's thesis was widely known before Duhem, tacitly if nothing else. Plus I recall reading a convincing paper that Poincare discussed it before Duhem made such a big deal of it. This was probably from a special issue of _Synthese_ on Duhem, e.g. 83(2). But I know my Duhem much better than I know my Poincare, so won't vouch for it.

  14. Duhem, The Aim and Structure of Physical Theory, 1906: “… the physicist can never subject an isolated hypothesis to experimental test, but only a whole group of hypotheses; when the experiment is in disagreement with his predictions, what he learns is that at least one of the hypotheses constituting this group is unacceptable and ought to be modified; but the experiment does not designate which one should be changed.” (p. 187) As stated this just concerns the physicist, but the point itself seems to be more general, or at least it seems naturally extendable, or at least there seems no reason not to extend it. But this theoretical-holism point is separate from the question of the revisabilty of data/observations in the light of theory or theories.

  15. 'An experiment in physics can never condemn an isolated hypothesis but only a whole theoretical group.' (Duhem, 1954, p. 183) also suggests that the scope is intentionally restricted, rightly or wrongly. Gillies argues that extending it is natural, as I recall.

    I'm not sure that I understand the last sentence, Nick. (Do you mean the issue of theory-ladenness is different? If so, I agree.)

  16. About Duhem and the thesis not going beyond physics, I remember Brad Wray (I think it was) told me (when I was a graduate student more than 10 years ago at UBC) that Duhem actually says explicitly somewhere that the thesis (as it is now known) is not applicable to chemistry. Presumably he then gives a reason why not. Is this a reason to not attribute the later version to him alone, since the argument of Duhem's itself might indicate why?

    As for Rawls, I have nothing specific to mention on the actual thread, but I always found it interesting that even _A Theory of Justice_ contains seeming links that do "remind" one that he seems to have to had very broad interests.

  17. Of possible relevance to this discussion, which I just happened across--at least as regards Popper-- is a passage I recently noted from Conjectures and Refutations where Popper alludes to “Duhem’s and Quine’s view”: Popper himself thought that “we can be reasonably successful in attributing our refutations to definite portions of the theoretical maze. (For we are reasonably successful in this—a fact which must remain inexplicable for one who adopts Duhem’s and Quine’s view on the matter.) (C & R, 1963, 242). Never mind that Popper did not supply an adequate account for warranting this.
    Feb. 3,

  18. Carnap, in "The Logical Syntax of Language" (1934), sec.82, observes the Duhem-point (at the time, as the association with Poincare shows, also to give (like CI Lewis as well) a justification for scientific conventions as respectable via empiricism. W. James, "The Conception of Truth in Pragmatism", makes very similar points to those of Quine's 2 Dogmas. I mantion each of these to add to the "in the air" idea, and a couple of titles that Rawls (as Rawls) and any student of Black would (have to be) familiar with who dares mention physical theory.


Note: Only a member of this blog may post a comment.