18 March 2010

Duhem, Quine, and Underdetermination

The Duhem-Quine thesis is identical to neither the Duhem thesis nor the Quine thesis,[1] which in turn are not identical to one another. The Duhem-Quine thesis acquired its name in a footnote to W.V.O. Quine’s 1951 article “Two Dogmas of Empiricism” as it appeared in his 1953 collection From a Logical Point of View.[2] Specifically, Quine says, “This doctrine was well argued by Duhem,”[3] the doctrine being “that our statements about the external world face the tribunal of sense experience not individually but as a corporate body.”[4] This is the only statement, vague though it is, that can rightly be called the Duhem-Quine thesis. First, it is here and only here that Quine establishes a link between his underdetermination thesis and that of Pierre Duhem. Second, it is here that the idea of a Duhem-Quine thesis has emerged in the philosophical vocabulary. Thus, apart from what is explicit in the Duhem-Quine thesis so stated, we have an exegetical basis to insist upon a separate Duhem thesis and Quine thesis. Further, these theses are to be maintained as separate even where they might concur. This paper will attempt no addition to this historical and minimal Duhem-Quine thesis (even if the latter has often been equivocated with the Quine thesis). The Duhem-Quine thesis (hereafter DQT), the Duhem thesis (DT), and the Quine thesis (QT) are identical neither historically nor philosophically.

Before moving on, it behooves us to comment that DQT is introduced into “Two Dogmas” against the second dogma of reductionism.[5] Quine makes use of a distinction between “radical reduction” and an attenuated reductionism. The former “sets itself the task of specifying a sense-datum language and showing how to translate the rest of significant discourse, statement by statement, into it.”[6] The latter “survives in the supposition that each statement, taken in isolation from its fellows, can admit of confirmation or infirmation.”[7] It is against the latter, attenuated form that Quine offers DQT. Quine’s point in stating DQT, then, is to say confirmation or infirmation in isolation is impossible, that reduction is impossible, not only attenuated reduction but also radical reduction, since the impossibility of radical reduction is a consequence of the impossibility of attenuated reduction. Thus, DQT is at the heart of QT’s stand against the dogmas of empiricism; indeed, it has come into being as a central philosophical component of QT.

If DQT really constitutes a bond between DT and QT, then we shall find that DT also includes DQT among its philosophical components; however, DT and QT must include independent philosophical components, or else all three would just be DQT. Further, if the uniqueness of all three theses is to be maintained, then disparities must exist between DT and QT, or else DT and QT would be the same, and DQT could be properly redefined in stronger terms according to a reunified DT and QT. Let this hypothetical thesis be called strong DQT`. If this were the case, Quine would have observed DQT just as a component of DQT`, which includes is the identical DT and QT. It is also possible that there are components of either DT or QT (or both) that are neither included in nor contradicted by the other. If DT and QT are non-identical in this way, then it remains possible to assemble DQT` from all the points of explicit concurrence. It is also possible to assemble another thesis which includes all these compatible components of DT and QT, which we shall call strong DQT``. DQT`` is another strengthened form of DQT from analysis of DT and QT. Lastly, one could find the existence of not only concurrent components, but also compatible components as well as incompatible components. In principle, it is still possible to assemble a weak DQT` and a weak DQT``, but only by eliminating the incompatible components in the case of the latter and both the incompatible and the merely compatible components in the case of the former. Such weak theses as these will be ahistorical constructs among a plurality of theses that includes the historical trio of DQT, DT, and QT. They will not enjoy the historical foundation of the strong formulations of DQT` and DQT``.

In order to show that no ideal formulation of DQT` or DQT`` is possible, maintaining the uniqueness of the DQT, DT, and QT, it is necessary to describe the components of DT and QT, since the minimal components of DQT have already been stated. Let us turn now to Duhem, since it has already been shown that QT includes DQT. In The Aim and Structure of Physical Theory, Duhem makes a multitude of statements that sound very much like DQT. For example, “Physics is not a machine which lets itself be taken apart; we cannot try each piece in isolation…Physical science is a system that must be taken as a whole; it is an organism in which one part cannot be made to function except when the parts that are most remote from it are called into play.”[8] As long as “our statements about the external world” include representative physical theories, it seems that Quine was almost right to call Duhem a defender of DQT, but DT is a narrower thesis than DQT.

The main limitation on DT comes from the fact that Duhem was concerned primarily with physical theory. Even when he includes physiology and chemistry by saying, “for the physiologist and chemist as well as for the physicist, the statement of the result of an experiment implies, in general, an act of faith in a whole group of theories,”[9] he must be read in terms of what he said immediately before:

When [scientists] make use of physical instruments…[they] implicitly admit the accuracy of the theories justifying the use of these pieces of apparatus as well as of the theories giving meaning to the abstract ideas…by means of which the concrete indications of the instruments are translated. But the theories used, as well as the instruments employed, belongs to the domain of physics.[10]

So DT is unquestionably a thesis about physics. To further elaborate Duhem’s view, let us pursue the above account of apparatus. According to Duhem, the physicist conceives an apparatus in two ways: the concrete apparatus and the schematic or abstract apparatus. The former is useless to physical theory unless it is translated into the language of physical theory, since physical theory is an abstract structure of relations that attempts to represent an ontological order. The abstract instrument is as much part of the theoretical structure as a mathematical equation.

It is in such a structure as the following that the abstract instrument is given its meaning, whereby the readings of the concrete instrument acquire abstract meaning:

The materials with which [physical] theory is constructed are, on the one hand, the mathematical symbols serving to represent the various quantities and qualities of the physical world, and, on the other hand, the general postulates serving as principles. With these materials theory builds a logical structure; in drawing the plan of this structure it is hence bound to respect scrupulously the laws that logic imposes on all deductive reasoning and the rules that algebra prescribes for any mathematical operation.[11]

For Duhem a physical theory is an approximate, symbolic representation of the physical world, based on certain postulates from which the theory is developed to completion by deduction. The conclusions of such a theory, if the theory is fully developed, are subject to experimental inquiry, which itself requires translation into the abstract. It is for this very reason that when the interpretation of an experiment finds itself in contradiction with a theoretically established conclusion, it is impossible to tell where in the theoretical construct there is error: “The only thing that the experiment teaches us is that among the propositions used to predict the phenomenon and to establish whether it would be produced, there is at least one error; but where this error lies is just what it does not tell us.”[12] By advancing this thesis, Duhem wants to demonstrate the impossibility of a Baconian experimentum crucis, which decisively determines the truth of one hypothesis by showing the alternatives,[13] and it is this impossibility that leaves us vulnerable to the problem of underdetermination. All of this represents a significant departure from DQT, which has nothing explicit to say about locating error in a theoretical structure or even about possibility of determination at all, and in this way DT is a stronger thesis than DQT. To put it in perfectly Duhemian words, DT is a stronger (DQT does not preclude an experimentum crucis) but narrower (it is limited to physics) thesis than DQT.

Limitation of DT primarily to physical theory is to be contrasted with the vastly broader scope that Quine intends for QT. QT is intended to cover far more than just physical theory; after mentioning the underdetermination of the “over-all algebra of rational and irrational numbers” by the algebra of rational numbers, Quine says, “Total science, mathematical and natural and human, is similarly but more extremely underdetermined by experience. The edge of the system must be kept squared with experience; the rest, with all its elaborate myths or fictions, has as its objective the simplicity of laws.”[14] The inclusion even of mathematical systems[15] under QT goes far beyond that with which Duhem would have been comfortable, the latter having said of physical theory, “we see nothing analogous to the certainty that a mathematical definition draws from its very essence, that is, to the kind of certainty we have when it would be foolish to doubt that the various points on a circumference are all equidistant from the center.”[16] Mathematical truths are safe from DT but not QT. Quine would agree that such doubt is foolish, but not because of the analyticity to which Duhem appeals in order to establish this apparent mathematical truth. On the contrary, Quine says, “I espouse a more thorough pragmatism,” by which he explicitly means pragmatism that denies the “imagined boundary between the analytic and the synthetic.”[17] In this way, even the mathematics that Duhem would take to be plainly analytic are, for Quine, no different from such myths as the existence of physical objects and the existence of the Olympian pantheon. It is just that mathematics and the hypothesis that the physical world exists both have pragmatic value, where Zeus does not. This matter will rise again below.

The thrust of this contrast is that QT is a much stronger thesis than DT, the scope of the former being impressively vast, the scope of the latter being limited to physical theory. DT takes the cleavage of the analytic and the synthetic for granted, as evidenced by Duhem’s view of mathematics. Also revealing Duhem’s tacit endorsement of analyticity is his assertion, “Certain fundamental hypotheses of physical theory cannot be contradicted by any experiment, because they constitute in reality definitions, and because certain expressions in the physicist’s usage take their meaning only through them.”[18] Duhem provides an example of such a hypothesis. For a freely falling heavy body, acceleration is constant. This hypothesis is immune to experimental refutation because the definition of a freefall involves constant acceleration. “Falling freely” may mean different things between ordinary and scientific parlance, but in the structure of physical theory, such a definition is analytic.

Quine does allow for any such thing. In the course of his argument that any account of analyticity is viciously circular, Quine touches directly on the idea of a “definition.” By way of summary, the problem of analyticity is traced out from the difference between statements of the class, “(1) No unmarried man is married,” and those of the class, “(2) No bachelor is married.”[19] The former “is not merely true as it stands, but remains true under any and all reinterpretations of ‘man’ and ‘married.’”[20] It is a point about the logical relations of the terms in question, not of the meanings of the terms, and this is the Achilles’ heel of the second class. In order to get (1) from (2), thereby showing the analyticity of (2), Quine tries to appeal to “synonymy” in order to substitute “unmarried man” for “bachelor.” In order to account for synonymy, Quine makes an abortive attempt to appeal to definition. In short, he concludes that there is only one sort of definitional activity that does not appeal to some pre-existing synonymy, “the explicitly conventional introduction of novel notations for purposes of sheer abbreviation,”[21] and thus definition does not account for analyticity. Synonymy remains the problem. Quine concludes that no account of synonymy can be given without appeal to analyticity, and no account of analyticity can be given without synonymy, hence the vicious circle.[22]

Quine would have to read Duhem’s account of constant acceleration being in the definition of freefall as resting on a synonymy between freefall and constant acceleration, but synonymy must appeal to analyticity between freefall and constant acceleration, which must appeal to synonymy, and so on. This is the allegedly vicious circle, and it represents an important difference between QT and DT. Quite simply, DT allows for the creation of experimentally independent, analytic truths within a physical theory. Far from the creation of novel notation for abbreviation (such as using “F” for “freefall”), this is a case of “the definiens…in the spirit of explication, improv[ing] upon the antecedent usage of the definendum.”[23] “Freefall” is a translation of certain more primitive components of a physical theory. QT in no way disallows a theory to contain such definitions, but it denies them their analyticity. Thus QT comes out again as a much stronger thesis than DT. When the theoretical structure is called into question by an experimental result, DT does not include these “analytic” definitions among possible errors. QT must include them all, excepting mere abbreviation.

At this point it should go without saying that DT and QT are distinct theses, QT being a much stronger thesis, first because Quine does not limit himself to the sort of physical theory that Duhem is explicating, and second because Quine denies the distinction between the analytic and the synthetic and reductionism, the “two (equivalent) dogmas of empiricism.” Further, QT includes DQT, but DQT exhausts neither the strength nor scope QT. For example, one would not suppose from DQT alone that QT would call mathematical “truths” into question, but DQT does not preclude that expansion. DT is also stronger than DQT, but it is narrower, so it cannot be said to include DQT. Thus, DQT is a part of QT, but it is not a part of DT, and DT and QT are distinct. Because of this, the only historical way to strengthen DQT is with QT, which is consistent to typical mention of the “Duhem-Quine Thesis” meanting, in reality, the “Quine Thesis.” Indeed, DT is quite distinct from both DQT and QT, and thus it is rather misleading to speak of DQT at all, not only in the minimal form above, but in any strengthened form that attempts to join together DT and QT. If this analysis is right, then to assemble any form of DQT is a highly unnatural and artificial thing to do, and it can do justice to the central components of neither DT nor QT, since those components have been shown to be incompatible in at least the ways noted above. There is, however, one interesting point on which DT and QT may come very close together, and that is in the response to their respective consequences.

It seems that the consequences of DT and QT led both Duhem and Quine to a very similar epistemological conclusion, one that can arguably be placed under the broad banner of virtue epistemology. Kyle Stanford says of Quine:

[Quine] argues (1955) that our actual revisions of the web of belief seek to maximize the theoretical ‘virtues’ of simplicity, familiarity, scope, and fecundity, along with conformity to experience, and elsewhere suggests that we typically seek to resolve conflicts between the web of our beliefs and sensory our experience in accordance with a principle of ‘conservatism,’ that is, by making the smallest number of changes to the least central beliefs that will suffice to reconcile the web with experience.[24]

Stanford goes on to say that Quine adopts these virtues on an explicitly pragmatic basis, since we are naturally disposed to adopt beliefs in accordance with them. D. J. Stump writes of Duhem, “It has not been generally noticed the extent to which Duhem’s ‘good sense’ is an ethical term and the extent to which Duhem thinks that the intellectual and moral virtues of the scientist determine scientific knowledge.”[25] This apparent agreement derives from a shared need of both DT and QT to make sense of the decision between underdetermined alternatives. Further, in the cases of both Quine’s “virtues” and Duhem’s “good sense,” we find that underdetermined decisions are not made on the basis of anything in the alternatives in question; rather, they are based on a quality of the decision-maker. The “virtues” that Quine names derive from natural tendencies, hence pragmatism, an explicit consequence of “Two Dogmas.” Duhem’s “good sense” is a developed quality in the physicist (or doctor, musician, etc.), which is much more properly called a virtue. Thus one can develop a physicist’s good sense, or a doctor’s good sense, or a musician’s good sense, thereby becoming a virtuoso in the relevant sphere.

Duhem finds his “good sense” in the Penseés of Pascal, and he develops the idea of intellectual virtue to a considerable degree in the middle of Physical Theory. It is here that he contrasts the typical English mind, which is ample but weak, with the typical French mind, which is strong but narrow. So Duhem specifies two intellectual virtues, strength and ampleness, and two intellectual “vices,” weakness and narrowness. Abstract minds, given the virtue of strength, “have no difficulty conceiving of an idea which abstraction has stripped of everything that would stimulate the sensuous memory…they are skillful in following…down to its final consequences, the reasoning which adopts judgments for its principles.”[26] Visualizing minds, given the virtue of ampleness, “have a wonderful aptitude for holding in their imaginations a complicated collection of disparate objects; they envisage it in a single view without needing to attend myopically first to one object, then to another; and yet this view is not vague and confused, but exact and minute.”[27] The latter lends itself to weakness, which is of course opposed to strength, and the former lends itself to narrowness, which is opposed to ampleness. It is ampleness that can guide judgment of that which is underdetermined by reason. It is this structure that Duhem invokes when he says:

Pure logic is not the only rule for our judgments; certain opinions which do not fall under the hammer of the principle of contradiction are in any case perfectly unreasonable. These motives which do not proceed from logic and yet direct our choices, these ‘reasons which reason does not know’ and which speak to the ample ‘mind of finesse’ but not to the ‘geometric mind,’ constitute what is appropriately called good sense.[28]

This statement, which would not be foreign to Quine, is Duhem’s solution to the problem of underdetermination. On the other hand, Quine says, “Conservatism figures in such choices, and so does the quest for simplicity,”[29] where conservatism and simplicity are pragmatic mental tendencies. It is by appeal to mental qualities that Quine’s response to QT overlaps Duhem’s response to DT. This common direction is suggests that it is a good means by which to approach DQT and various strengthenings of that thesis, as well.

Duhem and Quine both may think that intellectual character can justify the choice of one underdetermined alternative over another, but the means by which this character guides decision-making is different. We have already seen that each has his own account of intellectual “virtues,” but it is even more important to see that where Quine’s virtues are basically descriptive, Duhem’s are unmistakably normative. Quine’s idea of conservatism, for example, just describes our tendency to think conservatively, and his pragmatism does not say that we ought to think in this way. Duhem’s account is explicitly normative; indeed, the term “good sense” itself implies normativity by using the value term, “good.” Moreover, the aforementioned mental “virtue” of ampleness is unmistakably the trait by which we ought to think in order to make better judgments of the underdetermined, just as the “virtue” of strength makes it easier to trace out the structure of a complete physical theory to be subjected to experiment. It is because of DT that both mental “virtues” are needed among physicists, the strong to build the theories, and the ample to trace out their underdetermined problems.

Even if DT and QT are sufficiently distinct as to disallow the construction of a historically sensitive version of DQT, there is clearly a common thread among these underdetermination theses, in that their differences can in many ways be accounted for as differences of degree. Thus the language of strength, weakness, broadness, and narrowness has proved very useful in differentiating among the theses. It seems that this common thread has lead to a common solution in virtue epistemology, broadly conceived, but it is a solution that bears differences according to the different theses. The stronger thesis of Quine demands a weaker solution; thus he is left with his thorough pragmatism and non-normative “virtues.” As for Duhem, “good sense” is strong enough to allow even for scientific progress, in accordance with his evolutionary view of the history of science. Because of this, when Duhem believes that a theory generally reflects an ontological order, the belief is much stronger than Quine’s pragmatic assent to the utility of the theory. Quine pays a price for QT’s strength in his weaker, pragmatic solution while Duhem reaps the benefits allowed by his weaker DT.



[1] The distinction between the Duhem thesis and the Quine thesis is nothing new; indeed, Donald Gillies has written on the subject under the very title, “The Duhem Thesis and the Quine Thesis” (1993). In Philosophy of Science: the Central Issues. Martin Curd and J.A. Cover, ed. Norton, New York, 1998. 302-319

[2] This observation comes from Ariew, Roger. “Pierre Duhem.” Stanford Encyclopedia of Philosophy. (13 July 2007). sec. 2.1

[3] Quine, W.V.O. “Two Dogmas of Empiricism.” From a Logical Point of View. Revised ed. Harvard, Cambridge, MA:1961. fn. 17

It should be noted that this is not the last edition of “Two Dogmas” that Quine produced; however, these later changes have been taken into account. Citing from the 1953 version presents no exegetical problem to this paper.

[4] Ibid. 41

[5] For Quine, though, “The two dogmas are, indeed, at root identical” (41), so DQT is of course opposed to both dogmas.

[6] Ibid. 39

[7] Ibid. 40

[8] Duhem, Pierre. The Aim and Structure of Physical Theory. Philip Wiener, tr. Princeton University Press, Princeton: 1954 (1906). pg. 200

[9] Ibid. 183

[10] Ibid. 183

[11] Ibid. 205

[12] Ibid. 185

In sec. 2.1 of “Pierre Duhem,” Ariew calls this the “non-falsifiability thesis,” which he says is the “obvious corollary” to what he calls the “non-separability thesis,” which we have already seen. It says simply that hypotheses cannot be tested in isolation. Ariew’s division of DT is useful, as it amplifies DT’s departure from DQT. DQT endorses a broader non-separability thesis, but it says nothing about non-falsifiability.

[13] Duhem’s brilliant example on this point is Foucault’s demonstration that light travels faster in water than in air (189), which could be thought to displace emission theory and confirm the wave theory. These, however, are not the only two possible hypotheses; further, there could be assumptions at work that allow the theoretical structure to be modified such that the experiment does not even discount emission theory. At any rate, there is no decisive demonstration here, and thus no crucial experiment.

[14] Quine 45

[15] Quine directly states that “the abstract entities which are the substance of mathematics—ultimately classes and classes of classes and so on up—are another posit in the same spirit” (45).

[16] Duhem 211

[17] Quine 46

[18] Duhem 209

[19] Quine 22-23

[20] Ibid. 22

[21] Ibid. 26

[22] Ibid. 32

[23] Ibid. 27

[24] Stanford, Kyle. “Underdetermination of Scientific Theory.” Stanford Encyclopedia of Philosophy. (12 August 2009) sec. 2.1

[25] Stump, David. “Pierre Duhem’s Virtue Epistemology.” Studies in History and Philosophy of Science. Vol. 38, No. 1(March 2007), pg. 149-159

[26] Duhem 56

[27] Ibid. 56

[28] Ibid. 217 It should be noted that the “geometric mind” corresponds to the “abstract mind” and the “mind of finesse” corresponds to the “visualizing mind.”

[29] Quine 46

No comments: