Why does Quine expand the underdetermination thesis through his global holism?

Question

Quine's application of the problem of underdetermination took the thesis to be a problem not only for physics (as Duhem before him), nor even for the particular sciences, but for any and all theories about knowledge. There is as such a disagreement essentially over the strength of the thesis between the two philosophers most associated with the thesis of underdetermination. And there is put forward by Quine a bold conjunction between this thesis and global holism.

However, I am curious as regards the following argument:

• Quine views the fact that an infinite amount of theories can be relevant to a given topic or empirical observation as proof that a.) an infinite amount of theories period is relevant to any observation and b.) all such theories are entertained by theory makers, and so all of our theories are relevant to each other and contain by certain qualification the same empirical content, and/or empirical justification (as Quine puts it, "whatever observation would be counted for or against the one theory counts equally for or against the other" 1992).
• The mistake as I see it can be identified in the following thought problem: the observation that wheels tend to roll downhill can be said to be relevant to a theory about rotation because it is considered to be the case that wheels roll in 'rotations'. This observation can also be said to be relevant to a theory of wheels, or a theory predicting increase in velocity of a wheel, ad infinitum. In this sense there are an infinite number of theories related to the observation at hand. For the sake of the example let's use a theory concerned with rotation.
• Now, Quine would hold that the reason we think that 'rotations' can adequately describe a wheel is because a wheel supposedly spins around an imaginary axis located in its center. And this imaginary axis is understood in terms of a certain point or line on an abstract plain. As such, Quine would hold that belief in an abstract plain is entirely and wholly relevant to the simple observation that some wheel rolls downhill. This belief, under Quine's vision of revision and global application of the underdetermination thesis, can be destroyed or justified, moved this way and that, by the power of the observation of a wheel rolling down the hill.
• This however is obviously not the case. And it is not the case because this line of thought which is in some manner connected to the theory of rotations which actually is relevant to the observation, is not relevant in the same manner to the observation for the fact that it contains no inherent concern for the subject matter of the observation. Its relation to the empirical observation is wholly determined by its relation to a theory that is inherently concerned with the observation. The relevance of the abstract plain to a rolling wheel is one derived from some theory more directly relevant to the observation. This point can be clarified in the ambiguity that results if we hold the abstract plain in itself to be relevant to the observation of the wheel, for there are endless ways in which it can be said to be 'relevant', since there are endless aspects of an abstract plain and wheels rolling. It is only when conjoined to the particular theorizing about a particular aspect of an observation that a theory or belief can be said to be relevant to a given observation. As such, what makes a theory 'relevant' to a given observation is its concern for the aspect of the observation under consideration.
• But as such, it just seems to be evident that certain theories really aren't 'empirically equivalent', in that where certain theories are inherently about an aspect of an observation, other theories are only about such an aspect in a derivative way.
• Thus, Quine's notion of global holism seems to collapse.

Is this argument fair to Quine? If not, in what way is it a misread?

Answer 1

That Quine greatly expanded Duhem's thesis (about which he learned only after writing the Two Dogmas) is generally acknowledged. Zammito gives a detailed comparison in The Nice Derangement of Epistemes, summarizing as follows:"Quine has extended the original proposition of Duhem in two crucial manners. First, he has introduced the radical phrase "come what may" into the considerations, so that he can be taken to have endorsed "drastic" rearrangement of the web of our beliefs not merely as possible but as plausible. Second, he has practiced a "semantic ascent" upon Duhem's empiricism, transposing an issue of theory and experiment into an issue of language and, indeed, situating the issue at the most holistic level within this scheme of semantics. Thus, Quine offers phrases like: "the unit of empirical significance is the whole of science." At that level of abstraction, there could be no real distinction of the whole of science from the whole of language."

So I think your reconstruction somewhat undersizes and overconceptualizes Quine's holism. The reason he thought that "the unit of empirical significance is the whole of science" is not as much psychoanalysis of our belief reasons as his rejection of atomic "meanings" as semantic tools, "semantic globalization". Roughly speaking, his analysis of language led him to believe (as it did Wittgenstein) that concepts and propositions acquire significance only as placeholders in a web of belief (or language game), not as stand alone entities. Your observation that "wheels tend to roll downhill" is not even an observation since it involves "induction" over a class of observations, and "meanings" of "wheel", "roll", and "downhill", not to mention "tend to", you won't be able to explain to others (to make sure that the same "observation" is being talked about) without referring to other concepts, and ultimately to all of common practice that embeds them. He is unlikely to be impressed by making distinctions about aspects and attempts to analyze relevance for the same reasons that led him to reject the analytic/synthetic distinction, and characterize meaning as "will-o'-the-wisp". Quine simply saw that any attempts at distinguishing and piecemeal justification are transparently vulnerable to extensions of the same objections that took down the logical atomism of early positivists, and took the preemptive step of nipping the idea in the bud.

This said, Quine did walk back some of his more polemical and radical statements in his later works, like Two Dogmas in Retrospect. This walkback however was more pragmatic than principled: "Looking back on it, one thing I regret is my needlessly strong statement of holism. 'The unit of empirical significance is the whole of science... Any statement can be held true come what may.' This is true enough in a legalistic sort of way, but it diverts attention from what is more to the point... In later writings I have invoked not the whole of science but chunks of it, clusters of sentences just inclusive enough to have critical semantic mass". Pragmatic approaches to stratifying and parsing Quine's and Kuhn's holism that resonate with what you suggest were developed e.g. by Friedman in Dynamics of Reason, and especially by the structuralist school in philosophy of science (Suppes, Sneed), explicitly endorsed in Kuhn's Theory-Change as Structure-Change.