Outline:
Baconian view :: Quine-Duhem Thesis (pragmatism)
--> the underdetermination of theory and affiliated problems (the need for philosophy, the danger of circularity, and the threats to scientific realism and the "hypothetical-deductive" method of science, the method enshrined in...
Baconian view :: Quine-Duhem Thesis (pragmatism)
The "naive Baconian" view of science:
...science as the straightforward gathering of "facts" and their systematic organization into useful collections of facts. The idea is that nature will tell us what the facts are, provided we learn how to listen to nature properly. We perform our experiments, and nature rules on the outcomes. We simply have to be sharp enough to devise the right sorts of experiments to distinguish the true facts from the false conjectures. A central presumption of the native Baconian view is that "crucial" experiementation is the hallmark of science. A crucial experiement would be an experiment that unambiguously favored one or the otehr of two competing hypotheses under test, no matter the outcome. (63)
While an attractive and compelling view - almost no one is a Baconian anymore. In particular,
Pierre Duhem (french physicist, philosopher of science, early 20th ct.) "...argued around the turn of the twentieth century that crucial experiments are not possible in physics. The philosopher W.V.O. Quine, at least during one period of his distinguished career, took the radical view that crucial experiments are forthcoming in no domain of scientific inquiry; indeed, Quine in his more radical moments can be viewed as maintaining that crucial experimentation is a fairy tale of logic nowhere realizable no matter how smart human inquirers are. (64)
--> the Quine-Duhem Thesis:
(Klee's version): Any seemingly disconfirming observational evidence can always be accomodated to any theory.
In other words, the Quine-Duhem Thesis denies that the disconfirmation of a theory can be forced upon a practitioner by the evidence itself. Nature does not rule, we do; we do in the sense that our rejection of the theory under test on the basis of evidence is an inferential decision made by us, not nature. For, if we really wanted to, if we were willing to change enough of our other beliefs, we could render the evidence in question consistent with the theory under test. (65; emphasis added, CE)
In other terms we have used: nature may kick back - but that doesn't force us to change our theory.
All this follows straightfowardly if we move from the positivist view of theory to a holistic view - one that claims that
the terms and laws of a theory form one interdependent network. An important and obvious consequence of this is that it is always a large chunk of a theory that faces an experimental test, never one isolated prediction. It follows that a piece of seemingly disconfirming observational evidence never contradicts a single isolated claim, it contradicts the large chunk of theory that the claim that otherwise appears to be directly under test is part of....Obviously, therefore, we could "save" the claim that the context might seem to suggest is directly under test by adjusting other claims in the large chunk of theory with which it is connected. (65; emphasis added, CE)
It is important, on Klee's view, not to overstate the point here - as, for example, Larry Laudan does, by taking it to mean that "...there are no defensible, rational grounds for rejecting one theory on behalf of another where the competing theories are all consistent with the observational data." (65)
For Klee, this is not Quine's point:
His point was a logical one; or, put another way, his point was a metaphysical one: How theories are adjusted in the light of contrary observational evidence is a human affair, a decision that we, and not nature, make. So, as a matter of mere logical consistency, it would always be possible to render consistent any theory with any seemingly disconfirming evidence by making enough adjustments in other beliefs.
Quine was concerned to insist on this as an expression of his deep commitment to pragmatism. How we adjust theories to accomodate contrary observational evidence is a highly pragmatic affair: We adjust so as to maximize fertility of the resulting theory in generating new predictions, to maximize consistency with as much of our former system of beliefs as possible, to maximize the overall simplicity of the resulting theory, and to maximize the theory's modesty so that its claims are as narrowly drawn as possible. (65f.)
[see additional text here p. 66 - for these as necessary conditions, according to pragmatists, if we are to "track the truth".]
There are supporting examples - first of all, in the history of physics - which support this claim, e.g., the shift from Newtonian physics to Einsteinian physics:
...the observational evidence for the relativity of space and time did not logically falsify Newtonian absolute space and time, for absolute space and time could be saved by postulating the existence of bizarre "universal forces" that bend light rays, shrink measuring rods, and slow down clocks, forces which by their intrinsic nature are forever and ever undetectabl(because they affect any experiemental device desightedto detect them in such a way as to remain undetectable). Here was a case of two incompatible theories each consistent with all actual and possible observational evidence. [but the evidence did not force a choice between the two theories] (66)
[We have also seen the debate between Ptolemy and Copernicus as an equally classic case of this situation - i.e., where two competing theories explain the same data/observation/phenomena, so that the data do not force us to choose between the two.]
This situation is described by physicists as the underdetermination of theory by the observational evidence, "...the evidence cannot by itself determine that some one of the host of competing theories is the correct one." (66; emphasis added, CE)
If this is correct - then we're forced to turn to additional methodological principles "to break the competitive deadlock between the evidentially equivalent competing theories." (66) As we have seen, Quine proposes his pragmatic principles - but we have also seen the effort to use Ockham's razor, etc. As we have also seen, introducing such additional principles, however, inevitably forces us to make both metaphysical and epistemological assumptions about "nature" which underlie these principles - which
a) runs us into the domain of philosophy (as the domain that argues competing metaphysical and epistemological claims), and
b) confronts us immediately with the danger of circularity:
at issue: which theory gives us the best "picture" or account of "nature/reality"?
but: the observational evidence does not force us to choose, so that ...
we turn instead to methodological principles which rest in turn on
philosophical assumptions about "nature/reality" (metaphysics/ontology) as well as how we are to know that "nature/reality" (epistemology);
BUT: if we use a given set of philosophical assumptions about "nature/reality" to decide - via specific methodological principles - between competing theories about "nature/reality" -
THEN: the resulting theory of "nature/reality" issues from
what we already assumed to be true about "nature/reality" in the philosophical assumptions about "nature/reality" at work
in our choice of the methodological principles needed to break the deadlock between "evidentially equivalent competing theories."
In other terms:
we run the danger of using our assumptions about nature/reality (on the philosophical and then the methodological level)
to decide what theory about nature/reality we are to believe (when the evidence does not force a choice).
[We have seen examples of this throughout the history of science - e.g., Einstein assuming local reality in his development of the EPR paradox which seeks to demonstrate the truth of local reality and disprove the assumption of nonlocal reality in quantum mechanics.]
For Klee, moreover, the underdetermination of theory and the Quine-Duhem thesis regarding what we do when theory is not decided straightforwardly by appeal to observation and data also threaten two specific positions within the philosophy of science:
scientific realism (to be dealt with in ch. 10)
the "so-called hypothetical-deductive method," affiliated with Karl Popper's falsification criterion. (67)
Popper, working out of the Vienna Circle and positivism, developed his falsification criterion, as we have discussed, in order to meet a central problem in both the effort to confirm scientific hypotheses and to demarcate natural science from various pseudo-sciences (first of all, Marxism, Freudian psychoanalysis, and Adlerian psychology).
There is a central problem, as we have seen, with the effort to appeal to experience/experiment to "demonstrate" or "prove" a hypothesis - i.e., it falls into the fallacy of affirming the consequent:
If H [Hypothesis] is true, then E [experimental results] should follow.
E follows
Therefore, H is true.
BUT: how about if we turn this approach upside-down, and look for falsifying or disconfirming evidence? Popper's claim was that science is legitimate knowledge only if it can be falsified or disconfirmed, following the deductive form of modus tollens:
If H [Hypothesis] is true, then E [experimental results] should follow.
E does not follow.
Therefore, H cannot be true.
As attractive as Popper's effort to save the positivist program
from critiques of the sort raised by the Quine-Duhem thesis may
be (and as useful as this version of "the scientific method"
may be as a simplified introduction to science?) -
Klee's example of the experiment to determine whether or not there
are natural killer (NK) cells in a mammalian immune system makes
the point that Popper's theory of science is too simple,
and this in at least two ways:
a) Plato was right: you can't "move down" from a general principle (the hypothesis) to a specific deductive consequence (the observable prediction) without a sequence of intervening steps:
The immune surveillance theory of cancer does not by itself imply any observational prediction, and even Popper admitted as much about any general theoretical hypothesis (that is, he admitted that no general hypothesis by itself implies anything about an actual particular state of affairs. (72)
[To state it baldly: we cannot establish the first line of the modus ponens/modus tollens argument - the move from a general principle (hypothesis) to a specific observable (experiment) - without additional assumptions...] "What implies an observational prediction is that theory together with a myriad of interdependent beliefs, presumptions, guesses, and other theories." (72)
But this further means that...
b) ...the modus ponens part of the Popperian model of scientific method involves a complex conditional," - i.e.,
not H implies E (or, as Klee puts it, theory T implies observational prediction O)
but rather
H plus theory R plus theory M plus belief B plus assumption S plus assumption Y plus hunch C plus... imply observational prediction O.
But this means that
...the modus tollens part of the Popperian model of scientific method is logically undermined. From the falsity of observational prediction O we can no longer automatically infer that it is theory T that is false. All we may legitimately infer, as a matter of logic alone, is that at least one of T, R,M, B, S,Y,C, ... is false, and that is just what the underdetermination of theory by evidence is all about....(72)
Part of the further point is that this biological example suggests that the Quine-Duhem thesis is not limited to physics...
Klee points out - and will examine in chs. 8 and 9 - what he takes to be a misuse of the Quine-Duhem thesis and the recognition of underdetermination - namely, the rejection of any demarcation between science and non-science (including, in his example, "The New Age Society for Sublime Consciousness"), and various forms of social constructivism, postmodernism, etc., which use the thesis and the problem of underdetermination to build a case that "science" depends on a variety of nonevidential and "irrational" factors.
Larry Laudan, in particular, attempts to attack the Quine-Duhem
thesis in order to prevent just such problems. If such an attack
were to succeed, obviously, then the various worries that follow
from the thesis would be undermined.
As might be expected (remember the famous Medieval motto: when
in trouble, make a distinction!) - Laudan's critique involves
first of all making some important distinctions, i.e., between:
the simple claim that scientists are psychological inclined to hold onto a theory in the face of disconfirming evidence by adjusting other beliefs | such adjustments are justifiable, even rational, in the effort to "save" a theory |
"the evidence does not deductively entail a single correct theoretical account of the phenomena in question" (74) | even if we include deductive/probabilistic inference - the evidence still fails to imply a single correct theoretical account |
underdetermination = "any theory can be rendered logically consistent with any evidence that otherwise would seem to contradict it (by adjusting suitably elsewhere)." (74) | in addition to achieving logical consistency - "evidence underdetermines theory in the sense that any theory can be made so as to entail logically any evidence that otherwise would seem to contradict it." (74) |
given any theory and body of evidence - "there will be at least one other theory logically incompatible with the first theory which also is either compatible with or logically entails that evidence." (75) | "...there will be an infinity of such competing theories....All theories are epistemologically equal." (75) |
Laudan argues that radical critiques and rejection of science, resting on the Quine-Duhem thesis, must assume the stronger sort of underdetermination. Klee argues, however, that critics, while holding to egalitarian underdetermination, do not hold to the stronger version of the first three categories (see 75f.). As he points out, this raises the possibility that Laudan is attacking a straw person. (76)
QUINE (remember Quine?), however, holds to the weaker versions - descriptive, compatibilist, and nonunique: and he leaves indeterminate whether he holds to deductive or ampliative underdetermination.
In any case, the weak forms of underdetermination are apparently harmless for the claim that science constitutes objective, reliable, universally valid knowledge. In particular, the nonunique (vs. the egalitarian) view is harmless "because it cannot support the sorts of global attacks on scientific methdology that critics of science mount against it....[where these attacks amount to] the claim that science is a 'waste of time' in the sense that is methods cannot in the end guarantee a rationally justified choice between rival Theories of Everything."(76)
Klee also sees that Lauden appears to agree with Quine in his characterization of the compatibilist version of underdetermination as trivial - that is, compatibilist underdetermination does not lead to the "anything goes" model of scientific practice: on the contrary,
Quine is not an epistemological anarchist. Quine held that we apply pragmatic principles of choice to pare down competing theories....[e.g.,] simplicity, fertility, modesty, and conservativism. (77)
As well,
As a pragmatist Quine thinks that answering the question of whether our scientific theories accurately represent an inquiry-independent reality can be postponed anyway. What matters in the present is that we have a canon of methods, an epistemology of daily practice, which underwites enormous condifence in the claims of mature scientific theories....[Indeed] Quine thinks that those who believe that science truthfully represents an inquiry-independent reality have good cause to be encouraged even in the face of the Quine-Duhem Thesis. (77)
This hope rests in...
Are there NK cells in the immune systems of mammals? Pragmatism is a philosophical view that would answer this question with a hypothetical statement: There are NK cells in the immune systems of mammals if positing the existence of such cells maximally simplifies, renders maximally fertile, conservative, and modest, the sensory experiences (and our theoretical reflections on them) which we undergo in our immunological inquiries, as compared to what would result on those same criteria if we did not posit their existence.
In other words, there are NK cells if the NK-cell hypothesis "works." Nature, after all, pushes back against our theorizing, and if a hypothesis fails to do the work for us that we want it to do [i.e., explain and predict {instrumentalism} and perhaps accurately describe "reality" {realism}], that is a rational ground for becoming suspicious that the hypothesis might be barking up the wrong tree, ontologically speaking. (77)
This leads us (surprise, surprise) to the thorny philosophical issues of epistemology and ontology - and specifically to the issues of instrumentalism vs. realism. Quine, as Klee points out, uses the term "posit" as a verb, derived from its noun usage. For Quine, I "posit" (or, as Klee suggests as a rough equivalent, postulate) an entity such as an electron as part of a long-term process of theory-building which is checked by how well it "works" (this is left somewhat vague by Klee, if not by Quine). Posits which fail to work are discarded.
But what does this mean? Klee points out that pragmatists argue that the notion of posits "working" consistently as "the only evidence there could be that there are such things as electrons, that they exist as inquiry-independent entities that push back against our theorizing." (78)
Klee implicitly acknowledges the ambiguity of what "working" might mean - we must determine more precisely what we expect of a theory when we say it "works" or doesn't (the function of the following chapter). Still, at this stage we can note that this pragmatist point is at least partly correct: but notice carefully the argument involved:
If there are real, independent realities "X";
and if I posit a theoretical description of such "X" which, in the context of a larger theory, leads to observable predictions [H --> O]
then my capturing these independent realities in my theory (e.g., electrons) will lead to the (real) observable results predicted by the theory which posits these realities.
[Using Klee's example: if electrons really exist - and if I accurately posit such entities in my theory - then my theory should predict results that follow precisely because my theory has succeeded in "capturing" "reality."]
So far, no quarrel; quite true. But this is quite distinct from the claim that:
if my observations are consistent with what my theory has predicted (i.e., it "works") - I can conclude that the posits/hypotheses upon which my theory rests are real (ontological realism) and true (epistemological realism).
Indeed, as we have seen, such a claim is the classical fallacy of affirming the consequent.
Still more broadly: the fact that a theory and its posits may "work" does not commit us necessarily to realism (as the pragmatist might wish), but only to instrumentalism (I can only say the theory "works" as a successful "map" for predicting events - I cannot justifiably claim that it further represents a complete and accurate "picture" of "reality").
Klee acknowledges just this point as he characterizes pragmatism as attempting to make "an end run" around just the question of whether science provides us with a realist "picture" of "an inquiry-independent reality." (78)
"It all comes out in the doing," the pragmatist claims. We learn by doing, we know by doing, we believe by doing, so only by doing will these "metaphysical" questions be answered. What we must resist is the attempt to freeze knowledge in its current imperfect state while we try to read off of it what the ultimate Theory of Everything will contain. But a critic will wish of course to challenge the pragmatist's seemingly simplistic equation of what works with what is the case. Suppose some theory T works in all the senses that a pragmatist like Quine thinks count. Why should that be taken as a ground for thinking T is a true representation of some inquiry-independent reality? Why assume that workable theories track the objective truth? ....[In contrast with Quine's misconstrual of the point here,] The critic is claiming that workable theories don't have to work because they track the truth: They could do good work for us and still be false. (79)
In short - so far, there's nothing in the pragmatist account that pushes us towards realism, away from instrumentalism.