Quine considered his position to be a naturalistic one, and so not opposed to science as at least a good pragmatic form of knowledge. But I am unclear on the premise of his thesis of underdetermination. Why did he feel it necessary to challenge traditional logical positivism? Was there some crisis in the philosophy of science he was answering? It seems like a strange esoteric point to make - that there 'must' be alternative and equivalent theories. He didn't, from what I can tell, demonstrate it analytically or empirically. Or was he just making a point about how underlying subsidiary hypotheses are really what are driving the ideas?

I basically hold to Bergström's view that there has never really been a case in the history of science were multiple theories really were completely empirically equivalent. But my question is more to the underlying reason for generating the idea - Quine wasn't a scientific skeptic, from what I can tell and even if it is in some sense logically or theoretically true, does it matter in any meaningful way?


2 Answers 2


The Quine-Duhem Thesis was proposed by Quine as a critique of Popper's Falsificationism criteria for science.

IF every hypothesis is infinitely patchable, THEN it is impossible in principle to falsify any hypothesis.

Popper accepted this critique of falsifications, and abandoned explicit falsificationism as central to science. In his later years, he treated hypotheses as complex statements, with core assumptions, and ancillary ones, and the ancillary ones can be added or tweaked to deal with every falsifying test case. Popper then advocated for a version of Occam's Razor to sort between competing hypothesis collectives. As Occam is notably subject to rationalization, he suggested a somewhat different approach, where "predictive power" rather than "simplicity" was the Razor. The better hypothesis suite is the one that predicts many MORE things that could not be so.

Lakatos carried this central and ancillary distinction into his somewhat better post-Quine-Duhem approach. Note Lakatos' Research Programmes explicitly holds that multiple contemporary approaches to a problem are viable.

  • Very strange then that Popper, in effect, derailed Solomonoff's project at the dawn of Moore's Law and once having accomplished the feat of defeating a critical advance in the natural sciences -- most critically in the selection of social causation from the increasing wealth of data, only later admitted his mistake but without rehabilitating the Algorithmic Information Criterion for causal model selection. And so now we have Soros and an impending civil war over social theories. Apr 9 at 12:39
  • @JamesBowery -- There is an answer that was accepted that simplicity is computational. but I challenged that with references: philosophy.stackexchange.com/questions/93490/… Are you able to refute my counter answer? The war between democracy and the Authoritarians features the Authoritarians reviving the Biblical embrace of enforcement of One True Faith by the secular king, with the True Faith then justifying Authority. Soros is the most prominent defender of democracy worldwide against theocratic inspired authoritarianism.
    – Dcleve
    Apr 9 at 15:19
  • Choice of Turing machine must be made prior to the presentation of the set of observations otherwise it is yet another case of patching up theories in a posthoc fashion to avoid falsification. This is such a perennial fallacy in critiques of the Algorithmic Information Criterion for causal model selection as to qualify as a "philosophical nuisance". That is has been so simply dispatched so many times takes it beyond a mere nuisance to a philosophical malefactor. Apr 9 at 17:39
  • If you insist formalization of the argument, see the Gödelian approach to formalizing the aforementioned dispatch: NiNOR Complexity: jimbowery.blogspot.com/2023/10/ninor-complexity.html Apr 9 at 17:43
  • As to the selection of data being subjective, that is true and not to the point. Until Soros et al fund something like Hume's Guillotine: github.com/jabowery/HumesGuillotine Or, failing that, support Sortocracy: sortocracy.org They are working against their stated aims. Apr 9 at 17:54

I basically hold to Bergström's view that there has never really been a case in the history of science were multiple theories really were completely empirically equivalent.

I think the Galileo affair was one such example. At the time, there were two alternative cosmologies -- the heliocentric, advocated by Galileo perhaps too passionately; and the geocentric, particularly the model proposed decades earlier by Tycho Brahe. Each was offered as an explanation of the movement of the celestial objects. Galileo insisted that the heliocentric model was representative of the truth, but he did not have an empirical proof. This eventually lead to his legal troubles with the Church.

A more modern example would be the different interpretations of Quantum Mechanics, or at some point there were several competing versions of the string theory. The string theory itself has a competitor in the loop quantum gravity. Or think of different models of consciousness. Basically, science assumes1 that the Reality has to work, objectively, in some specific way (a.k.a. the objective/scientific truth). But until we have a definite proof one way or the other, we could have several competing theories/models of how the Reality might work.

Further update: According to the Britannica article on underdetermination,

Alternatively, one can interpret it as proposing that all the criteria of rationality and scientific method permit some means of protecting the favoured hypothesis from the apparently refuting results.

This is a fascinating proposition. Indeed, this is exactly what Galileo was doing -- defending, at the risk to his own life, the heliocentric cosmology against the apparently refuting evidence.2 This goes to the heart of the age-old question: What is truth? And if a scientific theory cannot be, generally speaking, falsified by the evidence, then what is the point of doing science?

As I understand it, the assumption behind science that our one and only objective Reality can be understood as a machine, as a very complex clockwork. To that end, scientists try to discover a model of how that machine of the Reality works under the hood -- a model of the processes driving the observable patterns (patterns like the apparent movements of the Sun, the Moon, of the stars and planets). Quine's point could be that those models don't need to be perfect, neither accounting nor being consistent with every available piece of evidence (with every available observation). Rather, we choose a model that offers what we feel is the best explanation of the available evidence.

Sometimes we might have multiple equally (un)convincing theories -- and then choosing the winner becomes as much a matter of personal taste,3 as of the amount of the evidence one is able to take into consideration. To Galileo, the discovery of moons orbiting Jupiter, as well as the mountains on the Moon, was a clear indication that there is no substantial difference between Earth and the celestial bodies -- that Earth, therefore, is but another planet orbiting the Sun. This was his truth, and as someone who dedicated his life to the pursuit of the truth, he didn't -- he couldn't -- back down. He could not un-know the truth he has uncovered, something he came to know for a fact.

1 Or, rather, this is the assumption on which the science is based upon.

2 For example, due to the primitive optics if his day, the stars appeared as disks when looking through a telescope, suggesting that they cannot be much farther from us than the planets. If that was true, then the movement of earth around the Sun would create a parallax -- an apparent movement of some stars against the backdrop of others. Yet, no star parallax could be observed.

3 The British physicist Paul Dirac, for example, claimed that it is more important that a theory be beautiful than that it conform with experiment, while Einstein stated that "the only physical theories that we are willing to accept are the beautiful ones." (link)

  • I get that, at least for a time, they were mathematically equivalent explanations (although I think that in some ways this is a good example of how theories aren't actually equivalent, given the ad-hoc complexity of the Brahe model). But it was only transiently undetermined.
    – BVinNV
    Feb 22 at 22:04
  • For me, this is exactly how science should progress: there are underdetermined hypotheses and we look for ways to make the determination. It might take a while, but that's OK. Once we have resolved the underdetermination, we might acquire newer technologies that, through better observation, measurement, etc., present another underdetermined debate. And so on, increasing resolution on the underlying phenomenon.
    – BVinNV
    Feb 22 at 22:04

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .