This is Philosophy of Science. Franz-Peter Griesmaier

contexts, both the claim that the universe is vastly bigger than usually estimated, and that phlogiston has negative, rather than the standardly assumed positive, weight, are hopelessly ad hoc. It’s just that Copernicus got lucky and was proven right by subsequent developments, while Priestley had massively bad luck and now looks like a fool.

Could their respective fates have been predicted at the time? If Popper is right about legitimate vs. illegitimate modifications in the face of observational counterevidence, it would have to be the case that Copernicus’ modification was arguably progressive, while Priestley’s was not. Remember that a modification is progressive to the extent to which the modified theory has more content, and thus faces more potential falsifiers, than its predecessor. Copernicus’ modification postulated a very large universe. Priestley’s modification postulated a substance with negative weight. Regarding Copernicus, it is difficult to determine whether heliocentrism in a large universe has more falsifiers than heliocentrism in a small universe. Thus, it is difficult to say whether this modification was indeed progressive in Popper’s sense. The difficulty might reside in Popper’s choice of the “scientific unit” that is to be judged as progressive or not. The unit for Popper is an individual theory. His student, Hungarian-born Imre Lakatos, proposed using a wider unit, viz., a research programme. Perhaps Copernicus’ modification can be seen as progressive in virtue of being embedded in a progressive research programme, as we’ll discuss in Chapter 15 on scientific progress.

      Let’s now turn to Priestley’s negative weight. There is one sense in which this proposed modification is not progressive, but rather badly ad hoc. Phlogiston was supposed to be the only substance that has negative weight. But could this hypothesis at least be falsified? In Copernicus’ case, the development of increasingly powerful telescopes provided good empirical reasons for adjusting the estimates about the size of the universe upward, until, by current estimates, we arrived at 91 billion light years. In principle, it could also have been falsified. This would seem to be impossible in the case of phlogiston. The obvious way would be to isolate phlogiston and then try to weigh it – but with what? We don’t have instruments for determining the value of negative weight. Moreover, with negative weight, would phlogiston also have to have negative mass? (Remember, weight is simply a function of mass and the gravitational constant). What would that be? Sure, the negative weight idea saved the theory from the mercury counterexample. But it would have been quite obvious that independent evidence was elusive if not impossible. In light of this, perhaps the right verdict is to say that the statement “Phlogiston has negative weight” looks quite a bit like the statement “The absolute is beautiful.” Neither one can in any clear way be falsified. Thus, by Popper’s criterion for progressive modifications, Priestley’s modification fails, as it introduces a nonfalsifiable, and thus merely protective, hypothesis.

      3.4 Conclusion

This completes our discussion of the use of evidence. We discussed the role of evidence in theory generation and seen that much of this role there depends on a scientist’s psychology and other factors. Theory confirmation, on the other hand, is supposed to be independent of such psychological factors. However, there are certain logical complications that afflict the relation between theory and evidence (e.g., the raven paradox). Popper’s falsificationism has promise to help with some of those problems. Whether his picture of science as a process of critical inquiry that discards theories one by one can be generally accepted is doubtful. The question will be revisited in Chapter 15. But for now we delve further into evidence to consider the ways in which it can be evaluated and why sharing it with fellow scientists doesn’t necessarily resolve disagreements among them.

      Notes

      1 1 Richard Swinburne, “The Paradoxes of Confirmation – A Survey,” American Philosophical Quarterly, 1971, Vol. 8, 318–30.

      2 2 French original published in 1914; English translation in 1954 by Princeton University Press. Quote from p. 199 f.

      Annotated Bibliography

      Vincenzo Crupi, 2020, “Confirmation,” The Stanford Encyclopedia pf Philosophy. Available at https://plato.stanford.edu/entries/confirmation A detailed presentation of many of the technical problems surrounding the notion of confirmation. It also includes many more details on Hempel’s model.

      Brandon Fitelson and James Hawthorne, 2010, “How Bayesian Confirmation Theory Handles the Paradox of the Ravens,” in E. Eells and James H. Fetzer (eds.), The Place of Probability in Science, Boston Studies in the Philosophy of Science 284. Available at http://fitelson.org/ravens.pdf. A thorough discussion of the raven paradox from the perspective of Bayesian confirmation theory, arguing that emphasizing the important difference between no confirmation at all and a small amount of confirmation resolves the paradox.

      Carl Gustav Hempel, 1945, “Studies in the Theory of Confirmation,” Mind 54(213): 1–26 and 54(214): 97–121. In this groundbreaking paper, Hempel develops his model of confirmation and introduces the Raven Paradox.

      Karl Popper, 1934/1959, The Logic of Scientific Discovery. London and New York: Routledge 2002. The classic statement of falsificationism, this is one of the most influential books in the philosophy of science. Popper argues for a decisive break with attempts to develop a model of confirmation, replacing it with a process involving bold conjectures and severe testing.

      Willard van Orman Quine, 1951, “Two Dogmas of Empiricism,” Philosophical Review 60: 20–43. Among other important contributions, this seminal essay introduces the idea of confirmation holism.