Today I attended a lecture by Massimo Pigliucci at Washington University in St. Louis. The title of the talk was "A Fresh Look at the Demarcation Problem and Why it Matters." Pigliucci’s aim was to help us distinguish between real science and pseudoscience.
He offered some a few examples up front to set the stage. It is fairly well accepted these days that Freudian psychoanalysis is pseudoscience whereas Einstein's theory of relativity is a prototypical example of legitimate science. Most science falls in between these endpoints. One example of a suspect science is string theory, which Pigliucci characterized as a favorite modern day "whipping boy."
Karl Popper had the same objective back in the 1930s, offering his falsifiability approach: a theory should be considered scientific if, and only if, it is falsifiable. Not only did Popper believe that he had provided a method for determining what is truly scientific; he also believed that he had solved David Hume's “problem of induction.” Induction is the process of generalizing from a smaller set to a larger as-yet-unobserved set; the induction problem, according to Hume, was that we cannot rationally justify induction, because this conclusion depends upon the assumption that nature will continue to be uniform. Pigliucci argued that Popper's falsification approach is not sufficient for it distinguishing between pseudoscience and science, because it is “vulnerable to the Duhem-Quine theses.” The problem, according to Pigliucci, is that one can often save a falsifiable hypothesis by tweaking it (as nineteenth century astronomers did when they worked to save Newtonian physics in light of the perturbations of Mercury by positing that there was an as-yet-unseen planet closer to the sun, a planet they named "Vulcan." It would also seem that there is another problem with falsifiability; some nascent fields don't yet have a thick collection of observations with which to work. Imagine that Aristotle announced the theory of general relativity, far before Einstein. It might have been impossible for him to offer a way to falsify his theory back then, but not because there was any problem with the theory itself; it would thus be declared to be not-science. Pigliucci addresses this situation (see below).
Throughout his lecture, Pigliucci referred to Larry Laudan's approach to the demarcation problem. Laudan has argued that the demarcation problem is "uninteresting and intractable," urging that we should completely stop using terms like "pseudoscience" and "unscientific." Laudan argues that philosophers have failed to point out necessary and sufficient criteria for distinguishing between pseudoscience and science. He therefore considers the demarcation project doomed. Pigliucci disagrees, pointing out that it is not necessary to find necessary and sufficient criteria for distinguishing between science and pseudoscience. Pigluicci draws upon Wittgenstein's work on family resemblances. Wittgenstein had challenged people to define the word "game" in terms of necessary and sufficient conditions. It seems like an easy task, but it is not. Pigliucci refers to this exercise as "humbling." Just because we can't set out the necessary and sufficient conditions doesn't mean we don't know what a "game" is. With regard to many concepts, including "species," and, yes, "science," no boundaries have been drawn, yet we have workable ideas for what these concepts are. Wittgenstein’s observations have been recognized and expanded by modern linguists, including George Lakoff, who has labeled such concepts as “radial categories,” indicating that prototypes serve as the most typical instances of such categories.
[More . . . ]