Popper

Demarcation

• Demarcation criterion that enables us to tell what is science and what is not. He started with the opinion that physics is science, but Freudian psychology, Marxist theory of history, and astrology are not. Freudianism and Marxism had converts with a religious fervor. The true believers could explain absolutely anything in Freudian or Marxist terms. A scientific theory "takes risks": It makes predictions that can be shown to be incorrect. Slogan: A theory is scientific if it is falsifiable.

• There is a reason Carnap and friends can't come up with a theory of confirmation: Scientific theories cannot be confirmed, only falsified.
  • Popper can rest content with the idea that the only logic is deductive logic. No observations of crows can show that all crows are black, but even one observation of a purple crow shows, logically proves, that NOT all crows are black.

• What makes a scientific theory a good theory, that is, one worth adopting? It is worth adopting if it has survived great risks, stringent tests, attempts to falsify it that look like they should have worked. A more positive way to put this, one that Popper is not completely entitled to, is, a theory is worth adopting if it has made surprising predictions.

• According to falsificationism, the only experiments worth doing are ones suggested by a theory. Popper denies the common idea that the data, observations, suggest our theories. Theories are outright creations of human imagination.

• Though Popper's view is generally thought of (correctly) as a rival to logical positivism, note that they have in common
  • The belief that science serves as an example of the best way we have of acquiring knowledge,
  • Empiricism.

---

The theory I just presented is often called naive falsificationism. Popper was in fact a sophisticated falsificationist. The chief problem is holism. Every experiment involves more than one theory, and so no experiment by itself does actually falsify a theory. The idea that only deductive logic is required seems to be wrong, but experiments, plus logic, plus some good judgment of a sort we are used to exercising, can falsify theories. It isn't simply a theory that is scientific or not, but a theory plus an attitude toward it. Popper's later version of his demarcation criterion is one that applies not to theories, but to the practices and procedures employed by researchers: a researcher is doing science if that researcher takes an attitude toward theories that makes them susceptible to empirical refutation. That represents a big change in the proposed subject matter of the philosophy of science and one that many of us would regard as an important advance.

Probability poses a related problem—what should Popper say about cases in which one seems to have shown, not that a conjecture is false, but that it is unlikely to be true.

Though deductive logic remains the only logic in Popper's "sophisticated" view, it doesn't explain what scientists do, or should do, in the way it seemed to in naive falsificationism.

The most serious problem, the one generally regarded as fatal, is that we never have any reason to believe any theory.

Popper tended to associate his conclusion that we never have any reason to believe any theory with something much less controversial: fallibilism, the widely accepted view that we can never be completely certain about factual matters, but his conclusion is much stronger—it is not unreasonable to take it to be a consequence of his views that we must always be completely uncertain about factual matters! In what way is a theory that has survived many tests better than a new one that has never been tested at all? Popper doesn't have a satisfying answer. (In fairness, note that advocates of inductive confirmation are unable to provide a justification, and so they are not much better off.)

The problems outlined for falsificationism are problems for falsificationism considered as a philosophical theory of how to determine which scientific theories are good—the evaluation problem. They are not problems for falsificationism considered simply as an answer to the demarcation problem: which theories are candidates to be scientific theories. To be a theory of scientific type, according to Popper, there have to be some kinds of empirical results that would lead you to abandon the theory.

Like the positivists, Popper is insisting that tenable theories must be associated with some empirical content. He reverses the experiment-theory order: roughly, for him, experiments, to be interesting, must be suggested by theory (the theory gives rise to the experiment). It is not the case that we devise theories to account for the results of experiment.

The two psycho-analytic theories ... were simply non-testable, irrefutable. ... This does not mean that Freud and Adler were not seeing certain things correctly. ... much of what they say is of considerable importance, and may well play its part one day in a psychological science which is testable. 69

One of the hardest things to do with a theory is to get it to make any predictions. A theory that is unscientific may, if suitably developed, become scientific.

Two examples:

1. Boyle vs Hobbes on the vacuum.
2. String theory

The current status of string theory is that, using an approximation to the actual theory and massive amounts of computing power to model the approximation, you can, with limited success, simulate the properties of the vaccuum (that is, of empty space).

The logical positivists rejected as nonsense anything without empirical content. Popper, who put theory first, recognizes that an unscientific theory can be a useful precursor to or part of or suggestive about, a subsequent scientific theory.

Popper abandons the semantic theory of the received view, but he persists with the idea that theories are central to science and that scientific theories must be sensitive to empirical outcomes.

---

The study of any of them seemed to have the effect of an intellectual conversion or revelation, opening your eyes to a new truth hidden from those not yet initiated. Once your eyes were thus opened you saw confirming instances everywhere 67

Only a theory that can explain "everything" can have such an effect. The reason I'm calling attention to this passage is for later reference: It is nearly verbatim how Kuhn describes the normal attitude of scientists toward the scientific theories with which they work.

Evaluation

According to Popper, science proceeds via a cycle of conjecture and refutation. A good theory (conjecture) is one that is not ad hoc, one that doesn't merely avoid known problems (refutations) of previous theories, but one that leads to bold, unexpected, surprising predictions. Popper always described the process as if it were to be carried out by a single individual, who should therefore be skeptical and test ideas. As we shall see later, the scientific community is structured to guarantee that something like that happens as a social process—scientists attempt to refute the theories of their rivals. There is a strong analogy between the cycle of conjecture and refutation and Darwin's theory of variation and natural selection.

Examples

• A bold conjecture, and the failure to refute it

My example was used by Kuhn (whom we shall study in the next lecture).

My account relies on

I should note that the main purpose of Worrall's article is to show that the use I have made of the story of Poisson's spot (which is a standard use) relies on a misinterpretation of what actually occurred. Worrall may well be right. I am not relying on the historical correctness of the example here but on the use of the story to illustrate what Popper means by a bold prediction.

In the early part of the 1800s, there were only two basic theories of what light could be: light either consisted of waves or of particles, and, moreover, the wave theory was taken to have long been discredited.

In 1819, Fresnel, then a young war hero unknown to the scientific community, submitted a paper to the prestigious French Academy for a prize competition concerning diffraction. Most of the members of the prize committee supported the particle theory, and the point of offering the prize may well have been to get someone to come up with a way of accounting for diffraction using particles. In his paper, Fresnel attempted to show how to account for diffraction phenomena in detail on the basis of a wave theory.

• • Aside: "Diffraction phenomena" are those we today ascribe to overlapping waves—which can form complex patterns, as anyone who has ever splashed about in a bathtub can attest. To observe diffraction, hold your pointer and middle fingers together in a normal, relaxed position, put them up to your eye, and look at a light through the narrow space between them. You will see the light, but you will also see dark lines ("fringes") in front of the light that appear to be between your fingers. Those are a diffraction pattern. You could actually learn a fair bit about diffraction using just that. For example, how do the fringes differ when you look at different colors of light?

Though Fresnel's theory was about diffraction in general, we shall only be interested in one part: how his theory accounts for shadows. Poisson, a member of the prize committee, noted that the shadow of a small disk cast by a tight beam of light will have a bright dot in the middle. That obviously absurd result seemed to torpedo Fresnel's theory. Arago, a member of the committee who supported Fresnel, actually went to the trouble of testing what happens, and, to everyone's surprise, the bright spot is there. It is, today, known as Poisson's spot, and you can easily find pictures of it by googling. The unexpected result was so decisive that, within a few years, the particle theory was all but dead, replaced by the wave theory.

• An ad hoc theory

According to Newton's theory, the planet Mercury should orbit the sun in an unchanging ellipse. In fact, Mercury's orbit is more complex: it moves along a rotating ellipse. Why does the orbital ellipse rotate? Astronomers looked for a new planet, Vulcan, smaller than Mercury and with an orbit closer to the sun to answer the question: the gravitational force of Vulcan on Mercury could account for the extra motion. Astronomers looked for such a planet without confirmed success. Einstein's theory of gravitation explains the extra motion without Vulcan: it arises, roughly, from the fact that the gravitational force propagates from the sun to Mercury with finite speed, instead of instantaneously as it does according to Newton. Once that was widely understood, the search for Vulcan was abandoned.

There is another way to explain the rotation of the orbital ellipse of Mercury: If Newton's theory is left intact except for the following tiny change, everything works out:

According to Newton,

That says: the gravitation force between two objects is proportional to the product of their masses, and inversely proportional to the square of the distance between them.

According to the new theory,

(I made up the number 1.9997—I haven't bothered to look up the actual proposed value.) That theory, it is said, works as well as Einstein's (that is, there is no evidence that favors one over the other), and it is much simpler. Advocates of the theory typically take themselves to have overturned Einstein. What is wrong with the theory? It is ad hoc: it doesn't make any "new" predictions, having simply been concocted to take care of the old ones.

Experiment

On both the received view and Popper's view, experiment is the poor cousin of THEORY. On the received view, experiment exists to verify theories, and to give a simple, direct meaning to observational terms. It is theory, and the complex definitions of theoretical terms that gets you somewhere. On Popper's view, experiments are the creatures of theory: one devises experiments as a way of testing theories.

Later, we'll see that experiment has a life of its own. Most experimentalists view theory and theorists as being an unfortunate fact of life they have to live with, not as having that all much to do with "real" science. ("Real" in the macho sense of the word: "real scientists do experiments.")

-- ShaughanLavine - 19 Sep 2005 - 17 Sep 2007