Last modified on Friday, September 18, 1998. Ó Malcolm R. Forster, 1998

The difference between science and non-science has practical ramifications for society:

• Parapsychology includes the study of such alleged phenomena as telepathy, clairvoyance, and precognition. In 1969 the American Association for the Advancement of Science (AAAS) admitted the Parapsychology association as an Affiliate member. Should they have done that?
• In Arkansas, U. S. A., there were attempts to have the biblical story of creation taught in schools alongside evolutionary theory (after earlier attempts to ban evolutionary theory failed). They argued that creationism is a just as much a science, and therefore deserves equal time.
• The Merriam-Webster's Collegiate^® Dictionary defines creation science n (1979): creationism; also: scientific evidence or arguments put forth in support of creationism. Should an authoritative dictionary presuppose that creationism is a science?
• Freudian psychology has a poor reputation in scientific circles. Is it a pseudoscience?
• Immanuel Velikovsky and Erich van Daniken wrote best sellers Worlds in Collision and Chariots of the Gods, which angered many scientists. Are these examples of pseudoscience.
• Thor Heyerdahl launched the Kon-Tiki expedition to support his theory that the polynesians migrated from South America. Was he a pseudoscientist?
• Gould wrote a book called The Mismeasure of Man about the IQ debate, and phrenology, which purported to predict the criminal nature of people from their skull shape and other characteristics.
• IQ testing has been used to screen children from entering high school, or college, in many countries for many years. Does it predict future academic performance reliably? Is there really such a thing as intelligence?
• Chemistry grew out of alchemy. One’s a science and the other is not. What’s the difference?
• The label 'science' carries a high degree of authority, and people need to understand when the label, and the authority, are deserved. Is there any clear difference between an unscientific study and a scientific one?
• More generally, an understanding of what science is carries us a step closer to telling the difference between good and bad science, and the limits of good science. If we understand how science works, we can be better and more informed use of scientific expertise.

If we wanted to know which subjects were generally accepted as science, we would probably find a fairly sharp and clear division between two categories. But we are interested in more than that! We want to understand the general characteristics of science that are different from pseudoscience. That is actually surprisingly difficult and controversial.

Exercise: Critically evaluate the following characterization of science (from the Encyclopedia Britannica): any system of knowledge that is concerned with the physical world and its phenomena and that entails unbiased observations and systematic experimentation. In general, a science involves a pursuit of knowledge covering general truths or the operations of fundamental laws.

The key to understanding Popper's demarcation criterion is to compare two examples. The first, Popper thinks is typical of science, while the second is typical of pseudoscience.

Example (a): Einstein's prediction of the bending of star light. For over 200 years prior to Einstein, Newtonian physics had enjoyed a period of unprecedented success in science. Many scientists thought that Newton's theory was the end of science, and many philosophers not only believed that Newton's theory was true, they thought that it was necessarily true. They sought to explain why Newton's theory had to be true. All that began to change with Planck's 1900 introduction of the idea that energy comes in small discrete packages (the quantum hypothesis) and Einstein's discovery of the special theory of relativity in 1905. Einstein's special theory of relativity was a way of reconciling some inconsistencies between the wave theory of light and Newtonian mechanics. Instead of modifying the wave theory, he modified some of the fundamental assumptions used in Newtonian physics (like the assumption that simultaneity did not depend on a frame of reference, and that the mass does not depend on its velocity). However, Einstein's special theory of relativity said nothing about gravity. Einstein's general theory of relativity was his theory of gravitation, which he had published by 1916. Many scientists were impressed by the aesthetic beauty of Einstein's principles, but it was also important that it be tested by observation. For most everyday phenomena, in which velocities are far smaller than the speed of light, there is no detectable difference between Einstein's prediction and Newton's prediction. What we needed was a crucial experiment in which Einstein and Newton made different predictions. In 1916, there were successful tests of Einstein's special theory. But crucial tests of the general theory were harder to come by. One such case was provided by the bending of starlight by the gravity of the sun. The period from 1900 to at least 1916 was a period of revolution in physics, and Eddington's confirmation of Einstein's prediction in 1919 helped to complete the change in physics.

"The idea that light should be deflected by passing close to a massive body had been suggested by the British astronomer and geologist John Michell in the 18th century. However, Einstein's general relativity theory predicted twice as much deflection as Newtonian physics. Quick confirmation of Einstein's result came from measuring the direction of a star close to the Sun during an expedition led by the British astronomer Sir Arthur Stanley Eddington to observe the solar eclipse of 1919. Optical determinations of the change of direction of a star are subject to many systematic errors, and far better confirmation of Einstein's general relativity theory has been obtained from measurements of a closely related effect--namely, the increase of the time taken by electromagnetic radiation along a path close to a massive body." (Encyclopedia Britannica)

"The theories involved here were Einstein's general theory of relativity and the Newtonian particle theory of light, which predicted only half the relativistic effect. The conclusion of this exceedingly difficult measurement--that Einstein's theory was followed within the experimental limits of error, which amounted to +/-30 percent--was the signal for worldwide feting of Einstein. If his theory had not appealed aesthetically to those able to appreciate it and if there had been any passionate adherents to the Newtonian view, the scope for error could well have been made the excuse for a long drawn-out struggle, especially since several repetitions at subsequent eclipses did little to improve the accuracy. In this case, then, the desire to believe was easily satisfied. It is gratifying to note that recent advances in radio astronomy have allowed much greater accuracy to be achieved, and Einstein's prediction is now verified within about 1 percent." (Encyclopedia Britannica)

"According to this theory the deflection, which causes the image of a star to appear slightly too far from the Sun's image, amounts to 1.75 seconds of arc at the limb of the Sun and decreases in proportion to the apparent distance from the centre of the solar disk of the star whose light is deflected. This is twice the amount given by the older Newtonian dynamics if light is assumed to have inertial properties. If light does not have such properties, as is generally accepted now, the Newtonian deflection is zero." (Encyclopedia Britannica)

Reconstruction of the example: Philosophers need a general characterization of the example: Let E be a statement of the prediction made by Einstein's theory. E states the direction that the starlight will be observed at the time at which the star was to be observed by Eddington. Let T be a statement of the general principled in Einstein's general theory of relativity. Let A be the conjunction of all auxiliary statements used to derive, or deduce, E from T. That is to say, the argument with T and A as premises, and E as the conclusion, is deductively valid. Symbolically, we may write this as:

T & A Þ E.

A will include assumptions like "the sun is spherical ball of mass M", "there are no other bodies close by to add to the sun's gravitational field," "If the sun were not present, then the star would be seen in the direction such-and-such," "the effect of stellar aberration on the direction of light is such-and-such," and so on.

Example (b): Adler's 'individual psychology'. Compare the following two (hypothetical) explanations of human behavior. (1) E₁: A man pushes a child into the water with the intention of drowning it. (2) E₂: A man sacrifices his life in an attempt to save the child. Popper claims that Adler's 'individual psychology' can explain both of these behaviors with equal ease. Let T be Adler's theory, let A be the auxiliary assumption that the man suffered feelings of inferiority (producing the need to prove to himself that he dared to commit some crime). Then T & A₁ Þ E₁. Let A₂ be the auxiliary assumption that the man suffered feelings of inferiority (producing the need to prove to himself that he dared to rescue the child). Now T & A₂ Þ E₂.

Definition: Let us say that a theory T predicts an event E if and only if there are auxiliary assumptions that have either been used successfully in other predictions, or are the simplest and most obvious assumptions that one would make in the situation, and that T & A Þ E. If there exists auxiliary assumptions such that T & A Þ E, where A is some ad hoc assumption that is introduced in light of the evidence E itself, then theory T merely accommodates E.

In example (a), Einstein's theory predicts the observational evidence, while in example (b), the theory is merely accommodates the evidence.

Popper describes the difference by claiming that Einstein's theory is falsifiable, whereas Adler's theory is not.

Remark: Popper also claims that the problem with Adler's theory is that it is too easily verified: "the world was full of verifications of the theory." Adler may have seen it like that, but was he right? My feeling is that mere accommodations do not count as verifications at all. Hence, I think that a verificationist could account for the difference between these two examples as well as, if not better than, a falsificationist.

Discussion Question: How does our previous distinction between ampliative inference and deductive inference enter into these examples, if at all. Popperians tend to think that there is no need for ampliative inference in science at all. Why might they think that? Are they right?

(Curd and Cover, pages 1-10) To understand a philosophical theory, like Popper's demarcation criterion, it is useful to see why simpler alternative proposals do not work.

Proposal 1: Science is distinguished by its empirical method. That is, science is distinguished from pseudoscience by its use of observational data in making predictions.

Objection: Astrology appeals to observation, but is not a science.

Proposal 2: Scientific theories, like Einstein's, are more precise in their predictions that Adler's psychology, or astrology.

Objection: While it is true that pseudosciences do often protect themselves from refutation by making vague or ambiguous predictions, that is not always the case. The 'predictions' of example (b) are precise enough for the purpose, and Einstein's prediction was not exact—it had to allow for many errors of observation.

Proposal 3: Science is explanatory, whereas pseudoscience is not.

Objection: If you buy into the auxiliary assumptions in Adler's psychology, then the theory explains the phenomena perfectly well. It is true we have little reason to believe that the explanation is correct, but that is a different issue.

Proposal 4: Science is distinguished from pseudoscience by its verifications, or confirmation.

Objection: Popper's objection is that "The world was full of verifications of those theories." I have remarked that that does not ring true in examples (a) and (b). Nevertheless, there seems to be some force behind Popper's point in other examples. For example, Einstein could have pointed to all the verifications of Newton's theory for low velocities and claim these as verifications for his own theory. Yet he did not. Why not? Because, says Popper, these were not risky predictions. They were not potential falsifiers of Einstein's theory.

Popper’s Proposal: Every ‘good’ scientific theory is a prohibition: it forbids certain things to happen. The criterion of the scientific status of a theory is its falsifiability, or refutability, or testability.

Note: Popper also anticipates a major objection to his criterion: namely, that any scientific theory can be protected from refutation by introducing ad hoc auxiliary assumptions. His reply is that the very use of ad hoc assumptions, in reducing the falsifiability of theory, also diminishes its scientific status. The problem with Popper’s reply is that it is not always, if ever, clear in advance that ad hoc auxiliary assumptions are needed to save the theory. This is essentially Kuhn’s point.

In the first set of lecture notes, I introduced the demarcation problem as a problem about the difference between good and bad kinds of ampliative inference. Popper rejects this formulation of the problem. He thinks it is wrong to think of theories as being inferred from, or induced from, the observational facts. Rather, the invention of theories is a question of psychology, which has nothing to do with the status of the theory as scientific. There are no scientific or unscientific ways of inventing theories. They can come in a dream or they can be constructed from the data—it does not matter. Rather, the essence of science is about how predictions are deduced from the theories. This way of viewing science is known as hypothetico-deductivism. The difference between science and pseudoscience rests solely on the 'deductive' part of the process.

Thomas Kuhn makes the following points against Popper (Curd and Cover, pages 11-19):

• The kind of examples Popper considers, like the 1919 test of Einstein’s theory of gravitation, is an example of extraordinary science, or revolutionary science. These are relatively rare in science.
• In non-revolutionary science, or normal science, the aim of research is to connect the experimental data to the background theory, by inventing the appropriate auxiliary assumptions. If a scientist fails, then scientist’s lack of ingenuity is blamed, not the theory.
• It is for normal, not extraordinary, science that scientists are trained, and...
• "If a demarcation criterion exists..., it may lie in that part of science which Sir Karl ignores."

Question: Kuhn concedes that "There is one sort of 'statement' or 'hypothesis' that scientists do repeatedly subject to systematic test. I have in mind statements of an individual’s guesses about the proper way to connect his own research problem with the corpus of accepted scientific knowledge." Thus, thinks (a) a demarcation criterion should refer to normal science, and (b) falsifiability does play a role in normal science. So, why doesn’t he apply a falsifiability criterion to normal science, and say that an alleged science is genuinely scientific if and only if its solutions to puzzles are falsifiable? As Kuhn says, this is not what Popper has in mind. But does the new criterion work?

Central Concepts in Kuhn’s Account of Revolutionary Change in Science: Kuhn denies that theories change by falsification in science, but he does not deny that theories are sometimes replaced (revised). What is his own account of ‘theory replacement’? Here is very brief summary of his positive account:

1. The process by which one paradigm is replaced by another is called revolutionary science.
2. An anomaly is a violation of "the paradigm-induced expectations that govern normal science" (Kuhn, The Structure of Scientific Revolutions, 1970, p.52).
3. A crisis in normal science occurs when puzzle-solving breaks down, either because no solutions are found, or because the discrepancy corrected in one place shows up in another.
4. A paradigm is overthrown only if the paradigm is in crisis and there is a second paradigm that shows equal or greater puzzle-solving potential.

Kuhn’s Demarcation Criterion: All of genuine science has a puzzle-solving tradition, while pseudosciences do not.

Objection: Until Kuhn says what a puzzle-solving tradition is, his criterion is rather vague. Why wouldn’t a research tradition that sought worked backwards from the fact to the auxiliary hypotheses count as puzzle-solving. It seems that Kuhn needs to add something like falsifiability.

Kuhn on Astrology:

1. Kuhn agrees that astrology is pseudoscience, but makes the point that it was not obvious that it was pseudoscience in the century it was practiced most. That is because its auxiliary assumptions, based on the configuration of the planets at the time of birth, were subject to genuine doubt. Not everyone was sure of their exact date of birth in any case. The problem is that similar arguments explaining away failed predictions are regularly used today in medicine or meteorology.
2. Astrology has no science to practice because practitioners had rules to apply, but no puzzles to solve. Most difficulties "were beyond the astrologer’s knowledge, control, and responsibility." In astronomy, however, if a prediction failed, a scientist "could hope to set the situation right." There was a puzzle-solving tradition.

Final argument:

• For a long period of time, there was a sense in which Ptolemaic astronomy was unfalsifiable by the means available (naked-eye observations). But that did not stop it from being science at the time. Moreover, when it was finally falsified (by Galileo’s observation of the phases of Venus, the moon’s of Jupiter, and Brahe’s observations of comets), it had already been rejected.

necessary condition: E.g., being enrolled in this course is a necessary condition for you getting an A for the course. That is, you will not get an A if you are not registered. Or equivalently, if you do get an A, then you are registered. In general: A necessary condition for an event or state of affairs X is one that has to hold for X to be true. A necessary condition is contrasted with a sufficient condition.

sufficient condition: E.g., getting an A in this course is a sufficient condition for passing this course. That is, it you get an A then you will pass. In general: A sufficient condition for an event, or state of affairs X is one that enough to makes X true.

necessary and sufficient condition: A condition is necessary and sufficient for a statement, or event, X if and only if it is necessary for X and sufficient for X. It is often expressed by the phrase ‘if and only if’ or the abbreviation ‘iff’. E.g., a necessary and sufficient condition for passing this course is to receive a passing grade while being registered for the course.

Popper says that the falsifiability is a necessary and sufficient condition for genuine science.

• Falsifiability is not a sufficient condition because astrology is falsifiable but not a science. In the 1960s, Michael Gauquelin examined the careers and times of birth of 25,000 Frenchmen, and found no significant correlation between careers and either sun sign, moon sign, or ascendant sign (see the Thagard reading). Gauquelin found some statistically significant correlations between certain occupations and the positions of certain planets at the time of their birth, but we can expect 1 in 20 random associations will be statistically significant by chance alone. Also, studies of twin do not show the correlation one would expect. These results cannot be explained away by supposing that almost all assumptions in every case were false. Therefore, such a study falsifies astrology in a sense that Popper would accept, and these studies were always possible, which proves that astrology was always falsifiable.
• Falsifiability is not necessary for science. The example of Ptolemaic astronomy shows this, because prior to the invention of telescope it was not falsifiable but it was still a science.

Counterexamples: In order to show that a condition is not a sufficient condition for X, we only need an example in which the condition holds, but X does not. In order to show that a condition is not a necessary condition for X, we only need an example in which X holds but the condition does not. In each case, the examples are called counterexamples. Astrology is a counterexample for the sufficiency of falsifiability for science, and Ptolemaic astronomy is counterexample to its necessity.

Here are some other arguments that make use of counterexamples.

• Astrology is not a science because it has mystical origins. Counterexample: Chemistry had its origins in alchemy and medicine had occult beginnings.
• Astrology is not a science because people believe astrology for irrational reasons. Counterexample: Many people believe in Einstein's theories for irrational reasons.
• Astrology is not a science because it assumes that gravitational influences of the planets influence us, but they are too weak to do that. Counterexamples: The lack of a physical foundation did not stop geologists from believing in continental drift, and we have plenty of evidence to prove that smoking causes lung cancer with knowing the details of the carcinogenesis.