Extinct and Living Approaches to Science

Essentialism

The idea is that an object or process has an "essence" or true underlying value waiting to be discovered. Although this idea is an old one (attributed to Aristotle), it is tenacious. In fields outside of biology and statistical mechanics, it may do little harm, but the idea is especially insidious in biology. The genetics of sexual recombination and environmental variation assure that no two individuals are exactly alike. Therefore, generalizations must be made statistically (e.g., concerning means or higher moments), and variations are of inherent interest. Evolution cannot be understood without studying variation.

Classical Justificationism

The idea is that science should be built on the solid foundation of proven propositions. Unfortunately, nothing can be proven in science, as the collapse of the absolute truth of Newtonian physics amply demonstrates.

Empiricism ("Just the facts, Ma'am")

Theory may be fallible, but the facts are not, so stick with the facts (observations). Lakatos (1970) defers his original example to kill naive falsification, but it effectively kills empiricism as well, so I introduce it here and use it again later. He points out that Galileo used a telescope to observe mountains on the moon and spots on the sun. These "observations," however, were not made with the unaided eye and depended critically on the verity of optical theory, so there is no natural demarcation between theory and observation.

Probabilism (Neoclassical Justificationism)

Probabilists asserted that since nothing could be proved, one should work with probabilities and weigh quantities of evidence. Doing science this way is like trying to swim by kicking one's feet in a vat of molasses (oscillatory motion at low Reynolds number). You do a lot of flailing, expending a lot of energy, but you don't get anywhere; you just go backward and forward. This approach is eminently sensitive to bias on the part of the investigator. Many popular "pseudoscience" books succumb to such biases. The author adopts an idea attractive to him or her, and then publishes all the evidence that supports it, without mentioning evidence to the contrary.

Dogmatic Falsification

Dogmatic falsification certainly is attractive. It admits the impossibility of proof. It highlights scientific honesty, defined as stating in advance what observations or experimental results would contradict the theory and cause one to abandon it. Roles become clear. In Lakatos (1970) words, "The theoretician proposes, the experimenter -- in the name of Nature -- disposes." The history of science as repeated in textbooks has been re-interpreted in this way. That there is no natural line between theory and observation (above under "empiricism"), however, is fatal to this mode of operation, as are insurmountable logical problems (the inability of factual propositions to be proved from experiments) detailed by Lakatos (1970). Insistence on "disprovable hypotheses," still quite prevalent among reviewers of proposals, dates back to this now untenable approach to science.

Beginning logic students have perhaps the most difficulty giving up this brand of falsification. Take the proposition that, "All swans are white." You could not possibly prove it unless you collected all the swans, and even then you should worry about whether past swans all were white and whether future swans all will be. Conversely, apparently all it takes is one black swan to disprove this proposition. The trouble is that this "disproof" requires a rock-hard definition of "swan" and unquestioning belief in observations. Was the animal in the distance and in a shadow? Did it fall in soot?

(Naive) Methodological Falsification

The lead (parenthetical) word in this label was chosen by Lakatos (1970) for its pejorative ring, but has perhaps had an effect opposite to the one intended. It has been largely ignored, perhaps out of embarrassment, and so a useful distinction has been lost. Methodological falsification as espoused by Popper (1965) is undeniably the most popularized view of scientific practice. The disproof of dogmatic falsification is relaxed to falsification. The impossibility of disproof is admitted, but the asymmetry of hypothesis testing is recognized and used to stack the deck intentionally to make false hypotheses easier to discredit than are true hypotheses. Standard hypothesis testing in statistics makes the same recognition in controlling type 1 or a error, and it is partly for this reason that it is easy to confuse good statistical null hypotheses with good scientific hypotheses. The former do not need to (and indeed often do not) make interesting predictions. This brand of falsification is undeniably an improvement on the dogmatic kind, and Popper (1965 and elsewhere) has made major improvements to the general practice of science. Among the many is the eloquent argument that observations without predictions are worthless: "The belief that science proceeds from observation to theory is so widely and so firmly held that my denial of it is often met with incredulity.. .. But in fact the belief that we can start with pure observation alone, without anything in the nature of a theory, is absurd; as may be illustrated by the story of the man who dedicated his life to natural science, wrote down everything he could observe, and bequeathed his priceless observations to the Royal Society to be used as inductive evidence. This story should show us that though beetles may profitably be collected, observations may not."

(Sophisticated) Methodological Falsification

A bone that Lakatos (1970) picks with Popper's exposition of methodological falsification is that it lacks mention of the creative side of science. Popper treats thoroughly both deduction from theory and testing of individual and alternative hypotheses, but is silent on where predictions arise. One is left with the impression that science is still a two-cornered fight between theory and observation (hypothesis and test). Lakatos' (1970) major contribution is a thorough re-evaluation of the history of science which demonstrates to my satisfaction that major theories of science have not been falsified in the absence of a superior, competing theory. He also convinces me that crucial experiments are myths that often are created by revision of history. Notably, the Michelson-Morley experiment is widely regarded as the crucial experiment that overthrew ether theories in favor of relativity, whereas it predates relativity considerably, was designed to test competing ether theories (which it did poorly) and was considered a minor experiment by Michelson himself (Lakatos 1970). Lakatos (1970) thus puts a very different slant on Kuhn's (1970) much more popularized but arguably mythical view of scientific revolution: The king is not overthrown until a new one is available. Science is a three-cornered fight among rival theories and observations.

Unlike Kuhn's analysis, Lakatos' cascades admirably down to more minor hypotheses and gives effective guidance regarding what an ordinary scientist can plan to accomplish from day to day. If one has but a single hypothesis, what is left if one succeeds in its falsification? Scientific honesty is much easier to define and achieve when competing hypotheses are made explicit; only then can one judge how and whether evidence can be collected to reject one in favor of its competitor. It is no accident that multiple, alternative working hypotheses have served scientists so well for so long (Chamberlin 1897 -- yes that is eighteen hundred ninety-seven). "What really count are dramatic, unexpected, stunning predictions" (Lakatos 1970). Lakatos applies a balance to prediction and testing, but in my opinion rightly emphasizes the overarching value of interesting, new predictions. His exposition explains why, "Important criticism is always constructive." I find this conclusion compelling and the clearest guide in reviewing others' work.