Comments and Articles
> Faith, Fact and Fiction: The demarcation of science and non-science
|I wrote this paper in 1988 for a university seminar on the philosophy of science. It turned up during 2008 when I was clearing the clutter before moving house. As the passage of twenty years didn't seem to have made it too out-of-date I decided that it could do with another airing.|
Reading Martin Gardner's book Fads and Fallacies in the Name of Science (Gardner,1957) in 1988 is a depressing experience. It is not depressing because of the contents, but because the contents are so topical and applicable more than thirty years after the book's publication. Except for a prediction that dianetics and its associated "religious" fraud, Scientology, would soon disappear, the book reads as if it had been written last week. If one were asked to state a single word which best represented the Zeitgeist of the late 1980s, a strong contender would have to be "superstition".
Shirley MacLaine makes far more money peddling her nonsense about reincarnation and astral travel then she ever did out of films. Peter Brock destroyed a career, a business, and long-standing friendships because he believed that orgone energy could make a racing car faster (see Tuckey (1987) for this sad story). Nancy Reagan consults an astrologer, and even if she doesn't use the predictions to sway her husband, suspicions are raised and his influence eroded. Steve MacQueen spent the money he had earned as an actor on useless laetrile treatments in a vain attempt to beat cancer. The Co-op Bookshop at Macquarie University carries books by Uri Geller on a shelf labelled "Popular Science". A self-proclaimed elite group of Sydney's investigative journalists accepted a made-for-the-occasion Californian channeller, virtually without question, gave him thousands of dollars worth of free advertising, and then denied responsibility when the fraud was exposed, instead criticising the ethics of the journalist who had set it up to test (and, as it turned out, confirm) their gullibility. These are just the well known people, and they can look after themselves, but what of the less known, "ordinary" people who may follow their examples. Someone must be buying all those books and crystals put out by MacLaine and the other New Agers. Five hundred people at least were attracted to a performance by the fake channeller, Carlos, mentioned above. Table 1 shows counts of the entries under various categorisations in the latest Yellow Pages telephone book for Sydney. Someone must be supporting all these businesses.
|alternative health services||85|
|relaxation therapy (including ganzfeld and sensory deprivation tanks)||66|
I feel that at this point I should state my own position and prejudices, in case they are not quite clear. I am an empiricist, specifically a scientific realist, and I am prepared to accept statistically-based induction as a reasonable basis for knowledge. I am a skeptic, but some things are so highly probable that it seems irrational to doubt them. I am a liberal, in the sense understood by Mill and Bentham, in that I would allow anyone to think and do whatever they liked, as long as no harm came to anyone else, although people must be protected to a certain extent from the effects of misfortune and their own folly.
People can have beliefs and knowledge bases for three different reasons. Firstly, they believe or know something for rational reasons based on the reality of the universe. As example, I see my empiricism, inductionism, and skepticism as cases of this kind of belief, as well, of course, as being the bases for all my other beliefs and knowledge in this class. The second class contains religious, moral, aesthetic, or ethical beliefs. These may not (and perhaps should not) be logically justified, but their basis in faith must be recognised. Beliefs in this class may or may not have an observable effect. Liberalism is an example of an ethical belief which affects others, but, as Blaise Pascal pointed out, belief in God may have absolutely no effect in this life but an enormous effect in the next. The third class is nonsense, which has no effect on anything, and cannot be justified in any way. I was predestined to be a methodical, rational empiricist because I was born on the cusp of Virgo and Libra.
All of these belief classes are valid, and most people have all three, although none are mandatory. People lacking beliefs in the third, nonsense, class can be extremely boring and humourless. People lacking religion, morals, and ethics are extremely unpleasant. Anyone without rational beliefs is probably mad. There is only a problem when the boundaries become distorted or overlap. These three classes are discrete and contain qualitatively different types of elements. It was no accident that I used the word "knowledge" only for elements of the first class, and said "beliefs" for the others. All this preamble is a way of saying that there is a qualitative difference between science (which is basically the first class), non-science, and nonsense. If this difference exists, then there must be a way to detect it. The next part of this paper will describe some ways in which science can be distinguished from other forms of belief. It is not sufficient, however, to simply prove that such demarcation is possible. I believe that it is essential to actually make and use the distinction. A properly functioning human requires a sort of metaknowledge, cataloguing the complete set of his beliefs so that their source and justification is always explicit. The final part of the paper will give some reasons for this mandatory categorisation.
It seems appropriate to start by looking for some a priori description of science. The Macquarie Dictionary defines it as "1.a. the systematic study of man and his environment, based on the deductions and inferences which can be made, and the general laws which can formulated, from reproducible observations and measurements of events and parameters within the universe, b. the knowledge so obtained." Other definitions include
I see science basically as defined by the dictionary, with an added aspect of skepticism. What separates science from other beliefs is the fact that its belief structure is always under challenge. This presents an apparent paradox, in that, of the three belief classes, the only one whose contents can be doubted is the one containing "truth" and "facts" about the world, and beliefs go into the "non-factual" classes if they rely on irrefutable absolute truth! This is not a problem, however if the strictly disjoint nature of the classes, as stated above, is observed. It is precisely this skepticism which demarks science (and other areas of knowledge) from pure belief - scientific theories can be tested and rejected if found wanting. It should be noted that I do not separate what is conventionally called "science" from other forms of skeptical enquiry which seek to gather knowledge. Any activity which expands knowledge by a process of theory invention, appraisal, modification, and so on should only be distinguished from other activities of the same sort by the subject matter, and this should make little or no difference to the epistemological basis for enquiry, although the problems, theories, and methods may differ enormously. With this interpretation, many of the conventional barriers between science and other areas of study break down. For example, archaeology, anthropology and history all fit the definition of scientific activity and are complementary. Certain areas of philosophy also can be included. Analysis of the poetic and aesthetic qualities of Shakespeare's work is not scientific, but investigation of historical soundness and the question of authorship are. (If Francis Bacon has a ghost, it must be gratifying for him to watch scholars using his methods to infer authorship from observations of word-length frequency distributions and other arcane data. Of course, if he really wrote the plays, he must be wondering when they are going to stop gathering data and start announcing proof.) What is important is the method of enquiry, and the source and evaluation of theories.
I would like to clarify here what I mean by "method". There has been a tradition of defining science in terms of activity - "science" is that which is practised by "scientists", following the "scientific method". This suggests that there may be one correct sequence of steps which if followed precisely will lead to discovery of the truth. This belief belongs firmly in the nonsense class. This is not to say that I agree with Feyerabend (1975) that there is no method at all, and that progress in science (if such a concept as progress has content within a context of anarchy) is some form of statistical artefact or an accidental byproduct of many people pursuing random interests. There is no algorithmic method which will always work. When I talk of "method", I mean only a system which allows for theories to be invented or proposed, for those theories to be tested for their predictive or retrodictive power or for agreement with observations of the real world, and for the results of such tests to be used in the appraisal of those theories, with this cycle being repeated, each iteration building on the results of the last. The requirements for a valid theory are set out in the next section of the paper.
My own field of endeavour, the development of computer software, can be seen as a model of the method set out above. At the lowest level, we use strictly algorithmic means of problem solution, but this breaks down as the pieces become larger. For more than thirty years, people have been trying to apply the low-level engineering and logical constructs to create some kind of metamethod, but all attempts have failed. There are an enormous number of ways of carrying out the process of production of something useful, simply because the range of problems to be solved is huge and intellectual capacities vary from creative genius to plodding dullard (for most of us positions on this continuum change frequently). This does not mean that there is anarchy, or that no progress is made, or that there is really no method. We delineate problems, suggest theories to solve those problems, build test equipment (programs) to test the theories, find new problems, and so forth. Much like science, in fact.
We have seen from the language used to define "science" that an a priori case can be made for positing a difference between science and non-science. Derived from that analytic difference, an empirical test can be devised. This is simply that the theories of science are open to empirical validation, whereas those of non-science are not. The following set of rules or criteria for evaluating a theory as scientific or not are based on those defined by Root- Bernstein (1984). At the highest level, there are four major categories of criteria - logical, empirical, sociological, and historical.
To be logically valid, a theory must be parsimonious, applying the principle of Occam's Razor, so that nothing unnecessary is postulated. It must be logically consistent within itself, in that it must not rely on contradictory assumptions. It must be logically falsifiable, so that conditions in which it would be proved to be untrue can be imagined, and tautologies can be avoided. (This is an essential part of devising any test procedure.) It must have clearly stated boundary conditions, so that there is no doubt about what data and observations are relevant. Creation science seems to be ruled out by these logical criteria, as it is impossible to imagine how a theory requiring an omnipotent creator can be falsified. Some of feminism looks a bit shaky too, with its inconsistent claims such as that women and men differ only through the social construct of gender, but that women are naturally more caring and peace-loving than men because they bear the children.
For empirical validity, a theory must be empirically testable itself, or must provide predictions or statements about past events which are themselves empirically testable. This appears to be a tautology, but it is important to stress it. There are some theories, such as those of astronomy which are not testable as such, but which make statements about how things should be. This leads to the next condition, which is that theory must have actually made verified predictions or retrodictions. If all of its predictions have been falsified, then it is hardly a valid theory. Astrology seems to be in this position. Other empirical criteria are that the theory must offer reproducible results, otherwise it is merely a statement about the observation of some highly improbable event, and would be of no use to someone designing an experiment. The theory must also define the criteria for deciding whether observations are to be considered as facts, artefacts, anomalies, or as irrelevant to the theory. If such distinctions cannot be made, so that anything is possible under the theory, then no test can be devised. These last two criteria are the ones which seem to provide the most trouble for parapsychology researchers - their results cannot be reproduced, and almost anything seems to be valid data.
Sociological criteria are similar to what Kuhn (1970) was referring to, when he talked about changes in paradigm, although I am applying them here without necessarily requiring a total revolution covering all theory in the particular branch of science. A theory must address the problems of past theories by resolving problems or anomalies which defied previous theories. It should provide a new set of problems to work on in the future, and posit a method for solving these problems. If it does not do these two things, then it offers nothing which cannot be learned by other means. These sociological criteria, together with falsifiability and reproducibility mentioned above, provide the main means for correcting existing theories. It is only by building on past theories and creating new ones that any progress can be made.
This leads to the last major class of theory criteria, historical. To be valid historically, a theory must meet all criteria met by its predecessors, or be able to prove their irrelevance. It must be able to explain all of the data observed within the paradigm of the replaced theory as either fact or artefact, but must not of course claim that something previously explained is now anomalous. Finally, it must be consistent with any ancillary theories which either attached to the replaced theory or have now become attached, if these theories have been independently established as valid.
In summary, then, a valid theory must explain more than its predecessors without contradicting anything which has independent validity; it must be possible to devise a test which could falsify it or it must make predictions which can be independently assessed; testing of the theory should be independent of the experimenter, that is, it must offer reproducible results; it must have a priori validity, in that it must be logically consistent and bounded in possibility. Validity is of course not proof, but if a theory satisfies all, or even most, of the above criteria, then it can be accepted as providing a useful description of reality. In other words, it is a theory which can be called part of science.
When choosing the title for this paper, I had little trouble with "fact" and "faith" as names for the first two categories of belief. For the last category, the term "fiction" was deliberately chosen over such equally alliterative options as "fallacy", "fantasy", "folly", or "foolishness" because it implies a consciousness of the fact that what is in this class is somehow "untrue". In other words, one of the elements in the "fact" category is a metastatement (or set of such statements) which partitions the holder's set of beliefs according to whether he holds them because they have been verified empirically (to some acceptable level of probability), or he holds them because some higher authority requires them or they support his ideological framework, or he holds them for no reason other than choice. That is, he holds them because they are facts, through faith, or knowingly as fiction. When I described myself as a liberal before, it may have seemed that this had little relevance to the subject of the demarcation between science and non-science. I would like, however, to repeat that this means "that I would allow anyone to think and do whatever they liked, as long as no harm came to anyone else, although people must be protected to a certain extent from the effects of misfortune and their own folly". The relevance is that, providing people have a correct categorisation of their beliefs as set out above, then almost any set of beliefs is acceptable. If people want to believe that UFOs contain little green men from outer space, if they believe that little pyramids will sharpen their razor blades, if they talk to plants, if they believe that their dogs are telepathic, or if they think that the inhabitants of Atlantis watched television, then there is no problem. The world needs eccentrics and people with differing views to make it an interesting place. The problems arise when people make categorisation errors - at this point "fiction" becomes "fallacy", "fantasy", "folly", or "foolishness", and their beliefs become pathological, and intervention to protect them and others is justified, even for the most liberal of observers.
There are still some who would argue that pathological beliefs are acceptable, and in some cases this may be so, but there are areas such as biology, medicine and health, anthropology, and sociology where confusion of either faith or fiction with fact can have literally lethal effects. More than six million people died because lunatics among the Nazis allowed racist ideology to dictate anthropology, although some people may have lived longer because the same (or similar) lunatics used astrology to select battle times and also impeded the work of real scientists trying to make atom bombs. Goodness knows how many died in the famines caused through the USSR's rejection of Mendelian genetics because it conflicted with a materialist ideology. In 1978, more than 900 people committed suicide in Guyana because a quack cancer-curer destroyed their boundaries between religion and reality. Less shocking, but closer to home and more immediate in time, doctors in Sydney hospitals are currently treating a spate of children who have been kept on virtually fat-free vegetarian diets, resulting in damage to their metabolic and digestive systems, and many doubtlessly well-meaning people are accepting a crackpot anthropology which is based on poor statistics, distorted history, ignorance of linguistics and archaeology, poorly done real anthropology, the "myth of the noble savage", and a good dose of fiction, and which could lead to the establishment of a system of apartheid dividing Australians who also happen to be aborigines from Australians who are not.
The issue of the demarcation between science and non-science is more than just a philosophical mind exercise. As shown above, when people lose the ability to distinguish between fact and fiction they may also lose the power of rational thought, and the consequent delusions can have tragic results. There will always be charlatans who know the difference but pretend there is none, just as there will always be cranks and crackpots who deny a difference because they cannot see one. It is the task of rational people to defend against both groups, and the only defence is a clear understanding that there is a real, qualitative difference between science and non-science. They are different, and the difference is absolute. Some beliefs can be proved, some can not, and nothing can be both.
Against Method. London:NLB, 1975.
Frazier, K. (Ed.)
Science Confronts the Paranormal. New York:Prometheus, 1986.
Fads and Fallacies in the Name of Science. New York:Dover, 1957.
The Structure of Scientific Revolutions. Chicago;UC Press, 1970.
Langley, P., Simon, H.A., Bradshaw, G.L. & Zytkow, J.M.
Scientific Discovery: Computational Explorations of the Creative Process. Cambridge, MA:M.I.T. Press, 1987.
Scientific Creationism Versus Evolution: The Mislabeiled Debate. In Montagu (1984).
Montagu, A (Ed.)
Science and Creationism. New York:0xford University Press, 1984.
On Defining a Scientific Theory: Creationism Considered. In Montagu (1984).
Creationist Pseudoscience. In Frazier (1986).
The Rise and Fall of Peter Brock. Richmond Vic:Greenhouse, 1987.