by Alan Sokal
The main problem in reading Feyerabend is to know when to take him seriously. On the one hand, he is often considered as a sort of court jester in the philosophy of science, and he seems to have taken some pleasure in playing this role.98 At times he himself emphasized that his words ought not be taken literally.99 On the other hand, his writings are full of references to specialized works in the history and philosophy of science, as well as in physics; and this aspect of his work has greatly contributed to his reputation as a major philosopher of science. Bearing all this in mind, we shall discuss what seem to us to be his fundamental errors, and illustrate the excesses to which they can lead.
We fundamentally agree with what Feyerabend says about the scientific method, considered in the abstract:
The idea that science can, and should, be run according to fixed and universal rules, is both unrealistic and pernicious.
(Feyerabend 1975, p. 295)
He criticizes at length the ‘fixed and universal rules’ through which earlier philosophers thought that they could express the essence of the scientific method. As we have said, it is extremely difficult, if not impossible, to codify the scientific method, though this does not prevent the development of certain rules, with a more-or-less general degree of validity, on the basis of previous experience. If Feyerabend had limited himself to showing, through historical examples, the limitations of any general and universal codification of the scientific method, we could only agree with him.100 Unfortunately, he goes much further:
All methodologies have their limitations and the only ‘rule’ that survives is ‘anything goes’.
(Feyerabend 1975, p. 296)
This is an erroneous inference that is typical of relativist reasoning. Starting from a correct observation – ‘all methodologies have their limitations’ – Feyerabend jumps to a totally false conclusion: ‘anything goes’. There are several ways to swim, and all of them have their limitations, but it is not true that all bodily movements are equally good (if one prefers not to sink). There is no unique method of criminal investigation, but this does not mean that all methods are equally reliable (think about trial by fire). The same is true of scientific methods.
In the second edition of his book, Feyerabend tries to defend himself against a literal reading of ‘anything goes’. He writes:
A naive anarchist says (a) that both absolute rules and context-dependent rules have their limits and infers (b) that all rules and standards are worthless and should be given up. Most reviewers regard me as a naive anarchist in this sense ... [But] while I agree with (a) I do not agree with (b). I argue that all rules have their limits and that there is no comprehensive ‘rationality’, I do not argue that we should proceed without rules and standards.
(Feyerabend 1993, p. 231)
The problem is that Feyerabend gives little indication of the content of these ‘rules and standards’; and unless they are constrained by some notion of rationality, one arrives easily at the most extreme form of relativism.
When Feyerabend addresses concrete issues, he frequently mixes reasonable observations with rather bizarre suggestions:
[T]he first step in our criticism of customary concepts and customary reactions is to step outside the circle and either to invent a new conceptual system, for example a new theory, that clashes with the most carefully established observational results and confounds the most plausible theoretical principles, or to import such a system from outside science, from religion, from mythology, from the ideas of incompetents, or the ramblings of madmen.
(Feyerabend 1993, pp. 52–3)101
One could defend these assertions by invoking the classical distinction between the context of discovery and the context of justification. Indeed, in the idiosyncratic process of inventing scientific theories, all methods are in principle admissible – deduction, induction, analogy, intuition and even hallucination102 – and the only real criterion is pragmatic. On the other hand, the justification of theories must be rational, even if this rationality cannot be definitively codified. One might be tempted to think that Feyerabend’s admittedly extreme examples concern solely the context of discovery, and that there is thus no real contradiction between his viewpoint and ours.
But the problem is that Feyerabend explicitly denies the validity of the distinction between discovery and justification.103 Of course, the sharpness of this distinction was greatly exaggerated in traditional epistemology. We always come back to the same problem: it is naive to believe that there exist general, context-independent rules that allow us to verify or falsify a theory; otherwise put, the context of justification and the context of discovery evolve historically in parallel.104 Nevertheless, at each moment of history, such a distinction exists. If it didn’t, the justification of theories would be unconstrained by any considerations of rationality. Let us think again about criminal investigations: the culprit can be discovered thanks to all sorts of fortuitous events, but the evidence put forward to prove his guilt does not enjoy such a freedom (even if the standards of evidence also evolve historically).105
Once Feyerabend has made the leap to ‘anything goes’, it is not surprising that he constantly compares science with mythology or religion, as, for example, in the following passage:
Newton reigned for more than 150 years, Einstein briefly introduced a more liberal point of view only to be succeeded by the Copenhagen Interpretation. The similarities between science and myth are indeed astonishing.
(Feyerabend 1975, p. 298)
Here Feyerabend is suggesting that the so-called Copenhagen interpretation of quantum mechanics, due principally to Niels Bohr and Werner Heisenberg, was accepted by physicists in a rather dogmatic way, which is not entirely false. (It is less clear which point of view of Einstein he is alluding to.) But what Feyerabend does not give are examples of myths that change because experiments contradict them, or that suggest experiments aimed at discriminating between earlier and later versions of the myth. It is only for this reason – which is crucial – that the ‘similarities between science and myth’ are superficial.
This analogy occurs again when Feyerabend suggests separating Science and the State:
While the parents of a six-year-old child can decide to have him instructed in the rudiments of Protestantism, or in the rudiments of the Jewish faith, or to omit religious instruction altogether, they do not have a similar freedom in the case of the sciences. Physics, astronomy, history must be learned. They cannot be replaced by magic, astrology, or by a study of legends.
Nor is one content with a merely historical presentation of physical (astronomical, historical, etc.) facts and principles. One does not say: some people believe that the earth moves round the sun while others regard the earth as a hollow sphere that contains the sun, the planets, the fixed stars. One says: the earth moves round the sun – everything else is sheer idiocy.
(Feyerabend 1975, p. 301)
In this passage Feyerabend reintroduces, in a particularly brutal form, the classical distinction between ‘facts’ and ‘theories’ – a basic tenet of the Vienna Circle epistemology he rejects. At the same time he appears to use implicitly in the social sciences a naively realist epistemology that he rejects for the natural sciences. How, after all, does one find out exactly what ‘some people believe’, if not by using methods analogous to those of the sciences (observations, polls, etc.)? If, in a survey of Americans’ astronomical beliefs, the sample were limited to physics professors, there would probably be no one who ‘regards the earth as a hollow sphere’; but Feyerabend could respond, quite rightly, that the poll was poorly designed and the sampling biased (would he dare say that it is unscientific?). The same goes for an anthropologist who stays in New York and invents in his office the myths of other peoples. But which criteria acceptable to Feyerabend would be violated? Doesn’t anything go? Feyerabend’s methodological relativism, if taken literally, is so radical that it becomes self-refuting. Without a minimum of (rational) method, even a ‘merely historical pr
esentation of facts’ becomes impossible.
What is striking in Feyerabend’s writings is, paradoxically, their abstractness and generality. His arguments show, at best, that science does not progress by following a well-defined method, and with that we basically agree. But Feyerabend never explains in what sense atomic theory or evolution theory might be false, despite all that we know today. And if he does not say that, it is probably because he does not believe it, and shares (at least in part) with most of his colleagues the scientific view of the world, namely that species evolved, that matter is made of atoms, etc. And if he shares those ideas, it is probably because he has good reasons to do so. Why not think about those reasons and try to make them explicit, rather than just repeating over and over again that they are not justifiable by some universal rules of method? Working case by case, he could show that there are indeed solid empirical arguments supporting those theories.
Of course, this may or may not be the kind of question that interests Feyerabend. He often gives the impression that his opposition to science is not of a cognitive nature but follows rather from a choice of lifestyle, as when he says: ‘love becomes impossible for people who insist on “objectivity”, i.e. who live entirely in accordance with the spirit of science.’106 The trouble is that he fails to make a clear distinction between factual judgments and value judgments. He could, for example, maintain that evolution theory is infinitely more plausible than any creationist myth, but that parents nevertheless have a right to demand that schools teach false theories to their children. We would disagree, but the debate would no longer be purely on the cognitive level, and would involve political and ethical considerations.
In the same vein, Feyerabend writes in the introduction to the Chinese edition of Against Method:107
First-world science is one science among many ... My main motive in writing the book was humanitarian, not intellectual. I wanted to support people, not to ‘advance knowledge’.
(Feyerabend 1988, p. 3 and 1993, p. 3, italics in original)
The problem is that the first thesis is of a purely cognitive nature (at least if he is speaking of science and not of technology), while the second is linked to practical goals. But if, in reality, there are no ‘other sciences’ really distinct from those of the ‘first world’ that are nevertheless equally powerful at the cognitive level, in what way would asserting the first thesis (which would be false) allow him to ‘support people’? The problems of truth and objectivity cannot be evaded so easily.
The ‘strong programme’ in the sociology of science
During the 1970s, a new school in the sociology of science arose. While previous sociologists of science were, in general, content to analyse the social context in which scientific activity takes place, the researchers gathered under the banner of the ‘strong programme’ were, as the name indicates, considerably more ambitious. Their aim was to explain in sociological terms the content of scientific theories.
Of course, most scientists, when they hear about these ideas, protest and point out the substantial missing piece in this kind of explanation: Nature itself.108 In this section we shall explain the fundamental conceptual problems faced by the strong programme. While some of its supporters have recently made corrections to their initial claims, they do not seem to realize the extent to which their starting point was erroneous.
Let us start by quoting the principles set forth for the sociology of knowledge by one of the founders of the strong programme, David Bloor:
1. It would be causal, that is, concerned with the conditions which bring about belief or states of knowledge. Naturally there will be other types of causes apart from social ones which will cooperate in bringing about belief.
2. It would be impartial with respect to truth and falsity, rationality or irrationality, success or failure. Both sides of these dichotomies will require explanation.
3. It would be symmetrical in its style of explanation. The same types of cause would explain, say, true and false beliefs.
4. It would be reflexive. In principle its patterns of explanation would have to be applicable to sociology itself.
(Bloor 1991, p. 7)
To grasp what is meant by ‘causal’, ‘impartial’ and ‘symmetrical’, we shall analyse an article of Bloor and his colleague Barry Barnes in which they explain and defend their programme.109 The article begins with an apparent statement of good will:
Far from being a threat to the scientific understanding of forms of knowledge, relativism is required by it. ... It is those who oppose relativism, and who grant certain forms of knowledge a privileged status, who pose the real threat to a scientific understanding of knowledge and cognition.
(Barnes and Bloor 1981, pp. 21–2)
However, this already raises the issue of self-refutation: Doesn’t the discourse of the sociologist, who wants to provide ‘a scientific understanding of knowledge and cognition’, claim ‘a privileged status’ with respect to other discourses, for example those of the ‘rationalists’ that Barnes and Bloor criticize in the rest of their article? It seems to us that, if one seeks to have a ‘scientific’ understanding of anything, one is forced to make a distinction between a good and a bad understanding. Barnes and Bloor seem to be aware of this, since they write:
The relativist, like everyone else, is under the necessity to sort out beliefs, accepting some and rejecting others. He will naturally have preferences and these will typically coincide with those of others in his locality. The words ‘true’ and ‘false’ provide the idiom in which those evaluations are expressed, and the words ‘rational’ and ‘irrational’ will have a similar function.
(Barnes and Bloor 1981, p. 27)
But this is a strange notion of ‘truth’, which manifestly contradicts the notion used in everyday life.110 If I regard the statement ‘I drank coffee this morning’ as true, I do not mean simply that I prefer to believe that I drank coffee this morning, much less that ‘others in my locality’ think that I drank coffee this morning!111 What we have here is a radical redefinition of the concept of truth, which nobody (starting with Barnes and Bloor themselves) would accept in practice for ordinary knowledge. Why, then, should it be accepted for scientific knowledge? Note also that, even in the latter context, this definition doesn’t hold water: Galileo, Darwin and Einstein did not sort out their beliefs by following those of others in their locality.
Moreover, Barnes and Bloor fail to use systematically their new notion of ‘truth’; from time to time they fall back, without comment, on the traditional sense of the word. For example, at the beginning of their article, they admit that ‘to say that all beliefs are equally true encounters the problem of how to handle beliefs which contradict one another’, and that ‘to say that all beliefs are equally false poses the problem of the status of the relativist’s own claims’.112 But if ‘a true belief’ meant only ‘a belief that one shares with other people in one’s locality’, the problem of the contradiction between beliefs held in different places would no longer pose any problem.113
A similar ambiguity plagues their discussion of rationality:
For the relativist there is no sense attached to the idea that some standards or beliefs are really rational as distinct from merely locally accepted as such.
(Barnes and Bloor 1981, p. 27)
Again, what exactly does this mean? Isn’t it ‘really rational’ to believe that the Earth is (approximately) round, at least for those of us who have access to aeroplanes and satellite photos? Is this merely a ‘locally accepted’ belief?
Barnes and Bloor seem here to be playing on two levels: a general scepticism, which of course cannot be refuted; and a concrete program aiming at a ‘scientific’ sociology of knowledge. But the latter presupposes that one has given up radical scepticism and that one is trying, as best one can, to understand some part of reality.
Let us therefore temporarily put aside the arguments in favour of radical scepticism, and ask whether the ‘strong programme’, considered a
s a scientific project, is plausible. Here is how Barnes and Bloor explain the symmetry principle on which the strong programme is based:
Our equivalence postulate is that all beliefs are on a par with one another with respect to the causes of their credibility. It is not that all beliefs are equally true or equally false, but that regardless of truth and falsity the fact of their credibility is to be seen as equally problematic. The position we shall defend is that the incidence of all beliefs without exception calls for empirical investigation and must be accounted for by finding the specific, local causes of this credibility. This means that regardless of whether the sociologist evaluates a belief as true or rational, or as false and irrational, he must search for the causes of its credibility. ... All these questions can, and should, be answered without regard to the status of the belief as it is judged and evaluated by the sociologist’s own standards.
(Barnes and Bloor 1981, p. 23)
Here, instead of a general scepticism or philosophical relativism, Barnes and Bloor are clearly proposing a methodological relativism for sociologists of knowledge. But the ambiguity remains: what exactly do they mean by ‘without regard to the status of the belief as it is judged and evaluated by the sociologist’s own standards’?