Feyerabend came to his infamous defense of "epistemological anarchy" via a Popperian youth, and the idea that theories could, should, and are falsified remained a deep strain in his philosophy. Although he took up a stand which was called "irrationalist" with respect to science, Feyerabend did have cleverly defended philosophical arguments which influenced many, in spite of the commonly widespread disagreement with his overall position. He wrote simultaneously with Kuhn, and the two were colleagues for a while at Berkeley, but he also had sharp disagreements with Kuhn's approving attitude towards what he had called "normal science."
As a vestigial Popperian, Feyerabend strenuously objected to Kuhn's defense of the monopolistic status of a reigning paradigm during "normal science." The absence of rival theories -or the refusal to give them any serious consideration as long as the accepted paradigm remains free from anomalies- presents what Feyerabend argued was a restricting barrier to the necessary and the actual historical growth of knowledge. However Feyerabend,. like Kuhn, accepted -to the fullest- both the theory ladeness of all "facts" (Hanson) and the holistic view of language meaning (Quine-Duhem). Thus Feyerabend reasoned from the assumption of the "theory-ladeness" of facts to the conclusion that the "facts" most likely to refute any given theory will not be facts observed by that theory, but rather by its rivals. Therefore, if the Popperian view that knowledge grows by refuting the old beliefs, the most rational way to make knowledge grow is to maximize the number of competing theories. Methodological rules, however, have the effect of restricting the permissible contending rivals (as well as to determine such matters as funding, institutional support, etc.) to those rivals deemed sufficiently "orthodox," as well as cutting off, before they have a chance (or funding or institutional support) to grow, those theories which are regarded as "unorthodox." From this analysis of methodological rules, Feyerabend was thus led to conclude that "liberation" entails rejecting all constraining rules of acceptable methodology. Ever the clever promoter for his views, he siezed on a motto which reflected the spirit of the sixties: "anything goes."
In particular Feyerabend attacked two constraints which he regarded as empiricistic dogma, which he called the "consistency condition' and the "thesis of meaning invariance".
The "consistency condition":
According to traditional empiricist analysis of change of scientific belief over time, earlier theories are "subsumed" as "limiting cases" of later, more general theories. The standard example was the case of the replacement of Newtonian mechanics by relativistic mechanics. The basic argument holds that Newtonian mechanics is not, strictly speaking, "refuted" by relativity theory, because in the range of ordinary human experience Newton's theory does give correct predictions. However, its range of application must be "restricted" or "limited" to just those velocities, v, which are very small in relation to c, the speed of light, i.e. cases where the ratio v/c is very small. Thus Newtonain mechanics is (allegedly) "reduced" to a "special case" of relativity theoy where we are dealing with small relative velocities. But in order to be a special case of relativistic theory, Newtonain theory must be logically consistent with relativity theory in this limited range of low velocities. No theory can be inconsistent with one of its special cases. Generalizing from this example, we can say that the "consistency condition" holds that later (more general) theories must be consistent with earlier theories.
If empiricists do indeed impose this consistency condition as a methodological rule on future research, then no new theory will be permitted to come forward for consideration (much less funding, institutional support, etc.) if it is held to be "inconsistent" with currently accepted theories. Feyerabend scours the historical record to try to show that in virtually all major revolutions this consistency condition is actually violated, and if it were to be imposed as a methodological rule (and to the extent that it is often so imposed), it would have a constraining effect retarding scientific progress scientific progress, which, according to Feyerabend, demands that we proliferate as many rival theories to "received" theory as possible. Thus the consistency condition should be abandoned. (Note that this is a normative claim about what scientists should do.)
The thesis of meaning invariance:
The question of how science progresses is in effect the question of when is it rational for a new theory to replace an old theory. Empiricists argued that in order for such a move to be rational, the successor theory must be about the same sort of things as its predecessor. In order for successive theories to be "rivals" they must offer competing accounts of the same phenomena. For this reason the meanings of those terms referring to such phenomena in both theories must be the same, or in other words, meaning must be invariant in the change from an old theory to a new one.Although many philosophers found such "liberation" to be potentially very frightening, Feyerabend celebrated this anarchistic attack on the constraints of any alleged scientific method.
Obviously if this condition is imposed as a methodological rule on the development of science over time, it will have a severely restricting effect on those new theories which which will be considered as potential replacements for an existing theory. Thus for the same reasons that he gives for rejecting the "consistency condition" Feyerabend argues we should reject the constraining influence of the empiricists' condition of meaning invariance. Furthermore, he again claims that the historical record displays that in fact scientists actually do reject such a constraint. This implies that terms change meaning in the shift from old theories to new theories, which, taken to an extreme, will imply radical meaning (i.e., semantic) incommensurability.