Description versus prescription
Kuhn’s own account of his disagreement with Popper seems a local affair. Popper’s episodes of falsification / refutation correspond – approximately – to revolutionary overthrow of paradigms and these are rare compared to mundane normal science and puzzle solving. But their disagreement is deeper. Popper hankers after a philosophical prescription (of how science ought, correctly, to go). Kuhn offers a description of a typical scientific process (the custom and habit of science). (Lakatos seems to wobble between the two. There are rules for theory change but only guidelines for choosing between research programmes.)
Despite his basic discipline, Kuhn does sometimes say what it is that scientists ought to do (a normative or prescriptive notion). He sheds some light on this on p207: these 'oughts' are merely internal to the theory. If science is proceeding as Kuhn claims that it does then scientists ought to do such and such to count as scientists.
Kuhn as historian
Kuhn’s account is ‘disciplined’ by history. If it is correct then it is correct as a historical account: a kind of historical science of the history of science. So Kuhn’s claims need not be justified philosophically. Some, however, do hang together philosophically. The role of tacit knowledge and the claims about incommensurability can be given some – at least – a priori justification.
Kuhn is the last methodologist of science. After him have come historians and sociologists who do not believe that there are such general patterns as philosophers like Popper and Lakatos and the historian Kuhn all believe.
Incommensurability
Kuhn claims that paradigms are incommensurable. There are a number of forms of incommensurability. Incommensurability of standards / methodological incommensurability is one:
“The incommensurability illustrated above whereby puzzle-solutions from different eras of normal science are evaluated by reference to different paradigms, is methodological incommensurability. Another source of methodological incommensurability is the fact that proponents of competing paradigms may not agree on which problems a candidate paradigm should solve (1962/1970a, 148). In general the factors that determine our choices of theory (whether puzzle-solutions or potential paradigm theories) are not fixed and neutral but vary and are dependent in particular on the disciplinary matrix within which the scientist is working. Indeed, since decision making is not rule-governed or algorithmic, there is no guarantee that those working within the same disciplinary matrix must agree on their evaluation of theory (1962/1970a, 200), although in such cases the room for divergence will be less than when the disputants operate within different disciplinary matrices.” [http://plato.stanford.edu/entries/thomas-kuhn/#4.1]
This looks to be significant and interesting even though it should be obvious to an unprejudiced eye. In effect Kuhn is pointing out that the standards of scientific rationality are themselves subject to change with changing scientific views. And thus there is no neutral standard to test theories. This means that scientific method cannot be simply algorithmic or rule-governed. So if scientific rationality is our best example of rationality in practice, then the requirements of rationality are not algorithmic or rule-governed but rather judgements about what to believe in particular contexts. That can seem quite surprising because we tend to think that the requirements of rationality are quite general / context-free.
Incommensurability of meaning is more radical. It follows from two claims:
1) Holism about meaning
2) The theory dependence of observation
If there is no neutral observation language then theoretical terms cannot be defined in terms of neutral (across theories) observational terms. Instead, there is merely a set of theory-observation terms.
Given holism, the meaning of such terms is given by theoretical context.
So changing the theory changes the meaning of all the terms.
So different theories are incommensurable.
A more radical view
In addition to that, Kuhn also uses that argument to justify – in part – a more radical view. Given that we have no access to a neutral description of the world (because of the theory dependence of observation: there is no neutral observational language and that is the only plausible neutral access to the world), we have no reason not to say that after a scientific revolution, scientists work in different worlds.
‘At the very least, as a result of discovering oxygen, Lavoisier saw nature differently. And in the absence of some recourse to that hypothetical fixed nature that he "saw differently," the principle of economy will urge us to say that after discovering oxygen Lavoisier worked in a different world.’ [Kuhn 1962: 118]
It seems that Kuhn is here asserting a kind of idealism: the world itself is changed by a change of the paradigm that we / our scientists believe or hold. So one question is can we believe in the lesser claim - the incommensurability of meaning - without this radical claim. One thing to press is that seeing and seeing-as or seeing-that are different. Sootica and I can both see my TV but only I can see that it is a TV or, perhaps, see it as a TV. So isn't Kuhn simply failing to spot the difference between seeing and seeing-that / seeing-as? Priestley and Lavoisier both see the same world; but they see it as different things. That is the bland way to capture the claim that Lavoisier 'saw nature differently'.
(It seems that Kuhn assumes that unless one can neutrally specify or describe the world - which is 'seen-as' in different ways - then we are forced to say that Priestley and Lavoisier live in different worlds. And we cannot neutrally specify the nature of the shared world, the world that is seen by both of them even if they see it as different things, in neutral terms if there is no distinction between theory and observation. But why should such a world have to be described in neutral terms? From the scientific perspective that Lavoisier has left us, we can describe the world in his way and explain how Priestley saw that world partly wrongly.)
So we can drive a wedge between the 'different worlds' claim and the more modest meaning-incommensurability claim. But hang on, do we even want to say that paradigms are meaning-incommensurable?
The obvious thought
“Surely we can compare theories. We know, at the very least, that Newtonian and Einsteinian physics are competitors. They are rival theories about the same sort of aspects of the world. If that is ruled out by the argument for the incommensurability of meaning then that shows that that argument must be wrong.”
The problem is earning the right to that obvious thought. What is wrong with Kuhn’s argument for meaning-incommensurability? Here is a thought.