Šešelja and Straßer’s critique fails to hit its target for two main reasons. First, the argument is not that Kuhn is a rationalist because he is a coherentist. Although Kuhn can be taken as a rationalist because of his commitment to epistemic values, coherence analysis provides a more comprehensive characterisation of cognitive process in scientific change than any of these values alone can offer. Further, we should understand Kuhn as characterising science as the best form of rationality we have outside (...) logic, which rules out algorithmic rationality and allows non-cognitive factors to play a role in theory change. Second, Šešelja and Straßer overemphasise the importance of a priori reasoning in Kuhn, which was only an alternative to his earlier historical-empirical approach. My suggestion is that Kuhn’s neo-Kantian historical cognitivism integrates the earlier empirical and the later a-prioristic orientations. According to it, that any understanding of the world is preconditioned by some kind of mental module that is liable to change, detected as a discontinuity in the historical record of science.Keywords: Thomas S. Kuhn; Rationality; Scientific change; Coherentism; Kantianism; Correspondence theory of truth. (shrink)

The incommensurability of two theories seems to problematize theory comparisons, which allow for the selection of the better of the two theories. If so, it becomes puzzling how the quality of theories can improve with time, i.e. how science can progress across changes in incommensurable theories. I argue that in papers published in the 1990s, Kuhn provided a novel way to resolve this apparent tension between incommensurability and scientific progress. He put forward an account of their compatibility which worked not (...) by downplaying the negative consequences of incommensurability but instead by allowing them to reach their natural end: a process of specialisation. This development in Kuhn’s thought has yet to be properly recorded but it is also interesting in its own right. It shows how a robust version of incommensurability—one which really does have severe negative consequences for scientists’ capacity to perform comparative evaluations of incommensurable theories—need make no puzzle of the progress of science. (shrink)