Kuhn’s account of scientific change is characterized by an internal tension between a naturalist vein, which is compatible with the revolutionary perspective on the historical development of science, and an aprioristic or Kantian vein which wants to secure that science is not an irrational enterprise. Kuhn himself never achieved to resolve the tension or even to deal with the terms of the problem. Michael Friedman, quite recently, provided an account which aspires to reconcile the revolutionary and the aprioristic elements of (...) the Kuhnian historiography by placing the latter in the neo-Kantian philosophical framework. In this paper, I suggest that Friedman manages to provide a convincing description of the structuration of scientific knowledge and, further, to present science as a rational enterprise. I argue that despite these virtues Friedman’s account faces a major problem with regard to the context of transcendental idealism which he is not willing to abandon. I also argue that discarding transcendental idealism in the way the philosophical perspective of John McDowell does, we can dispose of the flaws of Friedman’s account while retaining its virtues. (shrink)