Commentary on “is Science Cumulative? a Physicist Viewpoint”, by Bernard d’Espagnat

Bernard d’Espagnat’s concern is to meet some of the challenges of Kuhnian philosophy of science, especially as they concern the claim that scientific development is not cumulative. His main line of argument is to use the fact that, in physics, the equations of superseded theories frequently remain useful, for example, for making predictions. How can this be reconciled with the thesis of non-cumulativity? D’Espagnat argues that it cannot and that, therefore, this thesis ought to be rejected. In my commentary, I want to argue that d’Espagnat’s arguments fail to refute Kuhn’s claims with respect to cumulativity. Thus, I will defend Kuhn’s claims only against the arguments provided by d’Espagnat; I do not want to claim that Kuhn’s position is the ultimate truth.

I want to begin by briefly summarizing Kuhn’s (1970) own position as regards cumulative scientific change. The first thing to note is that Kuhn’s thesis of noncumulativity pertains only to revolutionary scientific change. Only scientific revolutions bring about the radical changes in the cognitive fabric of scientific theories that give rise to non-cumulativity. By contrast, normal science is fully cumulative. Normal science is likened by Kuhn to the practice of puzzle-solving, where the puzzles to be solved as well as the solutions produced stand in an appropriate similarity relation to some paradigmatic problem solutions. In fact, the very nature of normal science makes normal scientific development cumulative, as each newly solved puzzle is simply added to the stock of puzzles already solved and the latter are in no way affected by the coming into being of new puzzle solutions.