Antoine Brandelet

The epistemology of models has to face a conundrum: models are often described as highly idealised, and yet they are considered to be vehicles for scientific explanations. Truth-oriented—veritist—conceptions of explanation seem thereby undermined by this contradiction. In this article, I will show how this apparent paradox can be avoided by appealing to the notion of fiction. If fictionalism is often thought to lead to various flavours of instrumentalism, thereby weakening the veritist hopes, the fiction view of models offers a framework much richer than it seems at first sight. To do so, I will call upon the concepts of modality, counterfactual structure and credible worlds. In the end, veritism of explanation and fiction can indeed go hand in hand, but the scope of explanations we can hope to draw from models must be more precisely delineated.

John Stuart Mill’s System of Logic (1843) is often considered to be a work that defends an inductivist epistemology. In this article, I propose to question this status by examining the definition of induction set out in Mill’s book, and the consequences that can be deduced from it. I show that Mill’s inductivism implies a methodology of scientific research in which deductive reasoning is just as important, if not more important, than inductive methods, so that the classical opposition between induction and deduction is transformed into a complementarity. I then show how the concepts of natural law and causality emerge from these discussions, and examine how well the methods proposed by Mill are able to account for subsequent scientific discoveries.