The scientist's decision of accepting a given proposition is assumed to be dependent on two factors: the scientist's 'private' information about the value of that statement and the proportion of colleagues who also accept it. This interdependence is modelled in an economic fashion, and it is shown that it may lead to multiple equilibria. The main conclusions are that the evolution of scientific knowledge can be path-dependent, that scientific revolutions can be due to very small changes in the empirical evidence, and that not all possible equilibria are necessarily efficient, neither in the economic nor in the epistemic sense. These inefficiencies, however, can be eliminated if scientists can form coalitions.