Constructive empiricism, observability and three kinds of ontological commitment

In this paper, I argue that, contrary to the constructive empiricist's position, observability is not an adequate criterion as a guide to ontological commitment in science. My argument has two parts. First, I argue that the constructive empiricist's choice of observability as a criterion for ontological commitment is based on the assumption that belief in the existence of unobservable entities is unreasonable because belief in the existence of an entity can only be vindicated by its observation. Second, I argue that the kind of ontological commitment that is under consideration when accepting a scientific theory is commitment to what I call theoretical kinds and that observation can vindicate commitment to kinds only in exceptional cases. (c) 2006 Elsevier Ltd. All rights reserved.