This work considers the problem of the ontological implications of epistemic success (“the problem of success”), that is, what can be inferred regarding how the world has to be if one only assumes that there has to be some evaluable epistemic access to that world. What ontological assumptions are sufficient for having such access?

It is argued that three assumptions are sufficient for this task:
A1) One can act in the world.
A2) The way one’s actions in this world are constrained is not totally arbitrary.
A3) One is able to interpret some specifications related to how one’s actions would be constrained.

How these three assumptions can account for the possibility of evaluable empirical and theoretical knowledge is demonstrated by utilizing a novel ontological framework called Constrained Ontology. Constrained Ontology takes what is real to be the way in which our dealings with the world are non-arbitrary, that is, how our actions in the world are constrained. This means that the world is not taken to be real in terms of what it contains, but rather in how it affects our dealings with it.

This shift in ontological emphasis has several interesting philosophical features. It allows for non-circular evaluations regarding when a representation is faithful (Chapter 2). It allows for indirect evaluations of theoretical knowledge claims and the various theoretical methods used in science (Chapter 3). These methods include inductive inferences, triangulation, robustness analysis and consistency and evaluability conditions. The use of these methods, on the other hand, can be used in explaining the theoretical success of science, which means that a constrained ontology in a sense provides an ontological explanation for the success of science. Constrained Ontology can also be taken as a structural realist position that can account for scientific realist desiderata without breaking the empiricist spirit (Chapter 4). Finally, a constrained ontology can be used to show that stronger ontological assumptions are not necessary for accounting for the possibility of science being successful. In particular, this implies that one does not need to assume the existence of any substantial entities and all arguments to that effect have to appeal to some further desideratum than the mere possibility of science being empirically and theoretically successful. The success of science does not imply that there are substantial entities.

The work also utilises a condition of evaluability according to which distinctions are well defined only if there is some way to evaluate how they distinguish their objects. There must be some specified way to determine which cases fall to which side of the distinction. That is, a distinction is well defined only if there is some way of determining how it distinguishes the differing cases. Constrained Ontology can be seen as an application of this principle to ontology. What other implications this condition might have is a question well deserving of further study. — Constrained Realism : Ontological Implications of Epistemic Access (2020-06-03) by Petri Turunen