Constructive Empiricism: From a Theory of Empirical Adequacy to a Theory of Acceptance
Dissertation, The University of Rochester (
1995)
Copy
BIBTEX
Abstract
I begin chapter I by discussing two key distinctions that constitute the core of van Fraassen's constructive empiricism: a distinction between observables and unobservables and a distinction between acceptance and belief with regard to a theory. To support constructive empiricism, van Fraassen also deploys two epistemological principles: only actual observations are to be taken as evidence and possible evidence is all that can be rationally inferred from the actual evidence. I reject both principle and van Fraassen's construal of observation. As does van Fraassen, I make the distinction between acceptance and belief in chapter II. I contend that one does not need to be an antirealist to espouse this distinction between acceptance and belief. ;In the remaining chapters, I develop a quasi-Bayesian account of acceptance as an alternative to van Fraassen's. In chapter III, I consider three views of acceptance: van Fraassen's, Bayesian views, and the epistemic view. I argue that both van Fraassen's account and the epistemic account are open to Bayesian objections. However, Bayesianism assumes a controversial principle called "maximization of expected utility" , which might mitigate the impact of the Bayesian objections. ;In the final chapter, I consider different decision principles, e.g. Kyburg's decision principle, and Gardenfors and Sahlin's decision principle. I formulate a decision principle called the "weak-dominance-decision principle " based on Kyburg's theory . The interval-based probability yields, as a limiting case, the point-valued probability when we have complete information about the event in question. In this case, my decision principle reduces to MEU. Therefore, the same Bayesian objections against both van Fraassen's theory of acceptance and the epistemic theory of acceptance could be made from the perspective of my decision principle while escaping the perils of strict Bayesianism. My theory of acceptance is quasi-Bayesian, because WDP rests on interval-based probability