It's widely supposed that unification is an epistemic virtue: the degree to which a theory is unified contributes to its overall confirmation. However, this supposition has consequences which haven't been noted, and which undermine the leading accounts of unification. For, given Hempel's equivalence condition, any epistemic virtue must be such that logically equivalent theories must equally well unify any body of evidence, and logically equivalent bodies of evidence must be equally well unified by any theory. Yet the leading accounts of (...) unification in Bayesian terms, or those in terms of argument patterns, cannot satisfy these constraints conditions. The reason for this runs deep: these accounts of unification make unity depend on factors that vary between equivalent theories: the probabilistic relations of their components, or their relations to argument patterns. The solution is to abandon such accounts and instead adopt an account of unity based on worldly relations such as causation, rather than inferential relations. Such an account effortlessly satisfies the equivalence conditions, and so may describe the epistemic virtue of unity. (shrink)

According to the popular ‘epistemic virtue account’ (EVA) of theory choice, we should choose between scientific theories on the basis of their epistemic virtues: empirical fit, simplicity, unity etc. More specifically, we should use a rule that aggregates theories’ virtues into a ranking of the overall goodness of the theories. However, an application of Arrow’s impossibility theorem shows that, given plausible premises, there is no rule that can aggregate theories’ virtues into a theory ranking. The EVA-supporter might try to avoid (...) the impossibility result of Arrow’s theorem by asserting that we have more fine-grained distinctions between theories’ epistemic virtues than initially supposed. We show that implausibly fine-grained distinctions between virtue quantities are necessary to escape the impossibility result. This is shown via novel proofs of Arrow’s theorem for cases in which the quantities to be aggregated are measured on any combination of different scales of information, as is likely the case when aggregating epistemic virtues. (shrink)

Moderate realists hold that scientific theories are truthlike, rather than exactly true. Although scientific realism has been challenged by arguments such as the pessimistic induction, moderate realism hasn’t been challenged directly on the grounds that it makes scientific progress rely on inferences from theories that are only truthlike. This paper shows that moderate realism is incompatible with the claim that deductive arguments from scientific theories are reliable. Using truthlike claims as the premises of some patterns of deductive reasoning renders the (...) argument dramatically unreliable. The conclusion is not guaranteed to be true. Nor is the conclusion guaranteed to be at least as truthlike as the premises. Nor even is the conclusion shown to be likely to be true. This is because the consequences of truthlike theories are neither guaranteed to be true, nor even more likely to be truthlike than not. Truthlike theories cannot function like true theories in deductive arguments; instead they function as radically false theories would. In short, truthlike theories behave exactly like radically false theories for the purposes of their deductive consequences. And since scientists would not trust deductions from radically false theories, they should not trust deductions from truthlike theories either. Furthermore, this applies to a wide range of logics and patterns of deductive argument. The moderate realist must either reject bivalence, deny that theories are truth-apt, or accept that scientific theories are not used in deductive arguments. (shrink)