Abstract:

In a naive version the optimality version says the following: If G is a certain cognitive goal and M is a justifiably access-optimal method with respect to goal G (or with respect to a related goal G') and if  in regard to a certain given partition hypotheses H_1, ..., H_n and a given body of evidence E the method M recommends H_i, then (conditional on E): **It is rational to consider H_i as more probable than its alternatives, and hence more probable than 1/n**.

In this talk I discuss the following constraints and problems of this principle at hand of three applications (deduction, induction and abduction):

1) Constraint 1: The class of accessible methods must contain some (domain-specific) standard methods that that satisfy minimal rationality conditions.
2) Constraint 2: One must consider the assertions of all access-optimal methods that are justifiably access-optimal.
3) Problem 1: The option of an ongoing suspension of judgement must not be available.
4) Problem 2: The above diagnosis **...** is not enough; for practical applications one also needs a optimal probability estimation.
5) Problem 3: If the goal is abduction & theoretical explanation, then only the related goals of empirical success and simplicity (but not theoretical truth) is directly empirically testable.

Aufgrund der Corona-Pandemie werden die Vorträge des Forschungsseminars des Lehrstuhls für Theoretische Philosophie im Sommersemester 2020 per Videokonferenz durchgeführt werden. Bitte melden Sie sich, falls Sie an der Videokonferenz teilnehmen möchten, per Email (christian@phil.hhu.de).