Deterministic Falsificationism

There exists [34] a version of phenomenalism (section 2.1.1) which prohibits the entry into positivist knowledge of any theory, for which there is no conceivable piece of observational evidence which will entirely rule out the theory; in physics, Rutherford was [22,46] a proponent of this viewpoint, asserting that, rather than use statistics to interpret an experiment, one should perform a new experiment which discriminated between theories more decisively. The Bayesian mathematical model of this is simple: a positivist who subscribes to this view will not accept a theory $T$ as valid, and will therefore assign to it a prior probability $P(T)=0$, unless there is some conceivable piece of evidence $E$, for which $P(T\vert E) = P(E\vert T) = 0$. This restriction is unnecessary for a Bayesian, since Bayes' theorem is capable of generating inferences about the truth of any theory that meets the weaker version of phenomenalism (section 2.1.) In addition, it assigns a fixed prior probability to theories of this type, and is therefore, as far as these theories are concerned, an example of logical probability (section 3.1.) Garrett [24] calls this version of phenomenalism falsifiability, and also notes that it is not a reasonable principle for a Bayesian to adopt.