A fully logical inductive logic (Q807614)

The author maintains that Carnapian inductive logic incorporates an a priori bias in favour of more orderly worlds. Consequently he proposes an alternative approach to inductive logic that abjures this supposed bias. In order to do this he suggests the adoption of a fully logical inductive logic, that is, an inductive logic depending only on a priori assumptions. The assumption proposed in the paper is that there is a specific upper bound on the number of kinds of things. The author thinks that this is the simplest form of the principle of the uniformity of nature. In the reviewer's opinion this paper is not a progress in the studies on inductive logic and probability. The reason is as follows. The Carnapian parameter \(\lambda\) has been seen only as determining the (subjective) weights of logical and empirical factors. From this point of view there follows the idea to cancel empirical factors and assume a priori conditions that directly enable to learn from experience. But as a matter of fact, \(\lambda\) via \(\eta\) is a way to fix the correlation among individuals of the analyzed universe. This correlation can be positive, in this case it favours more orderly universes. It can be zero, in this case the probability only depends on logical factors. But it can also be negative, in this case it favours disorderly universes. Obviously this is the case if the individuals of the examined universe are negatively correlated. Dismissing \(\lambda\), we loose the possibility to analyze correlation. No inductive logic can accept such a loss.