Applied logic without psychologism (Q934820)

The author discusses broad criteria that might be used for deciding whether one logical system is better than another for representing problems in a given domain and solving them. He opposes the view that there is always one system, perhaps not yet formulated, that is intrinsically correct while all others are incorrect, and the related practice of treating syntactic rationality constraints or semantic structures as forced upon us on the basis of their formal elegance, simplicity, or initial intuitive appeal. He also opposes the view from the other end of the spectrum, that such assessments cannot be made except from within the perspective of a given logical system, so that external comparison of merits is impossible. Against these positions, he presents his own: when constructing a logic for a particular domain we need carefully to examine the fit between the two. We should clarify the structure of the domain, assess the expressive and deductive powers of the logic in application to that domain, as well as consider other factors such as the computational complexity of the system. There will be costs, benefits, and trade-offs, and we need to strike a reasonable balance between them. The discussion is illustrated by examples, drawn particularly from qualitative logics for uncertain reasoning (so-called nonmonotonic logics). Reviewer's comments: The author calls the first kind of view ``Fregean psychologism'', the second ``logical pluralism'', and his own ``pure invariantism''. The reader may find all three labels rather distracting. In the reviewer's opinion, the first approach has only tenuous connections with Frege, even less with psychologism; its main feature is its doctrinaire insistence on the existence of a single -- philosophically correct -- representation for each domain. The second approach certainly involves pluralism but, as the author remarks, is open to criticism precisely because it goes much further: it encourages tolerance of plurality to the point of complaisant indifference. And while the author suggests that, under his own approach, logical systems may be seen as schemes of invariance under suitable permutations, no attempt is made in the paper to give any specific content to this idea.