Hodges on the normativity of logic in The Reasoner

(This is a shorter version of something I just posted at NewAPPS, as M-Phi readers probably do not need to be introduced to The Reasoner.)

I'll be interviewing Keith Stenning for The Reasoner, and to prepare for it I was re-reading some of the previous interviews. I found the following delightful passage by Wilfrid Hodges (in the November 2011 issue):

Incidentally I have no patience at all with the view of Kant, followed by Frege and some modern writers, that logic studies how we ought to think and psychology studies how we do think. A logician can tell you that if you reason by rule X, then you will sometimes find yourself deducing false conclusions from true premises. It does follow that if you want never to deduce false conclusions from true premises, you ought not to use rule X. So for example you ought not to use rule X in a research paper in pure mathematics. But in real life, where time and memory are often limited and premises are often dubious in one way or another, rule X might be for practical purposes exactly what you need. One of the major achievements of logic of the last fifty years is to start taking seriously the constraints under which we reason, and the different aims that we can have in our reasoning. This expansion of logic gives many openings for collaboration between logicians and cognitivists.

I couldn’t agree more, and have argued at different occasions that the view of logic as having normative import for thought is entirely misguided. It is a relic of Kantian transcendental idealism that most philosophers still hold on to, but usually somewhat uncritically. What Hodges is referring to in “one of the major achievements of logic of the last fifty years is to start taking seriously the constraints under which we reason, and the different aims that we can have in our reasoning” is probably developments in computer science and artificial intelligence, as philosophers themselves (with honorable exceptions such as Gilbert Harman) remain by and large oblivious to the issue of constraints on reasoning. It is about time that we catch up!

(By this I do not mean to imply that logic has no normative import whatsoever: I'm questioning the idea that it has normative import specifically for thought. I do think that there is something deeply normative about logic, but it has to do with multi-agent, public sphere practices of argumentation rather than with mono-agent, private sphere practices of thinking. But that's a topic for a different occasion.)