Wednesday, August 26, 2015

Let's not model deontic constraints with infinite disutilities

It's natural to model deontic constraints in decision theory by assigning infinite disutility to forbidden actions. This temptation should be resisted. There are too many deontic theories with non-zero probability, and since an infinite disutility multiplied by a non-zero number is still infinite, we would have to take all these deontic theories extremely seriously. And that would lead to constant weighing of infinities against each other and/or an unduly restricted life that must obey prohibitions from fairly crazy (but not so crazy as to have zero probability) theories.

3 comments:

Heath White said...

You may find this relevant.

http://marginalrevolution.com/marginalrevolution/2015/08/the-difficulty-of-cross-theoretical-aggregation.html

Alexander R Pruss said...

Thanks for this very helpful reference.

entirelyuseless said...

Right, a deontic constraint does not assert that anything is infinitely bad. It just asserts that the action, as an action, is bad, although of course finitely bad.

The point of such a system is that something bad is bad, regardless of what you add on afterwards. That is why St. Paul says of those who say "let us do evil that good may come, their condemnation is just." Because by stipulation they are doing evil, no matter what good comes of it.