Thursday, July 20, 2006

Thoughts on Quine and belief revision

One way of reading Quine is to say that we each have a set of beliefs, connected in an inferential network. This set forms something like a web (hence the metaphor) with a group of core beliefs and a beliefs lying along the edges of the web. Experience impinges on the web only at its edges and modificaitons are made primarily along the edges. Changes can make their way to the interior of the web only if the recalcitrant experience is strong enough to require modifying more central beliefs to preserve as much of the overall structure as possible. This is somewhat appealling as far as it goes, but I realized I don't understand the mechanism behind the change. What I mean is that it sounds like we have beliefs and we get sensory experience of the outside world which constitutes experience; this experience either supports or conflicts with our existing beliefs. If it supports it, so much the better we think. If it conflicts, then we must modify our web, making as few changes as possible. One of the problems is that experiences don't come labelled as recalcitrant or not. They first have to be recognized as such. Somone can have experiences that would seem to undermine other beliefs she holds without seeing those experiences as the basis of conflicting beliefs. She could refuse to draw the inferences necessary to make the conflict apparent. Often the links between what we experience and how that experience impinges on our already held beliefs is separated by a large gulf, so problems are not immediately recognizable in most cases. That is one problem.

Another problem is how the web is revised. Suppose I have some experience E that I view (or come to view) as recalcitrant with regards to my web of beliefs. I could make the minimal changes in my web by changing the truth values I ascribe to various sentences along the periphery of the web. Or, I could change the strength of the inferential connections in different places. I could use E as the basis of a new theory that I will tack onto my web. Or, I could reinterpret E such that it actually supports rather than undermines my beliefs. Or, I could suspend any revisions while I try to figure out if my beliefs are actually compatible with E. Or, I could suspend my revisions because I think there is some crucial information that I need to make sense of the combination of E and my beliefs, e.g. I need to learn more about some topic before I can say for sure what status E has in regards to my beliefs. Or, I could reject the recalcitrant experience because I think that any conflict with my core beliefs is either merely apparent -and not real- or based on reasoning whose fallacies I don't currently detect but am sure of due to the strength of my core beliefs. That makes seven different ways of responding to recalcitrant beliefs, some of which do not involve any modification to my web, even along the edges. I think that Quine had the fairly straight-forward minimal revision (of the sort found in some of the current belief revision literature in logic) in mind. No doubt some of these options can be removed by stipulation or idealization, but there are a lot of options. I guess my points are that belief revision shouldn't be taken as a functional process, experience needs to be interpreted (or recognized) as recalcitrant, and recalcitrant experience need not entail even minimal revision.

No comments: