Rationing rationality
Mon 30 Jun 2014 09:21 AM
I wish I had read Rogier De Langhe's A Unified Model of the Division of Cognitive Labor a while ago.
I recently submitted the final draft of Science and rationality for one and all, my paper on reconciling individual and collective rationality. Referees had worried that I was misrepresenting what philosophers of science thought about the relation between the two.
In a passage I would have quoted in my paper if I had read it earlier this year, De Langhe nicely lays out the way that the matter is usually understood: "The fundamental normative question when dividing cognitive labor is to what extent agents ought to take into account the actions of others. This question typically reduces to the specification of a normative balance between the rational and the social: 'leaders' choose that theory which they deem intrinsically superior, and 'followers' choose that theory which others choose." Looking for a balance between the rational and the social presumes that they are two competing standards, and the thought that leaders respond to the intrinsically superior theory suggests that they are the genuinely rational ones.
In his analysis, De Langhe defuses the tension. He allows the space of possible theories to be dynamic, so that scientists might either accept the available theory that looks best or search for a better one. He writes that this "makes it possible to shift the question instead to finding a normative balance between the actual and the possible: 'exploiters' exploit existing theories, and 'explorers' explore new theories."
In conclusion, he writes, "adoption and production pull in opposite directions." (Once I write it out, this is perhaps just what Kuhn called the essential tension in science. Kitcher was explicitly inspired by Kuhn, but De Langhe's model improves on Kitcher's in capturing that idea.)
For my purposes, it is important to add that there is no neutral way to determine the importance of exploiting and of exploring. Both reflect personal and social values.