The Fragmentation of Reason: Precis of Two Chapters

My four distinguished critics have made my task a bit easier by aiming almost all of their comments at two of the chapters in The Fragmentation of Reason. In those chapters I offer arguments against two widely accepted accounts of what it is for one cognitive strategy to be better than another. The first account gives pride of place to the analysis of our ordinary notions of cognitive evaluation-notions like justification. The second links cognitive assessment to truth. Though there is a fair amount going on in these two chapters, their central arguments share a common structure. I'll begin by sketching that structure and saying something about how I would like the arguments to be understood. This will, I hope, be of some help to readers who are not familiar with the book. It will also provide a useful backdrop against which I can set out my responses to some of my critics' objections. My preoccupation with the evaluation of cognitive strategies began when I learned about the work of psychologists like Nisbett, Ross, Tversky, Kahneman and others whose experiments seemed to show that bright people in unthreatening environments often do a very bad job in a variety of reasoning tasks. A very natural reaction to this work is to wonder how we might improve matters. Are there steps we can take that will enable people to do a better job of forming and revising their beliefs? This question suggests two rather different lines of inquiry. One is largely empirical: What sorts of interventions will succeed in changing the way people go about the business of reasoning? The other is normative: What sorts of changes would be desirable? What is it that makes one strategy of