The Relevance of Coherence

Joseph Raz, Columbia Law School


Coherence is in vogue. Coherence accounts of truth and of knowledge have been in contention for many years. Coherence explanations of morality and of law are a newer breed. I suspect that like so much else in practical philosophy1 today they owe much of their popularity to John Rawls. His writings on reflective equilibrium,2 while designed as part of a philosophical strategy which suspends inquiry into the fundamental questions of moral philosophy, had the opposite effect. They inspired much constructive reflection about these questions, largely veering toward coherence as the right interpretation both of reflective equilibrium and of moral philosophy. In legal philosophy, Ronald Dworkin's work contributed to an interest in coherence accounts of law and of judicial reasoning.3

There were, however, other important influences on the growing popularity of coherence accounts in law and morality. They came from the application of a Davidsonian approach to ethics by Wiggins and McDowell.4 Their work points to the way coherence accounts chime in a vaguer, more pervasive way with the current philosophical climate. Coherence accounts fit well with the rejection of the Cartesian approach to philosophy. For one thing they seem a natural conclusion of the rejection of foundationalism, with its commitment to the view that all justified beliefs are justified by their relations to some incorrigible beliefs. Even those who accept that some beliefs are incorrigible would reject that. Moreover, under the impact of Quine's dual rejection of empiricism (with its belief in incorrigible foundations for all justified beliefs) and the analytic/synthetic distinction, many philosophers embraced holism, that is, the view that everything depends on everything. Coherence accounts, while not logically entailed by holism, seem to go well with it. If everything depends on everything, how is one to distinguish between truths and falsehoods if not by a test of coherence?

All this leads naturally to a frame of mind which, once one gets used to it, turns out to be oddly reassuring. We are all in mid-ocean on Neurath's ship.5 We cannot disembark and make a fresh start with a sound vessel, accepting only safe beliefs. We must use what we have, repairing our ship from within, jettisoning that which, in the light of our current beliefs, corrigible and possibly mistaken as each one of them may be, seems mistaken. "What is reassuring here?" you may ask. Is that not a recipe for skepticism? Not if one is convinced by Wittgenstein, or alternatively by the very different and incompatible argument of Davidson, that such skepticism is incoherent. Davidson's, rather than Wittgenstein's, arguments show the way toward coherence. Davidson concludes that it is incoherent to suppose that all or most of one's beliefs are false. Instead, he argues that to understand people presupposes accepting their beliefs as largely true. This confidence in the essential soundness of Neurath's ship seems to point to coherence as the inescapable solution to our puzzles.

I am not trying to describe a specific thesis here. My aim is to indicate some of the leading elements in the philosophical climate of opinion which make it congenial to coherence-based accounts-which make the air buzz with coherence. I will consider the merit and relevance of coherence in explaining the nature of law and of adjudication. In doing so, I will mention points derived from the writings of theorists who favor coherence. These borrowings notwithstanding, this is not an article about the work of any particular theorist. It is an exploration of the role and value of an idea, and of some of the different forms that it can take.

Herein lies a difficulty. How can one make sure that the main, the most promising and interesting uses of coherence have been examined? That is the claim I make for my discussion, but I'know of no way to prove it. It is possible that there are other more interesting and promising uses of coherence in explanations of law and adjudication than those here considered. With that caveat let us begin.6