Reflective equilibrium is the simple but compelling concept that a person reflects on conflicts between different beliefs that they hold, and revises their beliefs to reduce the conflicts, or to justify/unify some beliefs in terms of others.

Has anyone made a serious attempt at devising a logic that models this process? It seems that such a logic could be a closer match to human reasoning than any deductive logic. Conflict and unity could be measured as real numbers to optimize, so that we are not always dealing with black-and-white contradiction. We might also want some notion where we optimize only a small set of propositions at a time, rather than all of them at once, because that's what humans do.

In order for the process to converge towards a stable set of beliefs, there could be a need for real-valued propositional attitudes, where we do not simply hold P, we hold P with some level of resistance to changing it. Some beliefs that are foundational or encode sensory information might be immutable.

  • Seems to me that it's more like a proof technique than like a logic. Sep 7 at 23:46
  • 1
    Re your "we are not always dealing with black-and-white contradiction", isn't this just Bayesian inductive logic of science? Also per typical Hegelian dialectic logic, the synthesis from thesis and antithesis may not arrive at a simple equilibrium... Sep 7 at 23:49
  • 3
  • 1
    There are things like defeasibility logic, logics of belief revision, and generally non-monotonic logics that structurally/conceptually resemble the thematics of reflective equilibrium. To some extent, the built-in requirement of overlapping consensus and public availability of relevant information makes overly formal spellings-out inadvisable, though Rawls, for example, does carry out the formalism in a very fine-grained way when the "scene calls for it," so to say. Sep 8 at 4:37
  • 1
    @causative I'm going off Rawls' use, esp. in the final section of AToJ. Reflective equilibrium is coherentistic/Quinean, so an attempt at complete formalization would have to encompass subsets of balanced beliefs that can't be very determinable, so trying to formalize those subsets might not fit the "data set," so to say. Sep 8 at 9:56


You must log in to answer this question.

Browse other questions tagged .