Friday, November 11, 2011

Formulating Theories of Peer Disagreement

Just a quick thought... The "Equal Weight" View (roughly, that epistemic peers should "split the difference" between their credences) is often glossed as the view that you should give your peer's opinion the same weight that you give your own. But opponents of the view need not deny this (at least on one natural reading).

The best alternative views do not hold that your judging that p is better evidence for p than is your peer's judgment. Rather, they hold that this "higher order evidence" -- provided by the judgments of yourself and your peer -- does not exhaust the relevant evidence. While there's an epistemic symmetry at this level, we must also consider the first-order evidence (on which your original judgment was based), which might count in favour of one view over the other.

Indeed, even the most "steadfast" views give equal weight to your own and your peer's judgments (as such) -- for they give no weight at all to either psychological fact; they instead hold that the rational credence is entirely determined by the first-order evidence.

So it seems to me that we would do better to avoid this misleading way of formulating the issue. The question is not how to weight others' judgments against your own; it's about how to weight the higher-order evidence against the first-order evidence.

4 comments:

  1. http://www.overcomingbias.com/2009/02/share-likelihood-ratios-not-posterior-beliefs.html would seem to be relevant.

    ReplyDelete
  2. "The question is not how to weight others' judgments against your own; it's about how to weight the higher-order evidence against the first-order evidence."

    I agree wholeheartedly with the part before the semicolon, but have reservations about the part after. Why weight higher-order evidence against first-order evidence at all? They're evidence of different things.

    ReplyDelete
  3. Hmm, I was thinking that considerations of meta-coherence prevent any clean separation here. Though then perhaps my terminology is misleading, since I think that both kinds of evidence speak (whether directly or indirectly) to both the first-order and higher-order questions.

    ReplyDelete
  4. I could not understand the basic criticisms of the equal weight view Kelly advanced in the paper. Maybe that's because I don't know a lot about epistemology.

    His basic criticism seems to be that the EWV cannot account for the intuitiveness of cases where two peers with the same evidence E for some hypothesis H disagree and one of the peers realizes the other has made some sort of reasoning mistake. In this case the peer that realizes the other has made the reasoning mistake is somehow entitled not to give equal weight to the other person's views. But right off the bat, that seems to violate the hypothesis the EWV advocate holds as a precondition for the EWV: that they have the same evidence E. If one peer knows that the other has made a mistake that means that that other person is not aware that his reasoning is not sound and thus does not support his view. So it would seem that this peer has less (perhaps a kind of meta) evidence for H than the other with sound reasoning. Alternatively, the peer with sound reasoning has additional to the evidence she shares with her peer the knowledge that some a certain reasoning is unsound which relates to the differing views on H in question. But that is additional evidence and thus violates the hypothesis the EWV theorist is working under that both share the same evidence. Thus it seems to me that Kelly's criticisms doesn't even touch on the EWV.

    I must be missing something as this criticism seems to be so obvious that Kelly felt it ought not even be mentioned as something to be acknowledged in the paper. So can some reader who is more versed in the peer disagreement literature explain to me what I have missed?

    ReplyDelete

Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)