Epistemic Peer Disagreement

About Peer Disagreement

When two epistemic peers—agents who are equally informed, intelligent, and rational—disagree about a proposition, what should each believe? This is one of the central problems in social epistemology, revealing deep tensions between respecting evidence and respecting peers.

Three Main Views

• Equal-weight view: Upon discovering disagreement, you should split the difference, giving equal weight to your credence and your peer's. If you believed P(H) = 0.7 and your peer believed P(H) = 0.3, you should both move toward P(H) = 0.5.
• Steadfast view: Stick to your guns. Your credence is based on your assessment of the total evidence, and learning about disagreement doesn't provide new first-order evidence. Maintain your original belief.
• Total evidence view: The disagreement itself is evidence—about the reliability of reasoning, the ambiguity of evidence, or the difficulty of the question. Update using Bayesian conditioning on the fact of disagreement.

Convergence Dynamics

If both agents adopt the equal-weight view and repeatedly share their credences, they will converge to agreement exponentially fast. But if one or both are steadfast, persistent disagreement is possible even with shared evidence. The total evidence view lies between these extremes: agents may converge slowly, or remain separated if disagreement is strong evidence of genuine ambiguity.

Philosophical Implications

• Epistemic humility: The equal-weight view embodies recognition that you might be wrong, treating your peer's judgment as additional evidence.
• Higher-order evidence: Disagreement is evidence about evidence—a meta-level signal about the reliability of your first-order reasoning.
• Self-undermining worry: If you think your peer is irrational for disagreeing, are they still a peer? Does applying equal-weight to someone who rejects equal-weight make sense?
• Expert disagreement: This problem becomes urgent in domains with genuine expert disagreement: climate science, medical diagnoses, moral philosophy, etc.
• Rational polarization: Even initially similar priors can lead to persistent disagreement if agents weight their own evidence more heavily.

Historical Context

The peer disagreement debate emerged prominently in early 2000s epistemology through work by Thomas Kelly (steadfast), Richard Feldman, and Adam Elga (equal-weight). It connects to older debates about testimony, social epistemology, and disagreement in science (Kuhn's incommensurability, Feyerabend's epistemological anarchism).

The problem has practical stakes: Should you adjust your political views after talking to an equally informed friend who disagrees? Should doctors defer to colleagues with different diagnoses? Should scientists treat peer disagreement as reason to reduce confidence? The answers depend on whether peers are truly epistemic equals, whether the evidence is genuinely shared, and what policy best tracks truth over time.