Is rationality mainly about keeping your beliefs coherent with each other, or about believing what your evidence really supports—even when the two come apart?
My links: https://linktr.ee/frictionphilosophy.
1. Guest
Alex Worsnip is an Associate Professor of Philosophy at the University of North Carolina at Chapel Hill. His work focuses on rationality, epistemology, and more. His book, "Coherence and the Demands of Structural Rationality", was published by OUP in 2021.
2. Interview Summary
Alex Worsnip Alex Worsnip defends a ‘dualism’ about rationality: structural rationality is (broadly) coherence among one’s attitudes, while substantive rationality is having attitudes that are actually supported by one’s reasons/evidence. He motivates the split with a Superman case: someone can be substantively irrational for holding an unreasonable belief (“I’m Superman”), yet become additionally irrational when their beliefs clash (e.g., “Superman can fly” plus “I can’t fly”), where the new problem is incoherence rather than a fresh evidential mistake. Comparing this person to a “coherent twin” highlights the intuitive pull: coherence can make you more rational in one respect even while you remain (or become) less rational in another, purely evidential respect.
A major theme is how this distinction interacts with evidence. Worsnip argues that someone can be substantively irrational (failing to believe what their evidence really supports) without being structurally irrational, so long as they have a coherent “story” by their own lights—illustrated with a person who knows about expert consensus on climate change but dismisses it via a conspiracy hypothesis or a principled anti-testimony stance. That leads to a clean characterization: structural rationality pressures you to believe what you take your evidence to support, whereas substantive rationality pressures you to believe what your evidence actually supports. He then uses probabilistic reasoning to discuss the preface paradox and argues that “deductive closure” (believing all logical consequences of what you believe) is too strong to be a genuine requirement of structural rationality: probabilistic coherence can force low credence in a long conjunction, and it’s structurally incoherent to outright believe what you regard as more likely false than true.
Later, the discussion widens to normativity and reasons, especially instrumentalism (the idea that reasons are always tied to one’s goals/desires). Worsnip pushes back using epistemic cases where we plausibly have reasons to believe things even absent any relevant goal (his “unicycles in Mexico City” example), and he argues that instrumentalism doesn’t really “demystify” normativity because it must still take some fundamental normative principle for granted (roughly, that goal-advancement itself generates reasons). He also clarifies how ‘reasons’ relate to ‘ought’: reasons are pro tanto contributors toward what one ought to do all-things-considered, and he sees no special added suspiciousness in goal-independent reasons once we’re already comfortable with goal-independent ‘ought’ claims. The interview closes with a defense of philosophy’s value: engaging these questions is part of being reflective about ourselves and the world, and it also has practical payoff by improving clarity, critical thinking, and reasoning—especially through teaching.
3. Interview Chapters
00:00 - Introduction
00:49 - Dualism about rationality
07:33 - Substantive rationality
12:21 - Reasons and evidence
22:39 - Evidential support
29:01 - Unclear terms
36:05 - Brute vs. unclear
43:19 - Instrumental norms
51:06 - Demystifying by stipulation
1:00:13 - Nihilism and metaethics
1:07:09 - Analytic naturalism
1:11:11 - Avoiding false metalinguistic claim
1:12:47 - Preface paradox
1:28:09 - Deductive closure too strong?
1:33:49 - Inconsistency constraint
1:39:17 - Strengthening the hypothetical
1:42:24 - Epistemic contextualism
1:51:23 - Pascal’s wager
1:52:53 - Reasons and well-being
1:53:54 - Is well-being normatively loaded?
1:55:49 - Value of philosophy
2:00:19 - Conclusion
