Scott Alexander critiques Bryan Caplan's constraints vs preferences model of mental illness, proposing instead a goals vs urges framework that better explains both mental and physical health issues.

Scott explains why 'wanting' and 'liking' diverge by examining reinforcement learning mechanisms like context-dependence, intermittent reinforcement, and hyperbolic discounting, concluding that this undermines the concept of stable preferences.

Scott argues that the 'Cynic's Theory of Akrasia' (which says people's actions reveal their true preferences, contradicting their stated preferences) is functionally equivalent to the naive theory, just valuing unconscious over conscious preferences, and defends prioritizing conscious preferences because they're rational and morally informed.