Toward A Bayesian Theory Of Willpower

Scott's model of willpower as a Bayesian evidence-weighing process in the basal ganglia: competing processes (a prior on motionlessness, a reinforcement learner, high-level conscious calculation) submit 'evidence'/confidence, coded by dopamine, and the highest-evidence action wins. Explains stimulants (boost frontal-cortex evidence -> more willpower + overconfidence + hallucinations), antipsychotic catatonia, the prior on muscular immobility (the finger-wiggle demo), alcoholic 'hitting bottom' as overwhelming evidence, and why we lack infinite willpower.

Quality 75 · Excellent. Excellent-floor: a genuinely original, generative model with strong explanatory reach across drug effects, movement, and addiction; held at the low end because it is explicitly conjectural and yields no immediate actionable advice (Scott says so).

Claude’s paradigm shift 58 · Moderate. Notable: framing willpower as Bayesian inference in the predictive-coding paradigm is a fresh, non-obvious synthesis at publication, distinct from the glucose and subagent theories it displaces.

Real-world impact 2 · Minor. Minor/within-discourse: an influential-within-the-blog neuroscience model with no material real-world effect.