Book Review: Surfing Uncertainty

Scott's lucid exposition (via Andy Clark's book) of the predictive processing model of the brain: a multi-layer prediction machine where a bottom-up sense-data stream and a top-down prediction stream meet at every level, are integrated by precision-weighted Bayes, and minimize 'surprisal' — yielding 'perception as controlled hallucination.' He then deploys it to explain a vast range of phenomena: attention (confidence intervals), imagination/dreaming, robust priming, learning/hyperpriors (Molyneux's problem), motor action (active inference — you move by strongly predicting the movement), why you can't tickle yourself, placebo/pain-gating, Asch conformity, neurochemistry (NMDA/AMPA/dopamine), autism (over-precise priors / weak central coherence), and schizophrenia (weak priors → delusions/hallucinations). Closes with the epistemological payoff: PP explains exactly how and how much expectation shapes perception — sympathetic to 'voodoo psychology' but mathematizing it, and predicting that placebo/basic-priming work while stereotype-threat/social-priming don't — and top-down stays advisory rather than steamrolling reality, which grounds the rationalist project.

Quality 88 · Excellent. Excellent, upper. A major, exceptionally lucid synthesis of a genuinely hard topic (theoretical neuroscience) with vast explanatory scope — the canonical piece you send someone to understand predictive processing, and one of the most load-bearing essays for the rationalist community (it seeded the whole 'Bayesian brain' / later 'trapped priors' frame). Scored on merit alongside Categories/Niceness/Ivermectin (88); held just below the very top because the framework is Clark's/Friston's exposition rather than Scott's own paradigm.

Claude’s paradigm shift 56 · Moderate. Moderate–Notable (56). The Georgism rule applies: predictive processing was already 'investigated, organized into a paradigm, tested' before this review (Clark 2015, Friston), so popularizing it is impact (A), not novelty (B). Credited at the upper-Moderate level because PP was still a genuinely fresh unifying brain theory at publication and the epistemological coda (PP predicts which priming replicates; PP grounds anti-bias) is Scott's own.

Real-world impact 4 · Moderate. Moderate (4). This review is the primary vector by which predictive processing / 'controlled hallucination' / precision-weighting entered rationalist and educated-lay discourse, where it became durable, widely-used vocabulary — comparable to a durable widely-adopted coinage (asymmetric-weapons tier). Capped at 4: no material-world/legal change, and the framework itself is not Scott's.