A thread through the archive
Essays tagged “predictive coding”. Follow an idea across the archive.
Scott shares his monthly collection of interesting links from around the internet, covering topics from Jeremy Bentham's linguistic legacy to AI developments, with characteristic commentary and fact-checking caveats.
Scott argues that dismissing AI as 'just a next-token predictor' is like dismissing humans as 'just reproduction machines' - both confuse the optimization process that shaped an entity with how that entity actually thinks.
Scott Alexander proposes a theory of depression as a recalibrated happiness set point, drawing parallels with fever and anorexia, to explain why depressed people seek out sad stimuli.
Scott explores the puzzling relationship between congenital blindness, schizophrenia, and autism, discussing potential explanations and the need for further research.
Scott Alexander examines a review of a paper on serotonin receptors, discussing prediction error, suffering, and the effects of different serotonin receptor activations on problem-solving approaches and psychedelic experiences.
Scott Alexander discusses recent research unifying predictive coding in the brain with backpropagation in machine learning, exploring its implications for AI and neuroscience.
Scott Alexander reviews his intellectual progress in the 2010s across various fields, crediting his blog for accelerating learning.
Scott reviews a paper proposing that psychedelics work by relaxing priors in the brain, potentially treating mental illness but also risking side effects.
Scott Alexander argues that Free Energy/Predictive Coding and Perceptual Control Theory are fundamentally the same, and proposes using PCT's more intuitive terminology to help understand FE/PC.
Scott reviews Kuhn's 'Structure of Scientific Revolutions', finding its thesis on paradigm shifts in science intriguing but frustratingly vague, especially beyond physics.
Scott explores how acetylcholine-related cognitive enhancers might work by increasing sensory precision and learning rate, while speculating on potential tradeoffs and acknowledging the highly speculative nature of these ideas.