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. Longer summary

Scott Alexander explores the puzzling behavior of depressed people preferring sad music, despite it making them feel worse. He proposes a theory based on control theory and set points, comparing depression to conditions like fever and anorexia. The post suggests that depression might involve a recalibrated happiness set point, where the body defends an abnormally low mood through both biological and behavioral mechanisms. Scott also discusses potential research directions and connects this theory to predictive coding concepts. Shorter summary

Scott explores the puzzling relationship between congenital blindness, schizophrenia, and autism, discussing potential explanations and the need for further research. Longer summary

Scott Alexander discusses the intriguing psychiatric trivia that congenitally blind people never develop schizophrenia, but are 50 times more likely to be diagnosed with autism. He explores this in the context of the diametrical model of autism vs. schizophrenia, which suggests these conditions are opposites in many ways. Scott speculates on possible explanations, including the idea that blindness might shift certain neural hyperparameters towards the autism end of a spectrum. He acknowledges the complexity of the issue and the need for further research, mentioning studies on sensory deprivation and its effects on both conditions. Shorter summary

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. Longer summary

Scott Alexander reviews George from CerebraLab's analysis of a paper by Nutt and Carhart-Harris on serotonin receptors. The post explores two main points: the relationship between prediction error and suffering in the context of active inference theory, and the different effects of 5-HT1A and 5-HT2A receptor activation. Scott discusses how these concepts relate to depression, psychedelics, and problem-solving approaches. He also considers George's suggestion that using psychedelics for introspection might be more harmful than using them for fun or practical problem-solving. Shorter summary

Scott Alexander discusses recent research unifying predictive coding in the brain with backpropagation in machine learning, exploring its implications for AI and neuroscience. Longer summary

Scott Alexander discusses a recent paper and Less Wrong post that unify predictive coding, a theory of how the brain works, with backpropagation, an algorithm used in machine learning. The post explains the significance of this unification, which shows that predictive coding can approximate backpropagation without needing backwards information transfer in neurons. Scott explores the implications of this research, including the potential fusion of AI and neuroscience into a single mathematical field and possibilities for neuromorphic computing hardware. Shorter summary

Scott Alexander reviews his intellectual progress in the 2010s across various fields, crediting his blog for accelerating learning. Longer summary

Scott Alexander reflects on the intellectual progress he made in the 2010s, largely tracked through his blog posts. He discusses major updates in his understanding of topics like predictive coding theory, psychedelics, mental health, SSRIs, genetics, willpower, nutrition, x-risk, AI, scientific progress, cost disease, the replication crisis, placebo effect, ethics, politics, cultural evolution, social justice, game theory, and enlightenment. He notes that while some were major shifts in position, many represent 'deconfusion' - gaining a better map of the problem space. Scott credits his blog and reader interactions for accelerating his learning in the latter half of the decade. Shorter summary

Scott reviews a paper proposing that psychedelics work by relaxing priors in the brain, potentially treating mental illness but also risking side effects. Longer summary

This post reviews a paper by Friston and Carhart-Harris that uses predictive coding theory to explain the effects of psychedelic drugs. The authors argue that psychedelics 'relax' priors in the brain, allowing for new perspectives and potential therapeutic benefits. They suggest this mechanism could help treat most mental illnesses by allowing patients to break free from maladaptive priors. The post discusses the theory's implications, including potential downsides like HPPD and increased belief in pseudoscience. It also mentions connections to meditation and prior work by other researchers. Shorter summary

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. Longer summary

Scott Alexander compares two theories of cognition and behavior: Free Energy/Predictive Coding (FE/PC) and Perceptual Control Theory (PCT). He argues that while they've developed differently, their foundations are essentially the same. Scott suggests that understanding PCT, which he finds more intuitive, can help in grasping the more complex FE/PC. He provides a glossary of equivalent terms between the two theories and gives examples to illustrate how PCT's terminology often makes more intuitive sense. The post concludes by discussing why FE/PC is more widely used despite PCT's advantages in explaining certain phenomena, and suggests teaching both terminologies to aid understanding. Shorter summary

Scott reviews Kuhn's 'Structure of Scientific Revolutions', finding its thesis on paradigm shifts in science intriguing but frustratingly vague, especially beyond physics. Longer summary

Scott reviews Thomas Kuhn's 'The Structure of Scientific Revolutions', which argues that science progresses through paradigm shifts rather than simple accumulation of facts. Kuhn posits that scientific paradigms are comprehensive worldviews that guide research and interpretation of data. When anomalies accumulate, a new paradigm may emerge to replace the old one. Scott finds Kuhn's thesis intriguing but frustratingly vague in parts, especially in applying it beyond physics. He draws connections between Kuhn's ideas and predictive coding in neuroscience, suggesting both describe how pre-existing mental structures shape perception and understanding. Overall, Scott sees value in Kuhn's perspective but wishes for more clarity and examples from other scientific fields. Shorter summary

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. Longer summary

This post explores the mechanisms and potential tradeoffs of cognitive enhancers, focusing on acetylcholine-related substances like piracetam and nicotine. Scott discusses the role of acetylcholine in the brain's predictive coding model, where it's thought to increase sensory precision and learning rate. He speculates on how this might explain the cognitive-enhancing effects of these drugs, but also considers potential downsides like disrupted intuitive understanding or increased risk of psychotic-like thinking. The post ends by acknowledging the highly speculative nature of these ideas and noting some contradictory evidence. Shorter summary