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 Alexander draws parallels between OpenAI's GPT-2 language model and human dreaming, exploring their similarities in process and output quality. Longer summary

Scott Alexander compares OpenAI's GPT-2 language model to human dreaming, noting similarities in their processes and outputs. He explains how GPT-2 works by predicting next words in a sequence, much like the human brain predicts sensory input. The post explores why both GPT-2 and dreams produce narratives that are coherent in broad strokes but often nonsensical in details. Scott discusses theories from neuroscience and machine learning to explain this phenomenon, including ideas about model complexity reduction during sleep and comparisons to AI algorithms like the wake-sleep algorithm. He concludes by suggesting that dream-like outputs might simply be what imperfect prediction machines produce, noting that current AI capabilities might be comparable to a human brain operating at very low capacity. Shorter summary

Scott reviews a paper proposing a computational model of mood and emotions based on predictive processing, discussing its implications for understanding mood disorders. Longer summary

This post discusses a paper by Clark, Watson, and Friston that proposes a computational perspective on mood and emotions. The authors argue that emotions reflect changes in the uncertainty about the somatic consequences of action, while mood corresponds to hyperpriors about emotional states. The theory suggests that depression is a prediction of bad outcomes with high confidence, mania is a prediction of good outcomes with high confidence, and anxiety is a prediction of bad outcomes with low confidence. The post explores how this theory explains various aspects of mood disorders and their symptoms, including learned helplessness and the role of serotonin. The author finds the theory intriguing but notes some inconsistencies, particularly in unifying the concepts of 'prior on bad outcomes' and 'low precision of predictions'. Shorter summary

Scott Alexander attempts to understand and explain Karl Friston's complex 'free energy' principle in neuroscience, exploring its various interpretations and potential implications. Longer summary

Scott Alexander explores Karl Friston's work on 'free energy', a complex concept in neuroscience that has been described as a unified brain theory. The post delves into various interpretations of free energy, from its mathematical origins in Bayesian equations to its application in explaining perception, cognition, homeostasis, and action. Scott struggles to fully grasp the concept, acknowledging its complexity and the widespread difficulty in understanding Friston's work. He presents different perspectives on free energy, including its role in uncertainty reduction, its connection to predictive processing, and its potential to explain biological systems and the origin of life. The post concludes with a tentative synthesis of these ideas and raises questions about the implications of the free energy principle for machine ethics. Shorter summary