How to avoid getting lost reading Scott Alexander and his 1500+ blog posts? This unaffiliated fan website lets you sort and search through the whole codex. Enjoy!

See also Top Posts and All Tags.

Minutes:
Blog:
Year:
4542 tags
Show all filters
5 posts found
Apr 14, 2021
acx

[LINK] Unifying Predictive Coding With Backpropagation

5 min • 540 words • 85 comments • 46 likes • podcast
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
Mar 26, 2021
acx

Toward A Bayesian Theory Of Willpower

16 min • 1,991 words • 421 comments • 137 likes • podcast
Scott Alexander proposes a Bayesian theory of willpower as a process of weighing evidence from different mental processes to determine actions. Longer summary
Scott Alexander proposes a new Bayesian theory of willpower, disagreeing with previous models like glucose depletion, opportunity cost minimization, and mental agent conflicts. He suggests willpower is a process of weighing evidence from different mental processes: a prior on motionlessness, reinforcement learning, and conscious calculations. The basal ganglia then resolves this evidence to determine actions. Scott explores how this model explains the effects of dopaminergic drugs on willpower and discusses implications for understanding mental illness and productivity. Shorter summary
Jan 31, 2018
ssc

Powerless Placebos

10 min • 1,173 words • 115 comments • podcast
The post examines research suggesting placebos are largely ineffective, explores exceptions, and proposes explanations for observed placebo effects based on brain function and study design. Longer summary
This post discusses the placebo effect, focusing on Hróbjartsson and Gøtzsche's research which found that placebos have little benefit over no treatment in most conditions. The author explores the reasons behind this, suggesting that regression to the mean may have been mistaken for placebo effects in earlier studies. The post also discusses exceptions where placebos seem more effective, such as pain and nausea, and offers a potential explanation for the placebo effect based on the brain's predictive processing model. The author concludes by speculating on why depression might not show strong placebo effects in clinical trials despite common assumptions. Shorter summary
Sep 06, 2017
ssc

Predictive Processing And Perceptual Control

8 min • 961 words • 78 comments • podcast
Scott Alexander explores the similarities between Predictive Processing and Perceptual Control Theory, arguing that PCT anticipates many aspects of PP and deserves recognition for its insights. Longer summary
Scott Alexander draws parallels between Predictive Processing (PP) and Perceptual Control Theory (PCT), suggesting that PCT anticipates many aspects of PP. He argues that both theories share the concept of cognitive 'layers' acting at various levels, with upper layers influencing lower layers to produce desired stimuli. Scott notes that PP offers a more refined explanation for higher-level cognitive processes compared to PCT's sometimes overly simplistic model. He concludes by comparing Will Powers, the originator of PCT, to ancient Greek atomists like Epicurus, suggesting that Powers' work deserves recognition for its prescient insights, even if it has been superseded by more advanced theories. Shorter summary
Mar 06, 2017
ssc

Book Review: Behavior – The Control Of Perception

30 min • 3,846 words • 164 comments • podcast
Scott reviews Will Powers' book on Perceptual Control Theory, finding it convincing for explaining lower-level brain functions but skeptical of its application to higher-level concepts. Longer summary
This post reviews the book 'Behavior: The Control Of Perception' by Will Powers, which presents Perceptual Control Theory (PCT) as a paradigm for understanding human behavior and brain function. The review explains how PCT views the brain as a hierarchy of control systems, from low-level muscle control to high-level abstract concepts. The author finds the theory convincing for lower-level systems but becomes skeptical of its application to higher-level concepts. The review discusses the theory's strengths in explaining certain brain functions and its historical context as a response to Behaviorism, while also pointing out its limitations and potential unfalsifiability at higher levels. Shorter summary