Astral Codex Ten

predictive coding

5 min read

predictive coding

Article

predictive coding is a recurring concept in the Astral Codex Ten archive, appearing 5 times across 5 issues between March 26, 2021 and February 26, 2026. The archive places it in contexts such as “the predictive coding community uses a different one”; “Predictive coding is the most plausible current theory of how the brain works”; “First of all - predictive coding identifies suffering with prediction error”. It most often appears alongside 5-HT1A, 5-HT2A, active inference.

Metadata

  • Category: Concepts
  • Mention count: 5
  • Issue count: 5
  • First seen: March 26, 2021
  • Last seen: February 26, 2026

Appears In

Source Context

Recovered passages from the original issue text. When the raw archive preserved outbound links inside the source passage, they are listed directly under the quote.

Toward A Bayesian Theory Of Willpower
March 26, 2021 · Original source
The economic metaphor here is cute, but the predictive coding community uses a different one: they describe it as representing the "confidence" or "level of evidence" for a specific calculation. So an alternate way to think about lampreys is that the flee-predator region is saying "I have VERY VERY strong evidence that fleeing a predator would be the best thing to do right now." Other regions submit their own evidence for their preferred tasks, and the basal ganglia weighs the evidence using Bayes and flees the predator.
Inline links: predictive coding
[LINK] Unifying Predictive Coding With Backpropagation
April 14, 2021 · Original source
[epistemic status: I know a little about the predictive coding side of this, but almost nothing about backpropagation or the math behind the unification. I am posting this mostly as a link to people who know more.]
This is a link to / ad for a great recent Less Wrong post by lsusr, Predictive Coding Has Been Unified With Backpropagation, itself about a recent paper Predictive Coding Approximates Backprop Along Arbitrary Computation Graphs.
Inline links: Predictive Coding Has Been Unified With Backpropagation, Predictive Coding Approximates Backprop Along Arbitrary Computation Graphs
Predictive coding is the most plausible current theory of how the brain works. I’ve written about it elsewhere, especially here.
Inline links: here
Carhart-Harris On Serotonin
June 15, 2021 · Original source
First of all - predictive coding identifies suffering with prediction error. This conflicts with common sense. Suppose I tell you I'm going to stab you in the arm, you agree that I'm going to stab you in the arm, and then I stab you in the arm, and it hurts a lot. You predicted what would happen correctly, but you still suffered. The theory resolves this with a distinction between common-sense-level and neurological predictions: your brain is "set" to expect normal neurological feedback from your arm, and when it gets pain signals instead, that's a violated prediction, and this is the level on which prediction error = suffering. But there are other cases where the common-sense and neurological sense of predictions are more congruent. When you first step into a cold shower, you feel suffering, but after you've been in it a while you adjust your "predictions" and it's no longer as unpleasant. If you unexpectedly lost $25,000 it would come as an extremely unpleasant shock, but when you predictably have to pay the taxman $25,000 each year you grumblingly put up with it.
Singing The Blues
January 03, 2024 · Original source
I’ve previously endorsed predictive coding theories of depression and other illnesses. How does that interact with this perspective? This is even more speculative than the rest, and I don’t feel like I entirely get it, but here’s the completion my internal pattern-generator has spit out:
Many of the claims of predictive coding can be rephrased as claims about control theory, and vice versa. You have to slightly fudge things to make this work on homeostatic bodily processes, but this is the kind of fudging that Karl Friston has already worked into his free energy concept.
Inline links: Many of the claims of predictive coding can be rephrased as claims about control theory, free energy
In predictive coding, the equivalent of control theory’s “set point” is the “prior”. This suggests an elegant equivalence: an incorrectly fixed set point, like those in anorexia and depression, are the same thing as a trapped prior.
Inline links: a trapped prior
Next-Token Predictor Is An AI's Job, Not Its Species
February 26, 2026 · Original source
In neuroscience, predictive coding (also known as predictive processing) is a theory of brain function which postulates that the brain is constantly generating and updating a “mental model” of the environment. According to the theory, such a mental model is used to predict input signals from the senses that are then compared with the actual input signals from those senses.
These algorithms are a second optimization loop, “evolving” neuron patterns into forms that better promote fitness, reproduction, etc. The most powerful such algorithm is called predictive coding, which neuroscience increasingly considers a key organizing principle of the brain. Wikipedia describes it as:
Inline links: predictive coding, Wikipedia describes it as