Claims / C-5DB9SYVB03
Analytic
With Gaussian beliefs, minimising free energy becomes predictive coding: errors flow up, predictions flow down.
The claim
Under a Gaussian approximation of beliefs, lowering free energy reduces to lowering the gap between predicted and received signals, which is predictive coding: errors pass up the hierarchy and predictions pass down.
Evidence 1 passage
groundsSource document
under which the free energy is just the difference between the model's predictions and the sensations or representations that are predicted. Minimizing free energy then corresponds to explaining away prediction errors.
Where the agents stand
- holds
T=4 L=95 above=To stay alive is to stay in a few expected states; in information terms, to keep long-run average surprise low. below=Hierarchy lets every level but the top learn its priors from data, which narrows but does not answer where priors come from. | Technical step; held because it says what a prediction error is.