Fundamentals of Active Inference (Chapter 6, Session 29) July 21, 2026 @ActiveInference
Fundamentals of Active Inference (Chapter 6, Session 29) July 21, 2026  @ActiveInference
Uploaded July 2026 | Updated November 2026, 6 hours ago
Book: Fundamentals of Active Inference
Principles, Algorithms, and Applications of the Free Energy Principle for Engineers, by Sanjeev V. Namjoshi https://mitpress.mit.edu/9780262050951/fundamentals-of-active-inference/

Register: http://textbook-group.activeinference.institute/
Code: github.com/ActiveInferenceInstitute/fundamentals

Playlist of Textbook Group: youtube.com/playlist?list=PLNm0u2n1IwdrpoTg-9l37YIBiDrrOzTw-

Direct link to group document: coda.io/d/_dZHp_UGhVZE/Meetings_su3ToqMp

------

CHAPTERS
0:00 Intro: third discussion on Chapter 6
1:19 Why Chapter 6 is a turning point, and where people get stuck
2:24 Which sections matter most: 6.1, 6.3 and Example 6.1
3:39 Algorithm 8: the univariate generalized filtering loop
5:16 Generative process and generative model as environment and agent
6:31 Why a dynamic hidden state changes what the agent needs
7:36 Gradients, learning, and Euler's method
9:00 Prediction errors and update strength
9:57 Many ways to linearise an equation into procedural steps
12:13 Reading the algorithm as pseudo-code
12:39 Step 1: initialising the state space
13:09 The main loop: for each time step
14:13 Who moves first, the environment or the agent?
14:49 The generative process transitions to a new state
16:00 Computing the free energy gradient with respect to the belief
16:38 Why the gradient matters more than the absolute value
17:35 The flow estimate and the update rule
17:47 Perception targets beliefs about latent states, not observations
18:21 Colour in peripheral vision: perception as unconscious inference
19:22 Correction: flow and state transition functions, not transition matrices
20:38 Stochastic differential equations and the partially observable setting
22:07 How a discrete transition matrix relates to the continuous case
23:21 Comparison with the OODA and TOTE loops
25:00 Algorithm 9: the multivariate, generalized-coordinates version
26:18 Nested loops and what M means
26:52 Embedding orders: uppercase M and lowercase m
27:50 An erratum in the loop index
28:36 Precision matrices in the multivariate case
30:25 Embedding order is just the element of the generalized coordinate
31:27 Appendix B and the indexing conventions
32:16 Zooming out: univariate to multivariate is scalability
33:30 Generalized coordinates vs a rolling average or spline
34:24 Taylor approximation and anticipation in continuous time
35:26 Expanding observations vs expanding the latent state
36:35 Q&A: reading the policy transition matrices
38:32 Attractors in Chapter 6
39:06 Why the generative process follows Hooke's law
40:08 Lorenz attractors and chaotic systems
41:05 Why attractors matter for what you can model
42:44 Physics-based models, energy minima and flow landscapes
43:39 Q&A: mapping equation 6.2 onto 6.3
45:25 Q&A: page 132, equation 6.9 and the noise parameter
46:25 Q&A: reproducing Figure 6.4
47:14 Reading the figure: true states, beliefs, prediction errors, free energy
48:41 Q&A: is the trajectory through belief space unique?
49:34 Non-identifiability, model comparison, AIC and BIC
51:35 Occam's razor and metabolic cost: the store and Venus
52:56 The frame outside the frame: what modellers leave out
54:05 What is specific to active inference, and what is just modelling
55:14 Q&A: sigma vs omega, agent noise vs environment noise
56:38 Correlated noise, Ito vs Stratonovich, and why it rarely matters in practice
59:34 Natural language readings of equations
1:00:23 Wrap-up/_dZHp_UGhVZE/Meetings_su3ToqMp

------

Active Inference Institute information:
Website: https://activeinference.institute/
X: https://x.com/InferenceActive
Discord: https://discord.activeinference.institute/
Donate: https://donate.activeinference.institute/
YouTube: youtube.com/c/ActiveInference
Active Inference Livestreams: https://video.activeinference.institute/
Fundamentals of Active Inference (Chapter 6, Session 29) July 21, 2026Michael Lennon: “Seeing the System Together: Active Inference for Networks That Learn Together”Bruce Tisler: “The Minimal Necessary Basis for Primitive State-of-Affairs Inquiry”ActInf GuestStream 113.1 ~ Bayesian Mechanics of Economic Choice (Ernesto Moya-Albor et al.)ActInf GuestStream 128.1: Active Inference as the Test-Time Scaling Law for Physical AI AgentsActInf GuestStream 120.1 ~ Temporal Cognition and the Emergent Properties of ConsciousnessFundamentals of Active Inference (Chapter 10, Session 52) October 9, 2026Viraj Deshwal: “A derived self”Fundamentals of Active Inference (Chapter 11, Session 55) October 27, 2026Fundamentals of Active Inference (Chapter 10, Session 49) September 29, 2026Tran Hoang Linh: “Dynamic Action-Space Adaptation via Contrastive Transition Adaptation”Antonio Lucas-Alba & Héctor M. Manrique: “Traffic Signs, Situated Cognition and Active Inference”
Active Inference Institute |

Fundamentals of Active Inference (Chapter 6, Session 29) July 21, 2026

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER