Fundamentals of Active Inference (Chapter 7, Session 34) August 7, 2026 @ActiveInference
Fundamentals of Active Inference (Chapter 7, Session 34) August 7, 2026  @ActiveInference
Uploaded August 2026 | Updated November 2026, 2 days 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: the fourth and final session on Chapter 7
1:04 Chapter 7 as an addendum to Chapter 6
3:02 Looking ahead: Chapters 9 and 10 and discrete state spaces
4:02 Q&A: modelling mental states, valence and interoception
5:00 Figure 7.3: generative process vs generative model
5:58 Example 7.1: the Bayesian thermostat
8:58 Specifying the generative process
9:53 Euler discretisation of the ODE
10:52 The generative model: observation and state transition functions
12:47 Autonomous vs exogenous forces
15:46 The forward model linking action to observation
17:39 Live notebook walkthrough begins
19:39 Prediction error and the preferred state
20:40 The observation function: mapping depth to temperature
22:33 Setting the agent's preferred observation
24:32 Free energy as precision-weighted prediction error
25:34 Approximating the forward model by its sign
26:35 Computing prediction errors in code
28:30 The simulate function and its parameters
29:27 Perception only for the first five seconds
32:11 Running the simulation: reproducing Figure 6.7
32:47 Reading the plot: hidden state, action, belief, preference
34:32 Free energy across time: what action buys you
35:21 Changing the forward model: a disastrous run
36:49 Experiments with sensory precision (50, 100, 500)
38:45 Where the code will live
40:07 Figure 7.11: why action reduces free energy further
41:54 Surprisal depends on observations, not beliefs
42:36 The complexity and accuracy decomposition
43:08 Q&A: perception first, or action first?
44:56 The four-step agent loop
47:53 Q&A: the costs of action and why planning needs sophistication
49:33 Exteroceptive, interoceptive and proprioceptive observations
51:41 Does an agent need to read itself? Metabolic constraints
55:02 Homeostasis and what an engineering textbook leaves out
56:47 Q&A: applying active inference to physical AI systems
59:07 One-to-many and many-to-many mappings in the likelihood
1:00:45 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 7, Session 34) August 7, 2026ActInf GuestStream 114.1 ~ How do inner screens enable imaginative experience?Fundamentals of Active Inference (Chapter 3, Session 13) May 26, 2026Fundamentals of Active Inference (Chapter 7, Session 31) July 28, 2026ActInf ModelStream 020.1: Kenric Nelson: Learning extreme models with the Coupled Free EnergyStell: “Where the Theories Stop: What Free Energy and Integrated Information Actually Buy You”Patricia Craja: “Percepia: Perceptual Phenotyping from Conversation”Fundamentals of Active Inference (Chapter 10, Session 48) September 25, 2026Hyojun Choi: “Adaptive Prosociality Through Allostatic Regulation in Social Dynamics”Daniel Dragolich: “NUMEN: Bare metal inference”ActInf GuestStream 084.1 ~ The Nature of Habits and Agential Systems, Jesse GActInf GuestStream 121.1 ~ Conscious active inference (Wiest and Puniani)
Active Inference Institute |

Fundamentals of Active Inference (Chapter 7, Session 34) August 7, 2026

SHARE TO X SHARE TO REDDIT SHARE TO FACEBOOK WALLPAPER