Edição completa
Conteúdo da newsletter
Esta página preserva o texto arquivado da newsletter, links públicos e mídia.
Hello all,
As always — complete the Measurement form to have your research, education, and work updates included in a future newsletter. Please include links to your work, and aim to make the first paragraph short & clear so that we can include it easily in newsletter.
Big thanks to the many of you — fellows, interns and others in the ecosystem — who are doing awesome work and submitting updates!
Updates from the Institute
- The 6th Applied Active Inference Symposium will be held November 12-13, 2026. The primary events will be online and free. Also, stay tuned for announcements about possible in-person events around the world. We will share participant registration information soon. For now we are soliciting Presenters (direct link to submit) and Sponsors (more information). Contact us with any ideas or questions. How can we make this the biggest and most impactful Applied Active Inference Symposium yet, and what would be of Pragmatic and Epistemic value for you?
- The Fundamentals of Active Inference textbook group continues to meet weekly. All details and registration form here, the playlist of all meeting recordings is here, our interactive learning document is here, and all open source code examples are here. The textbook is “Fundamentals of Active Inference: Principles, Algorithms, and Applications of the Free Energy Principle for Engineers” by Sanjeev V. Namjoshi (2026).
- We have a new public site live at
https://activeinference.institute/
- . This is hosted via Github pages, here is the underlying repo. Things will continue to update there; for now we’d appreciate your feedback on functionality, and sharing the site around to some fresh eyes who might not know about Active Inference or the Institute.
Updates from Research Fellows
- Jake Hooper gave a talk with PsychedelX, titled “Psychedelics as a Tool for Neuroaesthetics”, link here:
- . The role Active Inference takes in this talk is relatively minimal, but there are some clear connections in the deeper implications of the points made in it. Further empirical work will attempt to use this framework when interpreting results, with more technical attention to active inference going forward.
Updates from the Active Inference Ecosystem
- cpomdp project (Dan Corva) reports a new publication “State-Dependent Observation Noise Reintroduces Epistemic Value in Linear-Gaussian Active Inference” (arXiv:2607.20306). cpomdp is a JAX-native continuous-state active inference toolbox, the continuous-state sibling to pymdp. One of its routines compiles a specified agent down to a fixed linear-Gaussian filter wherever that reduction is valid. For models with state-dependent observation noise, R(x), it wouldn’t compile. Rather than patch around it, Corva chased the obstruction and found it was structural: no fixed linear-Gaussian filter can reproduce these agents at all. That became a typed IncompatibleLinearizationError in cpomdp v0.4, and now a preprint .
- Precious Adeniyi working on the PVF-U (Primitive Valence Foundation—Unified) project reports on a three-tiered architecture for multi-agent AI systems. The project addressed the Emptiness Asymptote—a failure mode where agents freeze in predictable environments without social partners.
- Artificial Sentience project (Cleber Gomes) reports a new blog post: Dissociating Perception and Reality: A Predictive Coding Framework for State Estimation, Synaptic Plasticity, and Emotional Valence Under Non-Optimal Generative Weights. This post explores the epistemological and mechanical boundaries between physical ground-truth reality and an agent’s internal subjective reconstruction.
- Daniel Friedman reports a new publication and codebase on “Active Inference Power Suite: Conditional Statistical Power under Controlled Generative Settings”. The suite combines fixed-horizon procedures, analytic references, dependence/calibration experiments, and a discrete-state active-inference agent to explore topics related to statistical power.
Some highlight livestreams from the last month included:
- GuestStream #067.2 with Andres Corrada on “Who Judges the Judges?”
- GuestStream #128.1 with Omar Hashash and Christo Kurisummoottil Thomas on
Active Inference as the Test-Time Scaling Law for Physical AI Agents
- GuestStream #129.1 with Valeriia Zaiats on TRIAD & ITCE Launch: The Thermodynamics of Ethics
- GuestStream #130.1 with Alexander Hemming, Dylan Grove, Jacob Hohwy, and Karl Friston on
Understanding Integrative Complexity Through Active Inference
- ModelStream #010.1 with Sharath Satihish on
- If you would like to help make audio-visual content with the Institute, there are many roles available, for example to help plan livestreams. See this video for more information.
This page gives a 1-page summary of the structure of the Institute — a helpful entry point for those who are new, or learning about our form and functions.
Here is the entry point for 2026, and ways to get involved with the Institute and Ecosystem this year. Donate to the Institute to support our impact and sustainability. Contact us with other philanthropic and grant ideas.
Get in touch if you have general or specific feedback, ideas, or questions for the Institute.
More to come! Stay tuned via the newsletter and Discord.
Officers, Alexandra & Daniel
👣👣
