Author | Lightnews

Maxwell Ramstead

@mjdramstead.bsky.social

1.3K followers 150 following 120 posts

Cofounder @noumenal-labs.bsky.social. Honorary Fellow at the UCL Queen Square Institute of Neurology. Free energy principle, active inference, Bayesian mechanics, artificial intelligence, phenomenology

Posts Media Videos Starter Packs

Pinned

Maxwell Ramstead @mjdramstead.bsky.social · 1d

Beff shipped:
extropic.ai
A new era begins

Extropic | Home

Building thermodynamic computing hardware that is radically more energy efficient than GPUs.

extropic.ai

Maxwell Ramstead @mjdramstead.bsky.social · 1d

Beff shipped:
extropic.ai
A new era begins

Extropic | Home

Building thermodynamic computing hardware that is radically more energy efficient than GPUs.

extropic.ai

Reposted by Maxwell Ramstead

MadMind @madmind.bsky.social · Sep 11

Karl Friston in #mlst
Philosophy done right! So many references, obviously @drmichaellevin.bsky.social mentioned #academicsky #philosophy #neuroscience #strangeloop

youtu.be/PNYWi996Beg

Why Intelligence Can't Get Too Large (Karl Friston)

YouTube video by Machine Learning Street Talk

youtu.be

2 6

Reposted by Maxwell Ramstead

Kevin Mitchell @wiringthebrain.bsky.social · Sep 12

Super interesting, thought-provoking conversation between Mark Solms and Karl Friston open.spotify.com/episode/151a...

Karl Friston & Mark Solms: Is it Possible to Engineer Artificial Consciousness?

Spotify video

open.spotify.com

3 3 26

Reposted by Maxwell Ramstead

Hadi Vafaii @hadivafaii.bsky.social · Sep 17

What drives behavior in living organisms? And how can we design artificial agents that learn interactively?

📢 To address these, the Sensorimotor AI Journal Club is launching the "RL Debate Series"👇

w/ @elisennesh.bsky.social, @noreward4u.bsky.social, @tommasosalvatori.bsky.social

🧵[1/5]

🧠🤖🧠📈

2 10 36

Maxwell Ramstead @mjdramstead.bsky.social · Sep 15

Sorry to hear about your negative experience! My pleasure, don't hesitate to write me if you have any questions or want to discuss specific points :)

Maxwell Ramstead @mjdramstead.bsky.social · Sep 15

Yes! While Warren and myself have our disagreements, I like his work on PCT. IMO all these approaches are complementary and play together nicely. Along with friends (namely @adw.bsky.social who bravely led the project), we penned this integrative review. Hope it's of interest:
osf.io/preprints/ps...

OSF

osf.io

1 1

Maxwell Ramstead @mjdramstead.bsky.social · Sep 15

There’s a lot of cool work on multi-scale applications of the FEP. See, e.g.:
- www.sciencedirect.com/science/arti...
- www.sciencedirect.com/science/arti...
- arxiv.org/abs/1906.10184, especially the chapter on States, particles, and fluctuations

Variational ecology and the physics of sentient systems

This paper addresses the challenges faced by multiscale formulations of the variational (free energy) approach to dynamics that obtain for large-scale…

www.sciencedirect.com

1 1

Maxwell Ramstead @mjdramstead.bsky.social · Sep 15

3. Your point about top-down causation is key. IMO one of the most interesting aspects of multi-scale formulations of active inference is precisely how it handles multi-scale system dynamics, cashing out top-down influence in terms of constraints on system dynamics in a non-reductionist way

1 1

Maxwell Ramstead @mjdramstead.bsky.social · Sep 15

2. Not much work has been done on active inference and the neural code. The key departure from RL is that active inference uses an alternative objective function (the free energy functional), which you can read as an "ontological potential function" specifying object type (arxiv.org/abs/2502.21217)

Dynamic Markov Blanket Detection for Macroscopic Physics Discovery

The free energy principle (FEP), along with the associated constructs of Markov blankets and ontological potentials, have recently been presented as the core components of a generalized modeling metho...

arxiv.org

1 1

Maxwell Ramstead @mjdramstead.bsky.social · Sep 15

Great questions!
1. IMO active inference falls under the rubric of NeuroAI, (although I'd describe myself as a non-realist about these types of physics-inspired models, and as such I’d say the FEP isn’t a literal description of the brain, so it depends on the scope of NeuroAI, as your define it)

1 1

Maxwell Ramstead @mjdramstead.bsky.social · Sep 15

Love a good Feyerabendian sandbox. I'd argue that they're very closely related (and indeed, that the difference is often overblown by both proponents and critics), but they're also importantly distinct. We wrote a post on this that I hope you'll find interesting: www.noumenal.ai/post/filling...

Filling the gaps in active inference

Here we discuss key gaps in SOTA applications of active inference in AI - and how Noumenal Labs is working to fill them.

www.noumenal.ai

Reposted by Maxwell Ramstead

Guillaume Dumas @introspection.bsky.social · Aug 8

🤔 How can we study #consciousness between people, at the social level? 🧠✨ New #preprint co-led by Anne Monnier & Lena Adel: “Now is the Time: Operationalizing Generative Neurophenomenology through Interpersonal Methods” 🧵(1/3)

2 14 36

Reposted by Maxwell Ramstead

Takuya Isomura @takuyaisomura.bsky.social · Aug 11

Our new preprint on active inference in zebrafish, with Yuki Tanimoto, Makio Torigoe, Hitoshi Okamoto, and Hideaki Shimazaki
"Predicting individual learning trajectories in zebrafish via the free-energy principle"
doi.org/10.1101/2025...

Predicting individual learning trajectories in zebrafish via the free-energy principle

The free-energy principle has been proposed as a unified theory of brain function, and recent evidence from in vitro experiments supports its validity. However, its empirical application to in vivo ne...

doi.org

1 2

Maxwell Ramstead @mjdramstead.bsky.social · Jul 2

Currently, using active inference at scale involves trade-offs between explainability and the ability to learn models from data. Not using overparameterized models increases model explainability and auditability, but makes learning in high dimensional and volatile environments more challenging

1 1

Maxwell Ramstead @mjdramstead.bsky.social · Jul 2

It provides an alternative objective function that has useful properties, in particular enabling agents to balance the value of exploration and exploitation in policy selection. But IMO the differences between RL and active inference have been exaggerated a bit. See: www.noumenal.ai/post/filling...

Filling the gaps in active inference

Here we discuss key gaps in SOTA applications of active inference in AI - and how Noumenal Labs is working to fill them.

www.noumenal.ai

1 1

Reposted by Maxwell Ramstead

Jen @ladyjenpool.bsky.social · Jun 22

Computers used to scream every time they connected to the Internet. They knew. They tried to warn us. We did not listen.

57 3.7K 11K

Reposted by Maxwell Ramstead

Daniel Yon @danieljamesyon.bsky.social · Jun 13

Delighted to see ‘A Trick of the Mind’ reviewed in @theguardian.com as Book of the Day! 🧠 🍎

Also in the print edition tomorrow 🗞️

www.theguardian.com/books/2025/j...

2 17 60

Reposted by Maxwell Ramstead

arXiv q-bio.NC Neurons and Cognition @qbionc-bot.bsky.social · Jun 9

Luca M. Possati: Markov Blanket Density and Free Energy Minimization https://arxiv.org/abs/2506.05794 https://arxiv.org/pdf/2506.05794 https://arxiv.org/html/2506.05794

1 2 1

Reposted by Maxwell Ramstead

Hadi Vafaii @hadivafaii.bsky.social · May 19

Elegant theoretical derivations are exclusive to physics. Right?? Wrong!

In a new preprint, we:
✅ "Derive" a spiking recurrent network from variational principles
✅ Show it does amazing things like out-of-distribution generalization
👉[1/n]🧵

w/ co-lead Dekel Galor & PI @jcbyts.bsky.social

🧠🤖🧠📈

1 12 34

Reposted by Maxwell Ramstead

Fernando Rosas @frosas.bsky.social · Apr 23

Preprint time:
“Shannon invariants: A scalable approach to information decomposition”
arxiv.org/abs/2504.15779

Studying information in complex systems is challenging due to difficulties in defining multivariate metrics and ensuring their scalability. This framework addressed both challenges!

Shannon invariants: A scalable approach to information decomposition

Distributed systems, such as biological and artificial neural networks, process information via complex interactions engaging multiple subsystems, resulting in high-order patterns with distinct proper...

arxiv.org

1 8 31

Reposted by Maxwell Ramstead

Tiago Peixoto @tiago.skewed.de · Apr 5

Computational theory is about computers (i.e. "technology") in the same way that astronomy is about telescopes. Thinking that computation is not fundamentally important for biology because "a cell is not like a laptop" is to miss the forest for the trees. N/N

1 11 41

Reposted by Maxwell Ramstead

Ryan Smith @rssmith.bsky.social · Mar 27

Reinforcement Learning and Active Inference are two frameworks used in computational psychiatry, but these are rarely directly compared empirically. In this new article, we aimed to compare these in a more systematic manner by fitting each to multiple datasets: papers.ssrn.com/sol3/papers....

A Systematic Empirical Comparison of Active Inference and Reinforcement Learning Models in Accounting for Decision-Making Under Uncertainty

Reinforcement Learning (RL) and Active Inference (AInf) are related computational frameworks for modeling learning and choice under uncertainty. However, differ

papers.ssrn.com

2 14 38

Reposted by Maxwell Ramstead

John Scalzi @scalzi.com · Mar 20

It's not just that just about everything I've ever published is in this particular database and used without my permission. It's that everything I've ever published was used without my permission to develop such a shitty, flawed and fundamentally useless tool. I deserve compensation for THAT itself.

Damon Beres @damonberes.com · Mar 20

NEW: LibGen contains millions of pirated books and research papers, built over nearly two decades. From court documents, we know that Meta torrented a version of it to build its AI. Today, @theatlantic.com presents an analysis of the data set by @alexreisner.bsky.social. Search through it yourself:

The Unbelievable Scale of AI’s Pirated-Books Problem

Meta pirated millions of books to train its AI. Search through them here.

www.theatlantic.com

71 290 1.6K

Maxwell Ramstead @mjdramstead.bsky.social · Mar 18

Bluesky has become awesome and I am absolutely loving it.

1 11

Maxwell Ramstead @mjdramstead.bsky.social · Mar 17

Accordingly, emergent reuse, reassembly, and analogical reasoning must be key features in the design of machine intelligence — and open a path towards the development of collaborative, superintelligent AI systems 5/5

2 1