The evolution of brain architectures for predictive coding and active inference

Hello and welcome to the Welcome everyone. It's ActInfLab livestream number 38.0, February 10, 2022. We're going to be discussing the paper, The Evolution of Brain Architectures for Predictive Coding and Active Inference. Welcome to the Active Inference Lab. We are a participatory online lab that is communicating, learning, and practicing applied active inference. You can find us at some of the links here on the slide. This is a recorded and archived livestream, so please provide us with feedback so we can improve on our work. All backgrounds and perspectives are welcome here, and we'll be following good video etiquette for livestreams. It's going to be a solo stream though. Go to activeinference.org if you want to learn more about how to participate or contribute or get involved with any ActInfLab project. And check out this the platform link to see past and upcoming livestreams. The page looks like this. So you can see events that haven't happened yet, like 39, 40. And then also you can look back and you can see who is participating and read the papers and all of that. So, check it out. Today in ActInfStream number 38, the goal is to learn and discuss this cool paper, The Evolution of Brain Architectures for Predictive Coding and Active Inference. The paper is by Giovanni Pizzullo, Thomas Parr, and Carl Fristen from December 2021. And just like all videos, it's just an introduction to some of the ideas. It's not a review or a final word. So go check out the paper to learn more. And there's going to be an overview with first names and claims, abstract, and roadmap. All right, I'm Daniel. I'm a researcher in California. The big question that this paper is getting at is, what is the evolutionary neurophysiological basis of cognition and how do complex cognitive phenotypes arise? So, how do things develop and evolve? How they think and how does that change over evolutionary time? And shown here are three images representing three scales of analysis. of looking at ant cognition. So, on the left is a representation of the synapse with the glia wrapped around it, and the molecules and some of the mechanisms, because changes in those mechanisms can influence cognition. Then, in the middle is a 3D representation of an ant brain with the different brain regions, like the central complex and the optic and the olfactory lobes. And this represents the level of regional or micro or meso-anatomical variation. And that definitely changes over evolutionary time, just like the synaptic level. And then there's this behavioral ecological level. And that's where the ants are engaging in collective behavior and stigmagy. And so, how does this all work? How does this all work in today's ants and how has it evolved? And then expand that to other species and other questions. The paper was published right at the end of 2021 in December in the Royal Society of Publishing. And just to go over the aims and claims of the paper, this is in the author's words. There's growing consensus that the brains of humans and other phylogenetically derived or advanced organisms operate in a predictive manner across perception, predictive coding, and action control, active inference. Yet, the ways in which our advanced predictive abilities may have arisen during evolution remain unclear. The goal of this article is to sketch an evolutionary history of brain architectures for predictive processing. A central tenet of our proposal is that although prediction is often characterized as a complex cognitive function, it is not a late evolutionary addition of advanced animals like us. Rather, in distinction to a late-stage cognitive argument like saying language is what makes us an advanced cognizer, or semantic language with certain types of syntax. Rather, our complex predictive abilities, e.g. planning and imagination, emerged gradually, e.g. via phyletic gradualism, smooth changes through evolutionary time, or punctuated equilibrium, sharp changes through evolutionary time, but…