Textbook Group meeting for Parr, Pezzulo, Friston 2022 .

all right welcome back slw welcome back to cohort 6 couple quick updates as we mentioned with octopus and others there are math discussion sessions they're Tuesdays at 12 UTC octopus session videos has the links to the videos so feel feel free um these should be awesome freewheeling and super epistemic sessions and as with the material on the book itself all kinds of questions from the most background and basic and confirmatory for one on through the most speculative and and farsighted for for others it's always a contribution to to write it down and as the math comes to play and The Meta math and the math art it's going to be a great journey together um today we're in our second pass on chapter six which is the recipe for active inference modeling and also a new or different feature in this interval of activity compared to previous weeks is the uh third week Rhythm so in the uh third week for cohort 6 going through the second half of the book will really focus on applying active inference Andrew has written a tutorial utilizing PDP with a focus on agent based modeling for the social sciences so that should be pretty cool and additionally as uh RX and fur development continues we'll be learning more and more about that um in cohort 7 we haven't quite determined what the third week activity will be in that uh slot so we'll find out uh discuss later today with that cohort and and see what's useful in the third week for now though let's uh jump over to chapter six we can start with any section of six any idea or page or figure in chapter six that anybody wants to bring up we could look at any of the prior questions and also as chapter six is about applying and designing active inference models it it can be any question that someone has so just raise your hand or write it in the chat or unmute and go for it okay I'll just give one brief catch up to get us to six to to get us uh there in just a minute and then again just type a question or raise your hand so part one of the book they outline as kind of the epistemic and the background and learning part two is heading kind of home stretch heading downhill heading back home in into the applying chapter 1 and chapter 10 are like book ends they're very similar chapters chapter 10 is a bit more broad ranging and longer whereas chapter one introduces the ideas whereas they're similar because they provide perspectives in chapter one this concept of the low road and the high road is introduced and that's Revisited in more detail in chapters two and three low road is like the how and the what it's the actual architectures that are built it's the way that base theorem applies to perception and cognition and action in a unified cognitive model and then the high road comes from the free energy principle physics of cognitive systems basian mechanics the imperative to persist to remeasure and so on chapter 2 and three get us to active inference and chapter four is where we actually see the active inference generative model itself which is what's going to be given the recipe for in 6 chapter V goes through the generative model describes some of its analytical properties and introduces us to the idea that there's a discrete time and a continuous time form chapter five message passing in neurobiology gives a quick review of one of the systems of interest that has received the most study to date in terms of active inference modeling which is like Maman and human neuroanatomy that's part one of the book part two of the book goes into applying active inference chapter six is very provocative and fun and it's a recipe for Designing active inference models and as we'll I hope unpack a little bit today it doesn't require some advanced level of programming or anything like that to to get going and actually with augmented coding with templates and some other tools there's a lot that can be done just through conversation with the domain expertise or familiarity and the active inference ontology chapter…