Textbook Group meeting for Parr, Pezzulo, Friston 2022 .

all right welcome back cohort 7 we're in our second onboarding SL chapter one slash Etc and Andrew go for it however you want to begin yeah uh great so I know that today we're covering not just chapter one but just a more General like introduction to to the group and active inference um sort of meeting uh I will kind of quick quickly give like a rundown of this first chapter it's not very long there's not too much to say but um that said I thought it'd be useful just to kind of kind of wrangle together all these different themes here um that we're going to be seeing in chapter one but also going forward for the rest of the textbook so in chapter one we're very quickly introduced to this general idea that a we have uh agents which could be seen as living biological organisms or other kinds of self-organizing systems um but they conceptualized as agents and so agents take in sensory observations and elicit actions to either realize their desired outcomes or goals and Andor to make sense of their world by gaining information about it this leads to the development of adaptive behaviors or strategies that agents use for navigating their environment uh this process plays out in a recursive action perception cycle where observations and form agents decision making regarding the next action they should take uh we also learn that this this whole process occurs at various levels of complexity and differing time scales ranging from the simple and rigid moment moment by moment strategy I of nutrient seeking bacteria up to the more complex and flexible strategies of humans engaging in planning to realize their goals in the future active inference occurs at different physical scales as well as well from microbiology to Broad cultural and development developmental learning in collectives or human societies uh one way to frame active inferences approach uh for example is through the neats versus scruffies discussion in this chapter uh while so-called scruffies lay emphasis on the idio sequencies and specificity of various phenomena of interests for example different biological adaptations neuronal Dynamics or cognitive mechanisms uh each deserving of their own separate bottomup research from empirical data the so-called Meats aspire to a set of first principles for explaining and relating these diverse phenomena in a more top- down fashion while still engaging in empirical analysis for validation and so active inference also identifies with needs as it lays the normative framework for understanding the brain and mind and their biological and cognitive implications alongside model design and testing for empirical analysis um we're given a kind of General schematic of the rest of the book uh it's separated into two parts s the first half will focus on Theory the second half on practice so I'm just going to speak about the rest of the first half um the first half basically presents active inference as a normal normative framework with its distinct Concepts and theories um we're shown a kind of conceptual road map at how to arrive at active inference uh in you know your respective Learning Journey or however you would like to view that um and there are kind of two roads to reach active inference there's a low road starting with bases theorem uh which basically supplies the mechanics of active inference uh in its relation to what would be called the basian brain hypothesis uh it's basically a way of viewing uh processes or computations in the brain is playing out Thea basian principles where there's prior information uh new observations come in there's some kind of likelihood that relates uh prior information with new information and then uh produces a kind of posterior uh result you know a prediction about what's going on in terms of uh inference right inferring what is going on based on what you thought before as well as what new information you've received and then on the other hand the high road in chapter three uh coming much later begins…