Textbook Group meeting for Parr, Pezzulo, Friston 2022 .

hello everyone we're in our first discussion of chapter seven in cohort tree so let's head over to chapter seven and uh we will come to the questions but let's just start with anyone who wants to do you want to just give a general comment or any reflection or thought on seven or anything up to this point Ali and then anyone else uh yes so uh chapter seven is basically the application of uh the first half of chapter four in terms of uh modeling um discrete time situations uh and uh it it goes through a number of case studies uh in order to show how the matrices a b c and d can be constructed in various situations and the the challenges or uh rather the concerns uh we may have when constructing such matrices in order to have a viable discrete time model uh for each situation uh so yeah that's basically uh the premise of chapter seven awesome thank you totally agree just like chapters two and three with low row to the high road we're kind of like a pair chapters seven and eight are kind of like a pair because seven is going to go into discrete time modeling chapter 8 is on continuous time and hybrid modeling does anyone else want to make any just overall reflection or a comment on chapter seven all right I'm checking if there's any okay does anyone have any question at all otherwise we'll turn first to these questions start with the written questions we have and then just approach it from there does anyone have anything else that they want to consider first though okay okay I'm putting in the chat the um oop that's the um social sciences link but here's the chapter 7 Link in the chat and so you can follow along there um also please upvote the questions that are interesting to you because the the questions that have a bunch of upvoted points we're going to make the short videos for um okay so we'll just go to the I'm going to unvote all the questions from my you can still vote for them but I'm just going to unvote them so that once we get to them all all vote for them okay all right first question here's chapter seven chapter seven begins with the quotation what I cannot create I do not understand Richard feudman what do you think this means in the context of active inference specifically I can't remember exactly who added this in a in a previous uh cohort but um they they gave a mild and a stronger answer the mild answer is you have to learn by doing you actually have to build the generative model to understand active inference um and then the stronger answer it goes even beyond that and it's not just that you have to build the generative model you actually have to have something that escapes your hands and does something interesting to really test the method and understand it foreign anyone else want to add a a comment on this okay kind of a warm-up question but anyone else want to add any anything on this what I cannot create I do not understand otherwise we'll come into the more specifics of the chapter seven now okay so in that case I'm just gonna um go with the most upvoted questions all right this is on page 130 the textbook they wrote this Choice which we're going to look to the textbook in a second speaks to the classical exploration exploitation dilemma and psychology a dilemma that is resolved under active inference what do you think about this okay um so the um setting that's being considered is decision making under uncertain payoff or reward structure so you think there's two slot machines one of them pays out fifty percent of the time you don't know how much the other one pays out so you could continue to get 50 percent that would be exploit or you could explore and it might be better than 50 it might be worse than fifty percent and so that is sometimes called the exploration exploitation dilemma um does anyone want to give a comment or or a related question on how is active inference going to be approaching explore exploit Michael yeah I just um uh wanted to both uh affirm that I found the mathematical…