Textbook Group meeting for Parr, Pezzulo, Friston 2022 .

all right welcome back everyone it is uh October 10th 23 and we're in our first discussion on chapter two so before we jump into any questions or anything what is anyone's just takeaway from chapter 2 or what were they curious about or what something that kind of remains with them after reading feel free to just raise your hand hand or just go for it yeah hi it's Pablo here um in the page 35 action has a pic value and pragmatic uh value um brings Rewards so that for me for what I'm building is super important so yeah I I have also some other notes but and a question over there we can cover later so yeah for me this is like my pick on this awesome yes the idea that we don't have to coers epistemic value or learning or Information Gain into utility we don't have to like convert it into utility in order for that policy to be selected key aspect any other favorite quotes or fun parts of chapter two or parts they woke up sweating about or I for me tying you tying this back to the surprise and and and um maybe elaborating on what encompasses policy I I'm I'm putting strategy in there but I don't know if that's correct specifically in the low road cool like policy enumerates the possibilities for Action over a given time Horizon but strategy isn't just enumerating branching path so then what is strategy great questions any other just random chapter two thoughts you know when it comes to this idea of planning as inference something that I've been just kind of thinking about a lot lately is that um ultimately when I do something and the best example is just trying to walk somewhere all I really do is to hold that conscious intention to be somewhere and then then I'm kind of out of the loop other than having uh you know processing um sensory evidence that I'm actually getting there and as long as long as that's happening you know I I can sort of see this this Loop in my in my brain where I'm going and just enjoying myself looking at things and just thinking all kinds of other thoughts and so on so forth and the going kind of happens on its own all all I really do is to have a conscious intention I don't even necessarily have to feel my body as it's moving if I get into this state almost like a you know like a walking meditation then it really feels like you know being in a self-driving car so it's kind of interesting that um you know I'm starting by understanding all of this stuff like having a having a generative model to notice these things in my own awareness I'm starting to notice these things in my own awareness and I'm starting to be you know not only uh have have these ideas confirmed mathematically but also exponentially so just a little bit weird but um really fun awesome cool and and also no amount of propositional intention to walk you can think as hard as you want I want to walk but then actually the intention that gets you out of the chair is something different so what does that what does that mean or what does that have to do with action yeah and but but but so you have to you have to sort of um distinguish different kinds of intention so like there's this there's the intention as in kind of like it would be nice if or like wishful thinking I I I wish I could do this um but what I mean by intention is a very specific mental state and that mental state is usually not available to you unless you start training in meditation where it becomes obvious that it exists so it's something that like you know if if you think about like think about this in from the perspective of training neural network for robot to do something so you know robot will have some kind of goal in a neural network getting to a place or whatever so it must it must be some kind of activation in the neural network that's that's going to be triggering a sequence of other activations right or a bunch of activations that are going to go through some kind of cycle in time that that sort of thing and normally that's an unconscious thing but I I don't I don't…