Textbook Group meeting for Parr, Pezzulo, Friston 2022 .

all right hello welcome everyone it's October 3rd 23 and we're in cohort 5 in our second discussion of chapter one we'll jump over to the uh cohort five questions it's not too late to add more just please feel free to copy in anything that you'd like us to explore but before we get to any of the questions is there just any kind of top of Mind piece about chapter one or some specific passage or some figure or something that anybody wants to share about chapter one before we jump into the question specifically okay okay let's go into the questions we we'll just go through them and again you can continue to add more or upvote them as as we continue question about modeling as action it seems that modeling is still abstract and based on observing and making effort to understand Dynamics Dimensions behaviors of given systems I.E modeling still seems theoretical possibly distinct from applying models and interacting with systems in the wild of course observing and documenting are also actions however the systems and behaviors being observed would be the primary actors expressing a rich complex range of acting in cohort 4 chapter 1 discussion arose a question about observing versus acting I guess it's well documented there inextricably intertwined what are anyone's thoughts on this for me um the perception action Loop seems to be more in line with you how we're making um policy decisions and not necessarily at this point about how we're mapping how we're moving um and so you I don't know if that's not you if if we can if we can kind of avoid conflating those two just to get the gist of where we're at in the in the book for me seems to be Cent thank you um Jesse I'm still on the phone for like the last day um yeah that that reminds me of like Maxwell's demon you know in a certain respect where even through making those choices you know it's it's an action so even the generating of the model and making that decision it's work that's also very um interesting point um it seems like the the questioner recognizes that um making a model is something that a person can do like a a researcher or scientist may do it so if you're thinking from the perspective of the modeler of course modeling is an action um applying a model that could be both abstracted and still an action I mean you could abstract away from the rat in the teamas like an example that will come back to or a frog jumping out of the hands which we'll come back to you can abstract that situation and model it and then you could apply that model but once you're in the map as opposed to the territory then there might be things that the territory could do like the rat could jump out of the maze but your model might not reflect that um and this is also a very rich question about observing versus acting we can revisit but action is not just overt gross bodily movement Act is conceived of very broadly uh in arbitrary spaces and so there's several ways that even in a situation that might be seen as a passive a passive observational setting there's several ways which action comes into play so visually icting is a a situation that's been modeled a lot in active inference so looking through a visual scene the is still an active policy selection ongoing about where the eyes move but that's still bodily in a sense but even more cognitively perception um is augmented by attention and attention in active inference is modeled by an internal policy decision so kind of like where to shine the internal Spotlight or how diffuse or how Focus to shine the attention Spotlight that's modeled as an internal or Co covert policy so there's definitely a lot of ways in which action and perception intertwine through through the loop and otherwise and um this is ironically or not a theoretical question about modeling being theoretical or not whereas once somebody puts something on paper and gets into the process of iterated modeling then it just becomes resolved in the actual what it is that they are doing and…