Textbook Group meeting for Parr, Pezzulo, Friston 2022 .

All right, welcome. Thanks everyone for joining. This is the final recorded session on the first half of Cohort 4. It's August 22nd, 2023, and we are going to discuss Chapter 5. So before we look at what we might have ahead of us from questions or anything else, does anyone want to make any general comments on 5? I'm just, anyone can give a thought. But just since we're on Chapter 5 in the neurobiology, we talked, I think, a little bit in the Chapter 10 closing discussion for Cohort 3 about some of the topics like what is empirical evidence or what does any experimental setting have to do with free energy principle active inference. And we reflect a little bit on this recent paper that, like, with Isomura et al, that, like, broke through kind of another level, not necessarily even from a epistemic perspective. Oh, I guess in terms of people's awareness, it was epistemic in that sense. Then I was thinking about this was in 2018. I mean, it's the same title. I know that the work has proceeded a lot since then, so that's not to say they rehashed the same. But in some ways, because there is a difference between map and territory, between, like, models of X and X, in a way, there's, like, this eternal gap that just continues to swallow up experimental results, much to the chagrin of people who think that just one more experiment is going to bridge or map the gap. And that's kind of, like, what some of us are told as graduate students. A lot of graduate school has to do with, like, creating a data set, and people get into the zone of thinking that that one data set is going to be addressing a question, even a narrow one. And of course, like, if it's not helpful, then let's not waste our resources to do that. But at the same time, people were saying that the relationship between gene expression and behavior was resolved in the 90s. And it's the same language that comes out today. And then even in this kind of fractal free energy way. It's like, what were we looking for in the Petri dish again? Oli? Yeah, that was exactly what I was going to say. Because, in fact, I would love to see in newer editions of the textbook to see some of the new directions in what some people call radical biology to be incorporated much more than we see currently in the textbook. Because, actually, the works of scientists such as Michael Levin and even before that, I mean, Varela, Mutana, Mutana, and Kapra, they show some radically new approaches to see biology and even life itself through, I mean, the conceptions of autopilot and allopoiesis and other related terms. But I believe that closer incorporation of these concepts into the textbook can help to see the bigger picture, which connects Bayesian mechanics as an overarching theory of every, quote-unquote, thing to active inference. Because the way it stands now, and in particular, the way Chapter 5 stands now, there seems quite a big gap between what Bayesian mechanics represents as a theory of everything and its neurobiological representation or, I don't know, neurobiological evidence or whatever. But if there was, I mean, more material on these new directions of radical biology, I believe these gaps can begin to be filled and the bigger picture can be, can finally emerge as a kind of continuous picture with, without any significant and possibly confusing gaps in between. Okay. Anyone else want to give a thought on that? Just one speculation here. The map territory, theory practice, the sort of dualistic dialectic is well-known. It's about questions like, what does math have to do with biology? And what does statistics have to do with something like a system of interest? So that's kind of the classic. We're dealing with another theoretical distinction. Within the domain of theory, we're making a distinction, not a homogenization, but a distinction between the more theoretical physics axiomatic view and the lens or tradition or method or approach or process theory described by active inference. That…