Textbook Group meeting for Parr, Pezzulo, Friston 2022 .

Hello and welcome. We're in cohort two of the Parr et al textbook group. We are in our final discussion on chapter 10, the last of the topical discussions. So we'll see where it goes. Then next week will be the final meeting of this cohort, where we'll have unrecorded feedback, discuss some project ideas, and so on. So this will be the last recording in this series for cohort two. And please go to future textbook groups to fill out the feedback form and or share the link to register for future cohorts. That being said, let us turn to chapter 10 and then go wherever people are interested. So Michael, maybe you can restate what area you want to look at, and then let's see if it relates to anything in 10. Michael Schultz- Yeah, okay. So my question was related to priming effects. So priming is very much used paradigm in psychology, which assumes that the stimulus activates somehow a network of semantic related items or phonetic or somehow related items, which facilitates recognition of successive items and then according also action. And I was wondering how this is represented in active inference, because it's a kind of an action perception loop, and how the difference between more conscious or more higher level kinds of decisions are distinguished in this active inference framework. Whether we have a possibility to distinguish between these kinds of more priming action perception loops and more consciously activated action perceptions. Are there any papers? Is there any research that that somehow maps those different distinctions into the terminology that we have in active inference? Do you know? Michael Schultz- Yep. Ali, any thoughts? Ali Kishnicka Ali Kishnicka Nothing immediately comes to mind, but you go on, please. And then I'll think if I can add to it later. Okay. One, this one appears to at least address attentional priming in the context of motor behavior. I'll add the link into the notes section. But I think as we're coming to see, the framework is going to be minimally opinionated on complex, extremely heterogeneous cognitive phenomena like priming. You can construct a cognitive model that embodies that through any number of ways. Like you could imagine a nested perceptual model where the first, the higher level is what genre of movie we're watching, comedy or horror. And then the stimuli are the sounds. And then like hearing a laugh, then updates the belief that it's a comedy, not a horror. And then that makes certain stimuli more attended to in the future. But that wouldn't be like an opinionated active inference framework describing how priming is approached, but rather a way that you could use the kinds of techniques that we discussed in the textbook to write your own generative model that has some of the features that you'd like. And I'm not exactly sure how they've done it in this motor paper all the way from 2011. Okay. Ali, anything? Ali, anything? Yeah, I just wanted to add that. I'm not sure if it directly relates to this specific question or not, but at least I haven't seen much paper doing direct empirical experiments, trying to, I mean, trying to observe or confirm some of the ideas of active inference framework. In other words, most of the papers in the literature are mostly concerned with modeling the existing experimental data or empirical data. And, but there are some important works that's being done, not directly trying to prove or confirm the active inference framework, but they have done some research in other areas, but they've used active inference in the modeling of the phenomena being observed or studied. So in this case also, I vaguely remember seeing some related studies that's being done and ultimately modeled with active inference framework, but I'm not sure if there's anything specifically putting active inference claims as hypothesis and trying to confirm those hypothesis. So, as you said, active inference is not, cannot be falsified. So it's not really a hypothesis in this way, no?…