Social constraints (Discussion) — Course home

Hello, welcome everyone. It is October 25th, 2023. We're in active inference for the social sciences. And today we're in a discussion section. It is Avil and I. And we'll start off by just saying hello and kind of reflecting on where we're at with the course. Then we have the questions form that is on the course website. It's in the video description. Feel free to submit questions via this form. And then today we're going to look at the questions that have been submitted and not answered yet. And we'll just give some thoughts on them. Also look for more questions to arise in the live chat. So with that, what do you want to begin with, Avil? Feel free to give any kind of overview or statement. And then we'll jump to the specific questions that have been submitted before today and during today. The only statement here that I want to make is the last presentation I did. So prediction cause constraints to exist would be our wealth. Sorry of that, but I really, I don't have anything to say that is more specific right now. All right. Well, this should be pretty fun. Let's jump into some questions. Okay. What are the challenges and potential solutions in using active inference to model collective behaviors with various environmental constraints? So the challenges first is that in any model, you have to pick what you include and what you did not include. So when you talk about boundary conditions, when you talk about what's including the model, you have to make hypothesis. Like if I tell you model, I don't know a country that is under sanctions like North Korea, Russia. Okay. Okay. How do you model that? Do you make a set base for every possible products? Like every dimension accounts for a specific product or type of product? And you have zero if you have access to it. And you have zero if you don't have access to it and one issue access to it. Okay. But what if we have new things that happen that, that, that, that is not very useful if you're interested in, is in innovation or something. So we'll have to make specific what is that, uh, we'll try to specific questions that you want to ask to the model and using active inference has. And where it is relevant is that, uh, you can have, uh, uh, endogenous evolution of constraints using active inference. I think in a more meaningful sense than you'd have with a pure, let us say, heavy on or like non representational. Uh, framework or framework that is presentational, but where the space presentation is like hard coded. So you, you, there is a space of, uh, models, let us say of evolution of certain norms, where I think active inference would have, um, uh, an edge because if inference is built to represent normativity and intentionality, and it is not the case to my knowledge of other, like either dynamicist or cognitivist, um, frameworks for, uh, or modeling. Could you maybe restate and go into a little more detail? Was it non representational or indirect specification of what? Of the, um, beliefs of the system. So what's being directly specified, what's directly specified, and then what is being indirectly specified? Inactive inference. Inactive inference. Yeah. Inactive inference. Inactive inference. You have at least in a sense, uh, uh, correspondence between, uh, dynamical constraints, like information over where the system is and belief that the system holds that are akin to norms are by intentional states. So beliefs or intentions are things that are going on inside. And to my knowledge, there is no framework that makes, uh, an explicit. For more connection between the two, you'd have dynamic framework, uh, where you have stuff that happens. And there is no representation or intentional states. And you have things like a utility theory where you have to specify, uh, preemptively what are the intentional states of the agents, which are, um, not very relevant to represent, um, like the analogous evolution of, uh, social constraints within a collective system. So, um, um, situation…