Free Energy, Attentional Narrowing, and Cognitive Rigidity: A Hierarchical Predictive Processing Approach

Hello, welcome everyone. Hi. It is... Greetings. It's May 23rd, 2025, and we're in Active Inference Guest Stream 109.1 with Alexander Hemming presenting some shared work on emotional attention and predictive processing. So, Alexander, thank you for joining. Looking forward to your presentation, and anyone watching live, please feel free to write questions in the live chat during. So, thank you, Alex. Okay, sweet. So, emotional attention and predictive processing. For the actual presentation, I'll go through an introduction, a foundations of attentional narrowing, then a understanding of the neurology and a connection between the attentional narrowing and the predictive processing. I'll run through some computational models, so it'll give a more holistic understanding of this, and then we'll go through a discussion, which will include the applications and future directions. So, this is the team. At the moment, I've got me. There's a co-author for a book I'm writing as well called Dylan, and then there's Kirsten, who is a psychology researcher at the University of Birmingham. So, this is a project that we're working on. I'm going to briefly talk about this. This is a book that me and Dylan are working on. We're using predictive processing and active inference to understand the basis for how ideologies are formed. Kind of from a foundational way, building up to pretty much all the differences in politics, but it's difficult to go into at this point. If anyone has any questions about this, happy to answer as well. So, introducing the idea. Here, we have the whole idea, which is, in a sense, when you have a prediction and the prediction error goes wrong about something which is very emotional, then you tend to have an emotional salience of attentional narrowing. So, the threat uncertainty creates a predictive error minimization process, but that also creates a neuromodulation of a temperature parameter, which then creates a narrowing effect so you consider fewer alternatives. The thesis aims are to kind of create more of a unified understanding of the brain by adding together more of an understanding of what the actual parameters in the active inference equations link to. It's connecting it between a lot of different disciplines to try and understand more of how this works. Affect-based attention is one of the most prime parts of this. It's how does emotionally salient information go through a Bayesian distribution to try and make a different precision weighting compared to more calculated predictions, which is the normal direction of predictive processing. The research gap is predominantly about the lack of a link between processes such as dominant response facilitation and attentional narrowing and active inference. And so using those two together, it aims at addressing things which have not really been talked about, but also connecting a lot of this together. I'm running through quite a lot because we have quite a few slides in the time, but the foundation. So it starts originally in the 1908 theory of the Juerk-Stoldsen inverse U-graph, which many people might know from A-level psychology. Well, that's a UK thing, but yeah. And the idea is that if you have a certain amount of pressure, then your most reinforced responses will be facilitated. So let's say you're playing piano in front of an audience and you've been practicing for a very long time. You'll actually improve your performance of playing the piano until the stress gets so high that you choke and you do even worse than you did before. This went on to be understood by Easterbrook as not just all arousal creates that. It's not excitement which creates this attentional narrowing process, but it's actually a negative effect for it. So it's the uncertainty of some form of threat will then create a form of pressure. This form of pressure then facilitates a lack of consideration to alternatives. It's kind of automatic when it goes into the dominant response facilitation, but…