Temporal Cognition and the Emergent Properties of Consciousness

Hello, welcome everyone. It is September 17th, 2025. We are in active inference gueststream 120.1 with Emanuel Adfa Olasupo who will be presenting modeling thought in time, a physics of temporal intelligence as the next frontier of AI. There will be a presentation followed by a discussion. So if you're watching live, please write any ideas or questions in the live chat and thank you for joining Emanuel to you for the presentation. >> Thank you Daniel. I appreciate the um the invitation. Um it's uh I'm incredibly humbled by um you know the opportunity to to give this talk. uh the title actually is not model in time. uh but this is the paper that I think uh connected you and I uh and so I just wanted to start the presentation by sort of uh clarifying or motivating the talk based upon this paper this preprint that I had published um and just to give a short summary of what the paper is about um you know the argument is that the uh current LLMs that um are being used right now both in the industry and in academia. They um have what we call a temporal blindness problem which means that they um do not represent time um in the lived sense right as as humans do. And so if the goal is to achieve conscious AI or AI that mimics human cognition um and humans ability to manipulate thoughts and emotions then uh this paper argues that time is um a foundational uh fe feature or uh that needs to be represented akin to how humans um experience time in the sense of thinking in time and so on. I'll get back to this towards the latter part of the presentation but I just wanted to motivate uh the the lecture uh based upon uh this paper the the the actual title of the talk is uh uh temporal intelligence and the emerging properties of consciousness um indeed you know I'm going to sort of to break this talk into three components the first is I'm going to idea of course defined and process it, right? And so this is a type of process in that um one would argue supports temporal cognition by structuring how information is organized over time and enabling the system to become more predictive and generative and generate expectations about the temporal sequence of impendent inputs. Right? And so that coarse course defined process and that's the first part that I'll talk about and I and I'll talk about the mathematics that perhaps governs that process both within biological systems but also artificial systems. The second part of of this of this talk is going to be uh on this idea of predictive remapping. Um and and this is an example of what I would think of as temporal cognition and action, right? Um um this is informed by um this idea of active sensing, right? Uh this idea that um cognitive systems must move in order to compute information in the world, right? And so I will also uh speak a little bit about this. Um and then the third part is the more um um speculative or perhaps more um has is it has the potential to be more uh paradigm shifted if um one or myself or groups or others anyone can achieve this this idea of uh the emergence of thought right so so how does your thought emerge how do they flow and synchronize with other people's thoughts and Can we achieve a mechanistic view of consciousness? Right? Because that I think is key into achieving AGI, right? Artificial general intelligence. And so th those three will be the so so this typeight structure would be kind of how I um uh go about giving this talk. And so you know I normally approach questions with physics and mathematics. All right? So I'm a theoretical physicist. My my my case is theoretical physics. But you know I I I normally use the two to try to understand the phenomenons that I study and I particularly look at complex systems particularly biological systems to explore that and so that is important to sort of highlight my biases right we all have biases in which we use to sort of inform how we think about the world right um so this idea of course defined I've I'm illustrated here in…