A Neuro-Inspired Computational Framework for AGI: Predictive Coding, Active Inference, and Free Energy Minimisation

This paper proposes that foundational principles from theoretical neuroscience - predictive coding, the Free Energy Principle (FEP), and variational inference - offer a biologically grounded framework for artificial general intelligence (AGI). These approaches characterise the brain as a hierarchical inference system that continuously updates beliefs and selects actions to minimise uncertainty and surprise. In contrast to conventional AI systems, which typically rely on static architectures and offline training, biological agents engage in active, generative inference within dynamic, uncertain environments. We argue that it is this inference-based architecture - not just its behavioural outputs - that underpins the adaptability, generalisation, and resilience of natural intelligence. We outline a neuro-inspired computational framework built on hierarchical generative models, scalable variational inference (e.g., Variational Laplace), and Active Inference. Finally, we contrast this approach with dominant deep learning paradigms and discuss its implications for building interpretable, adaptive, and autonomous machine intelligence.