INI Colloquium

29.04.2026

A Scale-Free, First-Principles Account of Brain Function: The Free Energy Principle Meets Attractor Networks

Abstract:
In his talk, Prof. Spisak provides an introduction to the Free Energy Principle (FEP) as a unifying theoretical framework for understanding brain function, followed by a presentation of how - and what kind of - attractor neural networks (ANNs) emerge from the FEP, based on (Spisak & Friston, 2026). The FEP-ANN framework integrates variational inference with attractor dynamics and introduces the idea of "self-orthogonalization”: a consequence of predictive coding-like plasticity that enables the formation of structured, non-interfering neural representations. Framed as a computational model, this approach offers a mechanistic account of how neural systems may implement perception, inference, and decision-making. Particular emphasis will be placed on the neuroscientific implications, as well as on first empirical results that support the model’s predictions and illustrate its potential for explaining real neural data. The framework also has potential relevance for artificial intelligence, particularly in the context of continual learning and the development of adaptive, agentic systems.

Speaker:
Prof. Dr. Tamas Spisak (UK Essen)

Host:
Dr. Sandhiya Vijayabaskaran 

Venue:
NB 3/57 or via Zoom: https://ruhr-uni-bochum.zoom-x.de/j/68067472844?pwd=R3gxqnDEw39ULvjKSrJ0eaYLeUVRiv.1

Starting time:
12:00 PM