Monash Computational Neuroscience Laboratory

Our research focuses on understanding and engineering adaptive intelligence through the integration of neuroscience, active inference, generative modelling, and embodied AI. Broadly, we study how biological and artificial systems learn internal models of the world, use these models to guide behaviour under uncertainty, and continuously adapt through interaction with dynamic environments.

A major component of our work develops mechanistic generative models of brain function, combining Bayesian inference, dynamical systems, and large-scale neuroimaging to study cognition, psychiatric disorders, and altered states of consciousness. We are particularly interested in active inference as a unifying framework for perception, decision-making, learning, and self-organisation in both biological and artificial agents.

We also work on embodied and closed-loop AI systems, including the DishBrain platform developed with Cortical Labs, where living neuronal cultures interact with simulated environments in real time. This work investigates whether biological neural networks can learn predictive world models, exhibit adaptive behaviour, and provide insights into more energy-efficient and robust forms of intelligence.

In parallel, we develop multimodal foundation models for neuroscience that integrate brain imaging, structural connectivity, behavioural, and physiological data across scales. These models aim to provide interpretable and biologically grounded representations of neural dynamics while enabling downstream applications such as disease prediction, mechanistic inference, and adaptive AI systems.

Across these areas, our broader goal is to bridge neuroscience and AI to better understand intelligence, generalisation, robustness, and alignment in real-world settings.