Research

Our research is cross-disciplinary which combines engineering, physics, and computer science with neurosciences. This has enabled us to go beyond the traditional boundaries in order to understand how the brain implements cognition. Our lab research is on:

• Mathematical modelling to characterise how different part of the brain work together to process information and how this processing is disrupted with different brain pathologies;
• Develop neuroscience-inspired artificial-intelligence schemes (e.g. active inference) to understand how brain performs reasoning, learning, and planning;
• Use of classical psychedelics in combination with brain modelling to understand neural mechanisms underlying altered states of consciousness.

Areas of research/projects:

Computational neuroscience

We develop advanced computational and theoretical tools to understand how the brain is organised. Our research is geared towards developing new dynamic causal models (DCM) that can explain how the brain’s measured data is caused. We use these models to integrate multi-modal empirical measurements from functional, structural and diffusion magnetic resonance imaging (MRI).  In this stream of work we are:

• Developing multi-scale models of brain function
• Validating models using multi-modal brain imaging
• Using these models to gain mechanistic insights of brain’s functionally organisation

_{Effective connectivity within and between each network. (A) Effective connectivity matrix of the 15 brain regions. The 3 networks, core default network (cDN), salience network (SN) and dorsal attention network (DAN,) are highlighted using black lines. (B) The nodes and effective connections within and between each network have been mapped onto cortical surfaces. (C) A schematic summarizing effective connectivity between each network.}

Neural mechanisms of psychedelics

We use magnetic resonance imaging (MRI) to investigate the neural mechanism that underlie psychedelic-induced altered states of consciousness. Our research aims to answer the following important questions:

• What are the neural basis of subjective effects of psychedelics?
• How do psychedelics act in the brain?
• What are their phamacology and neurobiology?

_{This schematic describes that understanding neural basis of psychedelics-induced altered states of consciousness will require synergy among multiple research areas from brain imaging to mathematical modelling to context and setting.}

Computational neurodegeneration

The main problem that we tackle is to understand and predict how pathogenic protein effects on brain cells and how local tissue circuits translate to the large-scale network changes that produce clinical syndromes. The main questions that drive our research are:

• Why certain brain regions are targeted early and selectively?
• How the disease progresses to engender a network-based dysconnection?
• How to identify reliable neuronal markers of disease progression?

_{Effects of major disease factors associated with semantic dementia on effective connectivity of the semantic appraisal network. The left panels show brain cartoons representing connection changes in the right (R) and left (L) cerebral hemispheres associated with pathogenic protein deposition}

Artificial and biological intelligence

Neuroscience and artificial intelligence (AI) are burgeoning areas of research with a long history of inspiring each other. We are interested in the direction where we can use insights from neuroscience to help build intelligent machines that can excel in highly dynamic environments – just as biological brains can and current AI-based systems cannot. Recent work on active inference is a promising direction (a corollary of the free energy principle) which is an approach to understanding behaviour that rests upon the idea that the brain uses an internal generative model to predict incoming sensory data. We are developing this theoretical framework that can lead us towards achieving (near) biological intelligence in machines.

_{Blankets of blankets. This schematic illustrates the recursive procedure by which successively coarser scale (and slower) dynamics arise from subordinate levels in a complex system of interacting elements. Here we use renormalisation group theory for scale-invariant transitions between multiple scales or levels.}