Ongoing research

Projects

We test how neural systems discover structure, separate similar experiences and build the predictive knowledge that makes inference possible.

An abstract illuminated brain formed from flowing connections

Ongoing project

RING

Representing Inferred Neural Geometry

RING asks whether the brain builds map-like representations of abstract concepts. Participants learn modular relationships among novel visual stimuli over several weeks before completing an fMRI session.

The project tests whether the hippocampus, medial prefrontal cortex and visual cortex encode inferred conceptual distances; whether the hippocampus represents that structure at different resolutions; and whether people generalise only after first memorising the relationships they were taught.

Methods: Longitudinal learning, fMRI, multivariate analysis
Core question: Can abstract knowledge take the form of a cognitive map?

Read the preregistration

A digital brain represented by a stream of connected points

Ongoing project

DISTAL

Distance in Stimulus Representations Tunes Abstract Learning

DISTAL investigates why some visual associations are learned faster than others. We hypothesise that learning depends on the distance between neural representations in visual cortex—not simply on differences between the images themselves.

One group views a diverse set of images during fMRI, while another learns discriminations that imply a hidden hierarchy. Relating neural distance to learning rates lets us test how similarity creates interference and whether structural learning is insulated from visual similarity.

Methods: fMRI, representational similarity, learning models
Core question: Does distance between neural codes determine how quickly we learn?

A small luminous brain emerging above electronic circuitry

Ongoing project

UNPICK

Uncovering Predictive Processes in Comprehension and Knowledge

UNPICK develops a mechanistic account of comprehension: generalisable knowledge may emerge when cognitive systems learn to predict their own future inputs and refine those predictions after errors.

The project combines a naturalistic narrative task with controlled learning of abstract relationships. Intensive fMRI tracks how knowledge develops within individuals and asks whether understanding stories and making novel inferences rely on shared predictive mechanisms.

Methods: Intensive fMRI, narrative comprehension, inferential learning
Core question: Is prediction the engine that builds understanding?

Ideas in progress

Want to discuss the work?

Start a conversation