Theme: Mechanisms

Project Description

How does the brain remove thoughts that are no longer relevant? Could difficulties with this process contribute to persistent negative thinking in depression? This PhD project will interrogate those questions using an innovative computational neuroscience approach.

Working memory provides a mental workspace that holds information to guide behaviour. While much is known about how the brain maintains this information, the mechanisms that remove it remain poorly understood. This PhD project will investigate three possible mechanisms: weakening the connections between a memory and its retrieval context (“unbinding”); replacing it with new or neutral information (“substitution”); and resetting the brain activity that maintains it (“attractor reset”).

The project will combine computational models of neural circuits with behavioural experiments and magnetoencephalography (MEG) in participants with and without depression. The models will generate predictions about how each mechanism changes memory representations over time. Experiments will test these predictions using neural decoding and brief visual impulses that “ping” the brain, revealing memory representations that may persist even when they produce no detectable ongoing activity. This approach will help distinguish information removal from retention in a hidden state.

The student will receive interdisciplinary training in computational modelling, neuroimaging, reproducible research and clinically informed experimental design, supported by supervisors with expertise in memory decoding, neural-circuit modelling and psychiatry. People with lived experience will help shape study materials, meaningful outcomes and interpretation.

The project aims to explain how working-memory removal operates and how it relates to depression and repetitive negative thinking. Its findings could identify behavioural and neural markers for future studies and inform interventions tailored to specific mechanisms. Open datasets, models and protocols will support replication and reuse.