Neural and cognitive insights into categorisation and interference resolution in visual working memory
Year: 2026
Type: dissertation
Abstract: Working memory is typically constrained by limited capacity, yet humans are able to flexibly prioritise relevant information and adapt to varying task demands. This thesis investigated how the brain manages interference under competing representational demands, focusing on whether information is supported by shared neural codes which generalise across contexts, or by independent codes which minimise conflict. Using EEG decoding, behavioural experiments, and a pre-registered EEG protocol, this thesis examines how category representations adapt when tasks shift from independent to overlapping stimulus structures; how cognitive load, response configuration, practice, and ADHD traits influence performance in visual working memory; and whether training or response effector separation could reduce interference between prioritised and unprioritised working memory representations. In Chapter 2 we explored multivariate decoding of EEG data which indicated that category-selective neural patterns generalised across tasks when stimulus sets were independent, but cross-task generalisation was reduced when overlapping stimuli were introduced (consistent with adaptive changes to representational structure in response to interference). In Chapter 3 we explored how interference between concurrently executed working memory tasks could be resolved through scheduling. The behavioural results indicated that increasing cognitive load from the interfering task impaired performance, which was not reduced by practice. Some ADHD traits correlated with interference-based impairment, suggesting some inefficiency in scheduling multiple tasks. Chapter 4, a pre-registered EEG study, was designed to test whether training could reduce this interference between conflicting prioritised and unprioritised working memory representations could be resolved by neural orthogonalisation, whilst also evaluating whether response effector separation contributes to representational separation. This thesis supports the view that cognitive limitations reflect working memory capacity constraints, but also interference occurring due to overlapping representations, and that the brain attempts to mitigate these limitations via flexible reformatting, prioritisation, and learning-dependent optimisation of neural correlates.
Author: Ajmeria, Uma
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7218195437
