Classifying neural responses based on attention and working memory
|Classifying neural responses based on attention and working memory
|Cassia Low Manting <email@example.com>
|2023-08-17 – 2024-03-01
Human perception, attention, actions, and memories are all represented as neuronal
activity patterns in our brains. These activity patterns demonstrate an exceptional
complexity in how they are synchronized and distributed across the cortex. When studying cognitive function in conjunction with complex information, such as how humans perceive, memorize and selectively attend to individual sources in a complex information mix (such as listening to a voice among multiple voices, or viewing a face in a crowd), things become even more complicated. How does one identify and isolate the aspect of brain activity belonging to an individual information source within all this complexity?
In this project, we utilize and optimize a powerful combination of stimulation methods (so-called frequency tagging), neuroimaging methods (magnetoencephalography,
MEG) and novel analysis methods that have proven useful and rewarding in our recent research. Using these methods, we here aim to characterize the specific neural representations of individual information sources within a complex information mix, with the purpose of elucidating how cognitive processes such as selective attention and working memory modulates such representations. We will be implementing machine learning methodologies, especially but not limiting to classifiers, to distinguish between attentional states and learning conditions.
The project is an innovative and important step towards an improved and nuanced
understanding of the neural representation of sensory information, as well as of the
neural underpinnings of human selective attention in complex naturalistic auditory (e.g.
music) and visual (e.g. crowds) contexts.