Oncescu, Mihai Andrei (2026) Investigating Uncertainty in EEG-Based Brain-Computer Interfaces During Grasp Observation and Execution. Research Project 2 (major thesis), Behavioural and Cognitive Neurosciences.
|
Text
Full-Thesis-Final.pdf Download (4MB) | Preview |
Abstract
Electroencephalography-based brain-computer interfaces aim to decode motor intentions from neural activity; however, distinguishing between complex hand grasps remains challenging due to overlapping neural representations and the low signal-to-noise ratio of EEG. This study investigated uncertainty in EEG-based grasp decoding from both neurophysiological and machine learning perspectives. Eighteen participants observed visually noisy videos depicting one of three grasp types (pincer, precision, or power), executed the grasp, and reported their subjective confidence. EEG recordings were analysed to examine task performance, the centroparietal positivity (CPP), movement-related cortical potentials (MRCPs), and sensorimotor oscillations. Grasp type was classified using Linear Discriminant Analysis (LDA), Minimum Distance to Riemannian Mean (MDRM), and EEGNet. Higher subjective confidence was associated with greater behavioural accuracy, faster movement initiation, and larger CPP amplitudes. CPP and MRCP amplitudes also differed between grasp types, whereas mu and beta oscillatory activity did not. All three classifiers decoded the grasp type above chance, with EEGNet achieving the highest performance. However, classifier-derived confidence measures were not significantly associated with participants' subjective confidence.
| Item Type: | Thesis (Research Project 2 (major thesis)) |
|---|---|
| Supervisor name: | Sburlea, A.I. |
| Degree programme: | Behavioural and Cognitive Neurosciences |
| Thesis type: | Research Project 2 (major thesis) |
| Language: | English |
| Date Deposited: | 26 Aug 2026 14:37 |
| Last Modified: | 26 Aug 2026 14:37 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/37515 |
Actions (login required)
 |
View Item |