Show simple item record

dc.contributor.advisorOjemann, Jeffrey Gen_US
dc.contributor.authorWander, Jeremiahen_US
dc.date.accessioned2015-05-11T20:02:51Z
dc.date.available2015-05-11T20:02:51Z
dc.date.submitted2015en_US
dc.identifier.otherWander_washington_0250E_14129.pdfen_US
dc.identifier.urihttp://hdl.handle.net/1773/33099
dc.descriptionThesis (Ph.D.)--University of Washington, 2015en_US
dc.description.abstractBrain-computer interface (BCI) technologies can potentially be used restore function in patients with severe motor disorders, however, BCI devices currently do not perform well enough to warrant the risk or expense relative to other treatments. This is due in part to limitations of current BCI architectures, which utilize signals from a relatively small portion of the cortex and exclusively rely on a fixed mapping between neural activity and the output device. In contrast, when executing native motor function, the nervous system invokes a sophisticated series of bottom-up and top-down modulators that dynamically change the relationship between cortical function and motor output, based on an individual's capability, task demands, attentional focus and numerous other factors. Characterization of the neural correlates of these higher-order cognitive facets of BCI use is a critical first step in the development of such systems. The work below focuses on identifying the distributed neural correlates of BCI skill acquisition and goal-oriented task execution, and leveraging these signals to improve BCI performance. We identified multiple cortical regions that become very active during novice BCI use, and are subsequently less active in the experienced user. Activity changes within these regions suggest distributed cortical processing, but could also be explained by nonspecific co-activation, so it was then necessary to show that these regions also interact in a meaningful way. To this end, we demonstrated that during BCI use there are high-frequency amplitude-amplitude interactions taking place on local spatial scales and non-linear low-frequency to high-frequency phase-phase interactions covering larger cortical distances. Lastly, we characterized neural correlates of a user's intended action immediately before and during BCI use, and trained a machine learning-based system to identify these activity patterns and leverage them in a hierarchical BCI. These findings are directly applicable to BCI design. Next-generation BCI architectures will include signals from multiple cortical regions to allow for dynamic device control strategies. Furthermore, by leveraging BCI as a platform for scientific inquiry, we have been able to develop our understanding of the networks involved in acquisition and execution of the BCI skill, and the neural mechanisms of interaction enabling communication across these networks. Understanding these relationships is at the core of understanding the tremendous adaptive capability of the nervous system, and successfully translating brain activity into action.en_US
dc.format.mimetypeapplication/pdfen_US
dc.language.isoen_USen_US
dc.rightsCopyright is held by the individual authors.en_US
dc.subjectBrain-computer interfacing; electrocorticography; intention; learning; machine learning; signal processingen_US
dc.subject.otherNeurosciencesen_US
dc.subject.otherbioengineeringen_US
dc.titleNeural correlates of learning and intent during human brain-computer interface useen_US
dc.typeThesisen_US
dc.embargo.termsOpen Accessen_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record