S-Space College of Humanities (인문대학) Program in Cognitive Science (협동과정-인지과학전공) Theses (Ph.D. / Sc.D._협동과정-인지과학전공)
Large-scale functional brain network for attentional regulation of encoding: Oscillatory interactions based on bipartite graph filtration
부호화에 대한 주의 조절의 뇌 기능적 신경망: 양분 그래프 필트레이션 기반의 오실레이션 관계 연구
- 인문대학 협동과정 인지과학전공
- Issue Date
- 서울대학교 대학원
- functional brain network ; alpha oscillation ; gamma oscillation ; magnetoencephalography (MEG) ; memory ; attention ; graph filtration
- 학위논문 (박사)-- 서울대학교 대학원 : 협동과정 인지과학전공, 2015. 2. 이동수.
- To encode relevant information and also suppress irrelevant one is suggested to be important mechanism for efficient memory functioning in the brain. The neural correlates of such memory regulation has been suggested as intra-regional interactions between alpha and gamma cross-frequency power. The current study expanded the cross-frequency interactions into a large-scale regional network for memory regulation. I analyzed the data wherein twenty three healthy subjects were instructed to remember (Remember) or not to remember (No-Remember) the following picture item by a cue during magnetoencephalography (MEG) recording. In this setting, network was modeled by correlation between regions for alpha power during cue presentation and regions for gamma power during item presentation, yielding a bipartite graph. Then, graph filtration method was applied to bipartite graph, wherein the changes of connected components between two sets of nodes (i.e., region) were quantified into two invariant measures: a barcode and single linkage distance. This procedure was called bipartite graph filtration in this study. As a result, as for the global connectivity reflected by the barcode, task condition was significantly different for gamma power during item presentation. In addition, subjects with steeper barcode in Remember than in No-Remember condition showed higher compliance to task instruction. Also, for Remember condition, subjects with fast merging pattern in alpha power during cue presentation than the pattern in gamma power during item presentation showed better memory performance. These findings suggested that encoding regulation is achieved by large-scale regional interactions, and the cognitive function is captured by distinctive pattern of network. As for the local connectivity reflected by single linkage distance, the connectivity between the left dorsolateral superior frontal gyrus for alpha power and the left insula for gamma power was significantly closer in Remember than in No-Remember condition. The superior frontal gyrus was included in dorsal attention network, suggesting dorsal attentional regulation of later long-term memory in the insular cortex. In conclusion, encoding regulation was investigated by large-scale brain interactions using bipartite graph filtration, showing the method allows to examine the time-lagged interactions between different frequency powers in an MEG network study for the first time.