Organization of Networks Underlying Working Memory. (16th November 2020)

• Record Type: Journal Article
• Authors: Darrow, David P; Herman, Alexander; Saiote, Catarina; Park, Michael C
• Is Part Of: Neurosurgery. Volume 67(2010)Supplement 1
• Publication Date: 2020-11-16
• Subjects: Nervous system -- Surgery -- Periodicals; 617.48005
• DOI: 10.1093/neuros/nyaa447_665
• Languages: English
• ISSNs: 0148-396X

Abstract

Abstract: INTRODUCTION: Working memory has been studied as a paradigm using EEG and fMRI. Frontal and parietal cortices have been identified as playing the most significant role underlying the neural basis for working memory centered on the prefrontal cortex. No intracranial studies have reported on the spatial or temporal features associated with the N-back, a common working memory task. We sought to characterize the electrocortical event-related potentials and spectral activity during the N-back task during invasive recordings. METHODS: Four patients with pharmacoresistant epilepsy undergoing surgery for seizure localization completed N-back tasks with randomized degrees of load (0-, 1-, 2-, and 3-back) in the epilepsy monitoring unit in a randomized order. Event-related potentials (ERP) and spectral analysis were done in a block by block basis. RESULTS: ERP analysis (permutation testing, cluster-based multiple comparison correction, P < .025) revealed functionally selective effects of N-back load in the prefrontal cortex corresponding to DLPFC, and large but undifferentiated ERPs in the hippocampus. Spectral analysis revealed specific patterns of increased beta and gamma activity in prefrontal electrodes with gradient representation of increasing back. Cross-validated SVM classification accuracy was found to be significantly above chance for all bands above delta and highest in gamma (mean 56% +/- 0.05, chance 25%, P < .0001). CONCLUSION: We found regional specificity in the prefrontal cortex with task-dependent spectral differences, most significant in gamma and beta bands. In states of greater cognitive effort, increased cross-frequency coupling within a distributed network supports the maintenance of local representations as well as the accumulation of evidence for target matching.