Poster E40, Monday, March 26, 2:30-4:30 pm, Exhibit Hall C
Predicting cognitive performance on the basis of electrophysiological properties of resting state neuronal dynamics
Elena Cesnaite1, Keyvan Mahjoory2, Arno Villringer1,3, Vadim V. Nikulin1,4; 1Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany, 2Functional Brain Mapping Laboratory, Université Libre de Bruxelles, Brussels, Belgium, 3Clinic for Cognitive Neurology, University of Leipzig, Leipzig, Germany, 4Center for Cognition and Decision Making, National Research University Higher School of Economics, Russian Federation
Neuronal dynamics at rest as well as cognitive task performance differ substantially across subjects, however, the link between both remains elusive. In order to further address this link, we used long-range temporal correlations (LRTC) that have been previously shown to relate to critical dynamics, hypothesized to be beneficial for task performance. In a current study, we have hypothesized that stronger alpha band LRTC at rest would correspond to better cognitive performance. Multichannel resting state EEG has been recorded from 117 young, right-handed subjects (81 male, mean age 25.5, SD=3.14) and linked to cognitive performance: fluid intelligence, working memory and executive functions. Mean amplitude, peak frequency and LRTC of alpha oscillations were used as EEG predictors. Statistical analysis was performed using non-parametric Spearman correlations and cluster statistics to account for multiple comparisons. In addition, we performed inverse modeling in order to infer neuronal sources of best predicting regions. Among the three measures only LRTC in males predicted working memory performance: LRTC exponents positively correlated with an accuracy of a numeric 2-back task. Inverse modeling showed several regions primarily at the right temporal and parietal lobes to be most predictive. While areas in these regions have been previously linked to number processing in fMRI studies, our findings suggest that higher LRTC exponents at rest reflect their readiness to quickly process and update numeric information when task demands are present. *First two authors share co-first authorship of the work.
Topic Area: EXECUTIVE PROCESSES: Working memory