Analysis of Multimodal Physiological Signals Within and Between Individuals to Predict Psychological Challenge vs. Threat

摘要: Abstract Challenge and threat characterize distinct patterns of physiological response to a motivated performance task where the vary as function an individual's evaluation demands relative his/her available resources cope with demands. responses during have been used understand psychological, behavioral, biological phenomena across many domains. In this study, we aimed investigate individual group-level variations in responding series tasks that difficulty. The proposed approach is by documented differences observed tasks, such first focus on rather than comparisons. Then, through our analysis individuals identify sub-groups (i.e., clusters) share common varying difficulty perform across-subject within each cluster. This from existing studies which typically do not examine vs. subgroup-specific activity. Such enables us can be predict self-reported judgments challenge higher accuracy subgroup compared includes entire sample population single group. Specifically, three hypotheses were tested: (H1) will different sets (features) difficulty; (H2) there subgroups who salient features clusters differentiate their (H3) predicting shared all or computed features. To test these hypotheses, developed integrated analytic framework for multimodal data analysis. We employed experiment participants completed mental arithmetic increasing modalities collected. Analyses revealed best differentiated within-individual tasks. Support vector machine (SVM) classifiers trained using both group states. Results showed that, within-group classification model achieved self-report prediction alternative without feature selection.