Analysis of multimodal physiological signals within and between individuals to predict psychological challenge vs. threat. (February 2020)

• Record Type: Journal Article
• Authors: Khalaf, Aya; Nabian, Mohsen; Fan, Miaolin; Yin, Yu; Wormwood, Jolie; Siegel, Erika; Quigley, Karen S.; Barrett, Lisa Feldman; Akcakaya, Murat; Chou, Chun-An; Ostadabbas, Sarah
• Is Part Of: Expert systems with applications. Volume 140(2020)
• Publication Date: 2020-02
• Subjects: Challenge vs. threat -- Multimodal data fusion -- Mutual information -- Support vector machines -- Clustering -- Graph theory; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2019.112890
• Languages: English
• ISSNs: 0957-4174

Abstract

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