Effect of disturbance in working memory performace: An fNIRs study

This paper demonstrates a near infrared spectroscopic study of understanding effect of disturbance on working memory performance. Embedded prefrontal hemodynamic characters are obtained from recorded data which is filtered and preprocessed by adopting a mean normalization technique. Furthermore, a ReleifF algorithm based feature selection is adopted to sort out the most prominent feature set. A comparative study is drawn to demonstrate the effectivity of the proposed framework by comparing classifiers accuracy which shows that our proposed framework outperforms other conventional principal component analysis (PCA) based feature selection and classification techniques. In addition, correlations among different learning conditions, as well as recall conditions are calculated to demonstrate how diverse conditions (without and with disturbances) alter prefrontal hemodynamics. In addition, a voxel plot approach shows the changing oxyhemoglobin concentrations during different recall tasks (with and without disturbance during learnings). The novelty of this paper lies in understanding the role of disturbance during encoding of information which affects recall performance in working memory tasks.