Early detection of cognitive decline is crucial for maintaining well-being in aging populations. As a pilot study, this research explores the potential of auditory-evoked neurophysiological responses to predict cognitive function in young and middle-aged adults. We developed a deep learning model to estimate reaction time, a proxy for cognitive performance, from electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) data. Participants engaged in personalized soundscape creation and listening tasks, designed to elicit auditory processing relevant to communication. Our model demonstrated significant predictive accuracy, achieving an R2 value exceeding 0.8 for attention-related cognitive assessments. Our findings suggest that the synergy of passive BCI cognitive assessment, conducted within active auditory paradigms, and deep learning analysis of neurophysiological data, represents a highly promising non-invasive strategy for objective cognitive monitoring. This methodological advancement offers considerable potential for the scalable early detection of cognitive decline, which could facilitate more timely therapeutic interventions and ultimately promote overall well-being. 

https://2025.ccneuro.org/abstract_pdf/Kojima_2025_Soundscape_Neuroarts_Cognitive_Well-being_EEG_fNIRS.pdf