In recent decades, machine learning techniques have been increasingly applied to biological data, including functional magnetic resonance imaging, electroencephalography, and electromyography. A growing number of physiological psychology and psychophysiological studies now utilize deep learning, a machine learning method actively employed across various fields. However, understanding such research can be difficult without understanding the underlying mathematics. This paper introduces the essential mathematics required to understand the operating principles of the multi-layer perceptron, a fundamental deep learning model. It aims to provide a foundation for further exploring deep learning techniques.
The increasing availability of psychological and neuroscience datasets and the growing integration of machine learning techniques in cognitive neuroscience have expanded opportunities for electroencephalography (EEG) research using existing data. Researchers can now conduct sustained investigations through advanced analysis of openly available data, even when direct EEG measurements are not feasible. This study analyzed open EEG datasets recorded during music listening to address a research question distinct from the original study. We applied machine learning methodologies to the EEG data and explained the fundamental aspects of this approach. To facilitate replication and further research, we have published our analysis program online on OSF (https://osf.io/KYBEF). This study aims to establish a foundational framework for analyzing open data with machine learning, supporting future research endeavors in this field.
The performance of image-generation artificial intelligence using deep neural networks has significantly improved. This article presents an overview of the technology driving these advancements and explores its potential applications in brain research. The article also discusses how state-of-the-art generative models can create realistic brain images as data, highlighting their potential for future research.
This study assessed daytime sleepiness in young adults, including individuals with idiopathic hypersomnia (IH) and narcolepsy (NA), using subjective and objective measures. The participants included adults (N=42): 13 with IH, 10 with NA, and 19 with healthy controls (HC). We excluded HC participants with a score of ≥11 on the Japanese version of the Epworth Sleepiness Scale. Participants underwent nocturnal polysomnography in a sleep laboratory, followed by administration of the Japanese version of the Karolinska Sleepiness Scale (KSS-J) before each session of the Multiple Sleep Latency Test (MSLT) the next day. We divided HC participants into two groups based on their mean sleep latency on the MSLT: low sleep propensity (low SP; ≥8 min) and high sleep propensity (high SP; <8 min). The high SP group demonstrated a discrepancy between subjective and objective sleepiness. The correlation between KSS-J scores and sleep latency in each nap trial was low and varied by the group. A significant proportion of HC individuals in the high SP group exhibited shorter sleep latency similar to those with IH or NA despite not experiencing subjective sleepiness. We concluded that some healthy individuals show objective signs of sleepiness comparable to those with hypersomnia or narcolepsy, even without reporting subjective sleepiness, underscoring the significance of using subjective and objective measures to assess daytime sleepiness.