The Potential of Scalp EEG in Extracting Auditory-Language Information and Understanding Individual Differences in Neural Representation

In recent years, research on brain-machine interfaces or brain-computer interfaces has increasingly focused on establishing systems that enable communication through the extraction of linguistic information from brain activity signals. Studies integrating large-scale language models into brain-machine interfaces have demonstrated the generation of characters and phrases. However, many of these approaches rely on hand or mouth motion data, highlighting the challenge of directly extracting linguistic information. This review introduces global trends in this research field and examines the difficulties and potential of extracting linguistic information from brain-activity signals. In particular, it discusses the feasibility of noninvasive methods in comparison to invasive brain activity recordings, which are primarily being explored for practical applications in the United States, with a focus on the neural representation of linguistic information and individual differences.