Advancements in BCI-based neurofeedback systems for working memory

Abstract

Brain-Computer Interface (BCI)-based neurofeedback has emerged as a potentially effective, non-invasive method for enhancing working memory and facilitating cognitive training. Conversely, the methodological heterogeneity characterizing the current research presents a challenge in establishing consistent conclusions concerning system design and efficacy. This systematic review seeks to synthesize the existing body of evidence concerning the design, evaluation, and reported efficacy of BCI-based neurofeedback systems, with a particular focus on their application in enhancing working memory. Adhering to PRISMA guidelines, a structured literature search was performed across Scopus, Web of Science, PubMed, and IEEE Xplore, culminating in the inclusion of 21 studies in qualitative synthesis. The results indicate the clear dominance of EEG-based architectures, with frequency-domain neural features, particularly those derived from alpha-, theta-, and ratio-based measures, being the most prevalent signals utilized. Visual feedback interfaces were the most commonly used, even though some studies used gamified, mobile, or immersive environments. Regarding the evaluation of the results, sham and passive control groups were often used. However, the methods were highly heterogeneous, and long-term follow-up assessments were relatively rare. The reported effectiveness findings were positive but varied. Some studies showed improvements in working memory, while others reported mixed, task-specific, or null effects. A consistent pattern observed was that the successful alteration of the specific neural characteristic did not invariably correlate with demonstrable cognitive enhancement. Based on these findings, this review proposes design and validation guidelines aimed at improving methodological consistency, strengthening evaluation practices, and enhancing comparability across future BCI-based neurofeedback studies.