Modulating complex brain states using MVPA-based neurofeedback: A systematic review

Multi-voxel pattern analysis (MVPA)-based fMRI neurofeedback is a promising tool for modulating brain states and influencing cognition and behaviour, yet its effectiveness remains unclear. This systematic review investigates its efficacy across emotion regulation, fear conditioning, associative and perceptual learning, attention, craving, semantic neurofeedback and motor rehabilitation, incorporating a meta-analytic synthesis of neural effects. A systematic search of PubMed, Web of Science, IEEE, and DPBL databases identified 29 studies. Published studies exhibit considerable variability in protocol designs and methodological aspects related to the implementation of MVPA-based neurofeedback. Overall, the global meta-analysis revealed a moderate, statistically significant effect of MVPA-based neurofeedback on neural outcomes, with subgroup analysis indicating a similar moderate effect in emotion regulation. Most studies support neural modulation following neurofeedback training across multiple domains, though behavioural outcomes are less consistent. Fear reduction, attention and perceptual learning studies demonstrated both neural and behavioural changes, while associative learning and craving reduction studies showed evidence of neural regulation but unclear behavioural outcomes. Studies on motor rehabilitation and semantic neurofeedback demonstrated neural modulation but lacked behavioural assessments. Emotion regulation studies consistently supported neural modulation, however, only a few studies reported behavioural improvements. While these findings highlight the potential of neurofeedback, they also emphasise the need for standardized methodologies and clearer theoretical frameworks. Clarifying terminology used to define different approaches and addressing MVPA-specific methodological considerations - such as preprocessing, motion correction, and classifier selection - will be critical for refining neurofeedback protocols and unlocking the full potential of MVPA-based neurofeedback for both clinical and research applications.