Abstract
Alpha rhythm (8-13 Hz) is linked to relaxed mental state in humans. Earlier reports have shown that individuals can increase their alpha power if provided with a valid feedback, compared to controls who are provided invalid feedback. However, these results remain controversial, partly because controls may be in a different behavioral state, making it difficult to directly compare their alpha power with the valid group. We here address this issue by using an experimental paradigm in which an invalid feedback is given on a fraction of trials, such that both valid and invalid conditions can be obtained from the same participant. Using electroencephalography (EEG), we recorded alpha power from the occipital area from 24 humans (nine females) and played a feedback tone which could be valid (tone frequency proportional to alpha power), invalid (tone sequence from a previous valid trial; participants were unaware of this condition), or neutral (constant tone frequency). We found that during eyes closed-state, neurofeedback did not enhance alpha activity beyond pre-trained state within the experimental duration, probably because of saturation of alpha rhythmicity. However, for participants whose alpha power decreased over time within a trial, valid feedback helped them to sustain alpha more than invalid feedback. Further, alpha increase showed a weak negative correlation with their self-reported attentional load but was uncorrelated with relaxation levels. Our results reconcile many conflicting reports in the neurofeedback literature, and show that even under most stringent control, valid neurofeedback can help participants who are otherwise unable to sustain their alpha activity.