JMIR Publications

ABSTRACT

Background:

Neurofeedback trains individuals to modify their brain activity using real-time feedback and is increasingly explored as an adjunct to interventions for anxiety symptoms. However, most electroencephalography-based systems rely on research-grade hardware and therapist supervision, which limits reach and routine use. Consumer-grade EEG and mobile applications could support self-guided regulation outside specialized institutions, but it is unclear whether first-time users can complete such training autonomously, how usable these systems are in real-world settings, and how pre-existing neural state shapes user experience.

Objective:

This study aimed to evaluate the real-world usability of a fully autonomous beta-suppression neurofeedback system running on a tablet with a low-cost consumer EEG headband, to test whether adding a paced-breathing guide improves usability or subjective experience, and to examine whether resting beta activity before the session predicts perceived usability and relaxation.

Methods:

The system was deployed in an unsupervised public science exhibition. A custom Unity application guided participants through the procedure without experimenter involvement. During the task, a sky-themed display became clearer and displayed more stars as beta activity decreased and darker and cloudier as it increased; in the pacer condition this display was overlaid with a circular guide pacing six breaths per minute. Adults (N=52, age 18 to 66 years) were randomized to visual neurofeedback only or visual neurofeedback plus the breathing pacer. The primary outcome was perceived usability. Secondary outcomes were subjective relaxation and mindfulness. Resting EEG from a four-channel consumer headband was used to compute absolute band power. Between-group differences were analyzed with nonparametric tests; associations between beta power and subjective outcomes were examined with Spearman and partial Spearman correlations.

Results:

All 52 participants completed the workflow without assistance. Overall usability was good (System Usability Scale mean 72.3, standard deviation 14.7), within the range usually interpreted as acceptable to excellent. Usability, relaxation, and mindfulness did not differ significantly between groups (usability median 76.0 vs 70.0; Mann–Whitney U=315.5, P=.65). In contrast, baseline neural state systematically predicted user experience: higher resting beta power was associated with lower usability ratings (Spearman ρ=−0.31, P=.03; partial ρ=−0.30, P=.03).

Conclusions:

Our fully autonomous neurofeedback system can be operated successfully by first-time users in unsupervised settings and achieves usability levels comparable to established consumer applications. Adding a paced breathing guide did not substantially change usability or psychological outcomes at the group level. Instead, individual differences in resting beta activity shaped how usable and relaxing the intervention felt. These findings support the feasibility of self-guided neurofeedback in everyday environments and suggest that baseline cortical arousal may be an essential factor when developing and evaluating neurofeedback interventions. Clinical Trial: Not applicable. This study was not a clinical trial and was not registered in a trial registry