Fig 1.
My Virtual Dream: the ‘dreamery’ and the stage.
In front of an audience, twenty participants at a time experienced a two-part interaction within the dome. Based on the collective neurofeedback of all 20 participants, the ‘dreamers’, artistic video animations were projected on the 360° surface of the semi-transparent dome and soundscapes were generated based on a pre-recorded sound library and improvisations from live musicians.
Fig 2.
(A) EEG data observation and welcome messages. Players are ordered from left to right. (B) Solo relax. Number of particles indicates cumulative relaxation result (e.g., players 4 and 5 accumulated large particle cloud). Ovals on the top of the screen represent additional feedback such that light shading of the oval signifies a+ state (e.g. player 4). (C) Solo concentrate. Brightness of particles indicates cumulative concentration result (e.g., players 1, 2 and 4 have good result). Dot inside an oval indicates that player is in b+ state (e.g., players 1 and 4). (D) Group game, guided and freestyle.
Fig 3.
Each phase of the game ended with fireworks display, size and brightness of which were determined by the performance of the participants. Total duration of the game was 6.5 min. In the Tutorial individual thresholds for alpha and beta were estimated based on guided ‘relax’ and ‘concentrate’ conditions. Participants obtained individual visual feedback on their performance to either increase alpha or beta. Solo 1 and 2 games were qualitatively identical with the tutorial, however individual thresholds were used. During the ‘group-guided’ game each ‘pod’ obtained feedback about the collective performance in addition to the individual feedback. In the ‘Group—freestyle’ period, participants did not obtain specific instructions other than to attempt to synchronize as a group by targeting the same brain state.
Fig 4.
Results of the automated artifact rejection procedure.
EEG traces from the left frontal channel of 5 randomly selected participants are shown during the first 20 s of the Solo 1-concentrate condition. Shaded areas indicate rejected intervals.
Fig 5.
Relaxation vs. concentration states (example left frontal channel).
(A) Mean RSP curves across 8 guided conditions in alpha and beta frequency ranges. (B) Left: mean participant scores with respect to the main effect of relaxation vs. concentration across 1–50 Hz frequency range. Error bars indicate 95% confidence intervals (CI) Right: frequency pattern associated with the main effect. Frequencies with reliable positive (negative) bootstrap ratios exhibited greater (lower) RSP for concentration conditions compared to relaxation conditions and are indicated by red (blue) circles. As hypothesized, concentrate conditions showed more power in beta range and less power in alpha range, compared to relax conditions.
Fig 6.
Training effects on brain states (example left frontal channel).
(A) Relaxation training effect. Left: mean participant scores with respect to the relaxation training effect, with error bars representing 95% CI. Right: associated frequency pattern with reliable positive (negative) bootstrap ratios indicated by red (blue) circles. Overall effect of relaxation training is a decrease in 17–18 Hz and 35–45 Hz frequency range. (B) Concentration training effect. Left: mean participant scores with error bars representing 95% CI. Right: associated frequency pattern with reliable positive (negative) bootstrap ratios indicated by red (blue) circles. Overall effect of concentration training is an increase in beta power (20–40 Hz frequency range) and a decrease in delta power (<3 Hz). (C) Reliability of concentration training effect across a series of analyses with varying number of participants, N. For each N we plotted mean estimate of the reliability measure maxp. Reliable results (maxp < 0.05, i.e., below the red dotted line) are consistently obtained when the number of participants is >>200.
Fig 7.
Neurofeedback performance measures.
(A) Group mean alpha performance aP and beta performance bP for all participants taken together (yellow bullets), with 95% CI’s shown as error bars, beta learners (gray bullets) and non-learners (white bullets). Conditions where neurofeedback did not depend on the respective band of interest are shown in desaturated color. Black asterisks indicate conditions which expressed reliable PLS difference between learners and non-learners. (B) Analysis of differences in baseline RSP between beta learners and non-learners. Top: mean participant scores with error bars representing 95% CI. Bottom: associated frequency pattern for the left frontal channel with reliable positive (negative) bootstrap ratios indicated by red (blue) circles. High power in delta range and low power in beta/gamma range during baseline predicted subsequent beta learning.
Fig 8.
Time-of-night effect (example left frontal channel).
(A) Mean RSP curves from 29 sessions during Tutorial-relax condition. (B) Top: omnibus correlation between RSP values and time-of-night across all guided conditions. Error bar indicates 95% CI based on bootstrap resampling. Bottom: associated frequency pattern, represented with bootstrap ratios across conditions. Reliable positive (negative) bootstrap ratios are positively (negatively) correlated with time-of-night and are indicated by red (blue) circles. As the night progressed, there was as a gradual shift towards more power in high frequencies and less in low frequencies.
Fig 9.
Sex differences in RSP (example left frontal channel).
(A) Mean RSP curves for males and females during Solo2-concentrate condition. (B) Sex effect across all guided conditions. Top: mean participant scores with error bars representing 95% CI. Bottom: associated frequency pattern, represented with bootstrap ratios across conditions. Reliable positive bootstrap indicated by red (blue) circles identify frequencies (35–45 Hz) where females have more power compared to males. Weak trend by which males exhibit more power alpha range (blue bootstrap ratios) is not consistently reliable across conditions.