Muse 2 EEG during the MR session — 1 CSV row per second. Charts use a fixed 15 s smoothing window (same length as the tutor sampling window). Both engagement traces are on when the page loads.
Frontal (F) = mean of AF7 & AF8. All (A) = mean of TP9, TP10, AF7 & AF8. Band powers are 5 s averages. Only the listed band, engagement, gamma, and blink columns are stored.
When the neuroadaptive admin is ON, Frontal_EngagementIndex and All_EngagementIndex are 0–1 scores after the 2 × 60 s dual-task calibration (\(E_{min}\) / \(E_{max}\), 15 s window). Frontal matches the tutor / video fade signal; All uses the same phases with the four-sensor ratio.
XR_session_Simplified_<date-time>_<id>.csv — one row per second from SimplifiedMetricsLogger.
| Column | What is logged |
|---|---|
| Timestamp | Wall-clock time of the row (HH:mm:ss.fff) |
| Frontal_Delta_Avg | Mean delta (0.5–4 Hz) over 5 s, AF7 & AF8 |
| Frontal_Theta_Avg | Mean theta (4–8 Hz) over 5 s, AF7 & AF8 |
| Frontal_Alpha_Avg | Mean alpha (8–12 Hz) over 5 s, AF7 & AF8 |
| Frontal_Beta_Avg | Mean beta (13–30 Hz) over 5 s, AF7 & AF8 |
| Frontal_EngagementIndex | Frontal engagement. After calibration: 0–1 tutor/video score. During baseline: live AF7/AF8 ratio |
| Frontal_GammaCoherence | Pearson r of gamma between AF7 and AF8 over 5 s (−1 to +1) |
| All_Delta_Avg | Mean delta over 5 s, TP9, TP10, AF7, AF8 |
| All_Theta_Avg | Mean theta over 5 s, all four sensors |
| All_Alpha_Avg | Mean alpha over 5 s, all four sensors |
| All_Beta_Avg | Mean beta over 5 s, all four sensors |
| All_EngagementIndex | All-sensor engagement. After calibration: 0–1 using that scope’s floor/ceiling. During baseline: live four-sensor ratio |
| All_GammaCoherence | Gamma correlation left (TP9+AF7) vs right (TP10+AF8) over 5 s |
| BlinkNow | 1 if a blink was detected in that 1 s interval, else 0 |
| Connection_AF7 | Muse HSI fit at AF7. 1 = good (green), 2 = mediocre (orange), 4 = poor / no signal (red) |
| Connection_AF8 | Muse HSI fit at AF8 (same scale) |
| Connection_TP9 | Muse HSI fit at TP9 (same scale) |
| Connection_TP10 | Muse HSI fit at TP10 (same scale) |
| AiAnswerCount | Running count of AI tutor speech segments in this session |
| AiAnswerStartTime | Wall-clock start of the current or last AI answer (blank if none yet) |
| AiAnswerEndTime | Wall-clock end of the current or last AI answer (blank if none yet) |
Wall-clock time when the row was saved (HH:mm:ss.mmm). Chart time is relative to the first row.
Mean delta (0.5–4 Hz) power over 5 s, averaged across AF7 and AF8. Associated with deep sleep / low arousal.
Mean theta (4–8 Hz) power over 5 s, AF7 & AF8. Often linked to drowsiness, memory, or internal focus.
Mean alpha (8–12 Hz) power over 5 s, AF7 & AF8. Often increases when relaxed with eyes closed.
Mean beta (13–30 Hz) power over 5 s, AF7 & AF8. Often associated with active thinking and alertness.
After calibration finishes: 0–1 from the 2 × 60 s dual-task (the tutor / video score). During the two baseline minutes it is the live AF7/AF8 ratio so you can see the sensors move.
Pearson correlation of gamma power between AF7 and AF8 over 5 s (−1 to +1). Higher = more synchronized frontal gamma.
Mean delta power over 5 s, averaged across all four Muse electrodes.
Mean theta power over 5 s, all four sensors.
Mean alpha power over 5 s, all four sensors.
Mean beta power over 5 s, all four sensors.
All four sensors (TP9, TP10, AF7, AF8). After calibration: 0–1 using that scope’s own floor/ceiling from the same 2 × 60 s. During baseline it is the live all-sensor ratio.
Correlation of gamma between left (TP9+AF7) and right (TP10+AF8) over 5 s.
1 if a blink was detected in that 1 s interval, else 0. Useful for spotting movement artifacts.
Muse headband fit per electrode at each 1 s row. 1 = good (green strip), 2 = mediocre (orange), 4 = poor / no signal (red), 0 = no data (black). Chart lines also change colour/dash from these values. Shown as strips under Timeline 1.
Running count of Convai tutor speech segments (logged when the NPC actually speaks, not A-button presses). The pink strip and “AI answers” counter use this.
Wall-clock start of the current or last AI answer (blank if none yet).
Wall-clock end of the current or last AI answer (blank if none yet). Together with start time this draws the interaction-frequency strip.
Click or drop MP4
Drop XR_session_Simplified_*.csv
Frontal + All bands, both engagement indexes, gamma, blink
Workflow: trim CSV/video ranges first, then mark a clear video cue and align it to the matching CSV time (AI answer bar, engagement change, or lesson start). Use nudge or [ ] keys for fine adjustment.