Heartbeat-evoked responses and self-reported meditative depth

EEG (64 channels), electrocardiography, respiration, electromyography and electrodermal activity recorded from experienced Vipassana meditators during two sessions of meditation with repeated self-reports of meditative depth. The dataset derives from the experiment described in Reggente et al. (2024) and accompanies an article analysing heartbeat-evoked responses across reported depth (reference to be added on publication). The dataset contains the raw recordings only; no preprocessed data are included.

Participants

Participants were experienced Vipassana meditators who self-confirmed proficiency in Vipassana as taught by S.N. Goenka, with at least five years of meditation experience and sessions of at least ten minutes practiced at least five times per week. They reported no history of hypotension, neurological or psychiatric disorders, or use of central nervous system medications. All study materials were approved by the Advarra Institutional Review Board (Columbia, MD; Protocol No. Pro00063946) and informed consent was obtained from each participant through Castor eConsent.

participants.tsv lists the participants in this dataset. The in_analysis column marks those included in the analyses of the associated article, and the other columns give the age, sex and meditation practice reported at screening (participants.json). Participant labels are sequential; study record identifiers and session dates are not included.

Design

Each session (ses-1, ses-2, at least one week apart) consisted of six blocks performed with eyes closed: resting state (5 min), pre-up meditation (5 min), meditation with a sustained attention passive oddball sequence (SAPO; 35 min), an active oddball task (5 min), meditation without the sequence (SILENT; 35 min) and a cool-down meditation (5 min). The two 35-minute blocks were interchanged between a participant's two sessions. During these blocks participants indicated the deepest level of meditation since their last report (1 to 5) followed by their confidence (1 to 5) with a clicker, either spontaneously (emerge) or when prompted (probe). See task-meditation_eeg.json for the full task description.

Recording

• EEG: BrainAmp DC 64 (Brain Products; two stacked 32-channel units, hence the "dual-amp" system of Reggente et al.) with a 64-channel BrainCap TMS fitted with sintered Ag/AgCl multitrodes plus one ground and one reference, 500 Hz, online reference FCz, ground AFz, impedances below 10 kOhm. The cap was used unmodified: the holes it allows were never cut, and no transcranial magnetic stimulation was part of this study. The scalp was exfoliated with NuPrep (Weaver and Co.) and the electrodes filled with Abralyt 2000, a high-chloride abrasive electrolyte gel (Neurospec; EASYCAP).
• Physiology: Cognionics Aim II system, recording electromyography from the left and right sternocleidomastoid muscles, respiration by bio-impedance on the left and right pectoralis major (the same leads recorded the ECG), and electrodermal activity from the non-dominant palm.
• All streams were acquired and synchronized with NeuroPype Experiment Recorder (Intheon) over the Lab Streaming Layer and saved as XDF files.

Conversion to BIDS

The conversion code is in code/ (step0_convert_to_bids.m for the recordings, step0_convert_questionnaires.m for the questionnaires). For each session, the XDF streams were read with clock synchronization and jitter removal (EEGLAB xdfimport plugin) and cropped to the time span common to the EEG, physiological and marker streams. The physiological channels were resampled onto the EEG sample times by linear interpolation, and each marker was placed at the nearest EEG sample. Intervals of more than 0.1 s between consecutive time stamps were treated as missing data (see Known issues). No filtering or re-referencing was applied.

Channels: 64 EEG channels, EMG1 to EMG4 (the ExG channels of the Aim II), ECG, RESP, GSR and PPG. The photoplethysmography channel is not used in the associated article and is flat in 36 of the 78 recordings. The Aim II oxygen saturation, heart rate, temperature and trigger channels held constant values in nearly all recordings and no physiological signal in any, and the packet counter is a device counter; these channels are not included. The device labels every channel in microvolts; the respiration, electrodermal and photoplethysmography channels are left in the raw units of the device (n/a in channels.tsv).

Events

Each events.tsv keeps one row per marker, with the original marker text in value, its category in trial_type, the block in block, and the depth and confidence ratings, oddball response times, durations logged at the end of sub-blocks and blocks, and clicker keys in their own columns. All columns and levels are described in task-meditation_events.json, which also carries the HED annotations (HED 8.4.0). Ratings are given as registered: values of 0 and above 5 occur. In the earliest recordings (sub-005 and sub-006) the rating markers do not name the block; the block given is that of the preceding report or prompt window. A free-text note written by the experiment software in every file is omitted.

Questionnaires

Questionnaires were completed with Castor ePRO and are in phenotype/, in one file per questionnaire and session (for example maas_ses1.tsv and maas_ses2.tsv) with one row per participant, each described by its JSON sidecar:

• Before the session: Brief Profile of Mood States (poms_pre), Mindful Attention Awareness Scale (maas), Cognitive and Affective Mindfulness Scale-Revised (camsr), Five Facet Mindfulness Questionnaire (ffmq) and Freiburg Mindfulness Inventory (fmi).
• After the session: Brief Profile of Mood States (poms_post), Meditation Depth Index (medi), Toronto Mindfulness Scale (tms) and task ratings (session_ratings).
• After the first and the second 35-minute meditation block: ratings in block1_ratings and block2_ratings.

Each file gives the scores followed by the item responses. Free-text answers are not included, and the item wording of the Brief Profile of Mood States is not reproduced. sub-007 has no questionnaires for session 2, and a questionnaire missing for a session has no row.

Known issues

• sub-021, ses-1: the EEG amplifier stopped mid-recording. The file keeps the 31.8 minutes common to all streams.
• sub-005, ses-2: the EEG amplifier stopped at about 87 minutes, during the SAPO block. The file keeps the 86.6 minutes common to all streams.
• sub-006, ses-1: the marker stream ends at the break after the SILENT block, so the post-meditation block is not included. The file keeps the 103.7 minutes common to all streams.
• Gaps in the EEG acquisition. Where no EEG sample was recorded for more than 0.1 s, the samples on either side are stored contiguously, a boundary event marks the gap, and markers registered during the gap are omitted. Times are approximate, from the start of the recording:
  • sub-004, ses-2: one 330.6 s gap at about 97 min (128 markers omitted, including two depth and two confidence ratings of the SAPO block).
  • sub-014, ses-2: one 330.6 s gap at about 77 min (38 markers omitted, including two depth and two confidence ratings of the SILENT block).
  • sub-009, ses-2: two gaps of 46.9 s and 421.1 s at about 58 and 59 min (no markers).
  • sub-002, ses-1: six gaps totalling 16.0 s at about 44 min (no markers); ses-2: six gaps totalling 8.3 s at about 29 min (8 markers omitted), and about 500 interruptions of one or a few samples that are not marked.
• sub-028, ses-2: a 43.7 s gap in the physiological stream at about 52 min. The physiological channels are empty (NaN) during this gap.

References

Reggente, N., Kothe, C., Brandmeyer, T., Hanada, G., Simonian, N., Mullen, S., et al. (2024). Decoding Depth of Meditation: EEG Insights from Expert Vipassana Practitioners. Biological Psychiatry Global Open Science, 100402. doi: 10.1016/j.bpsgos.2024.100402