Loneliness EEG - Roving Oddball Task

# Loneliness EEG — Roving Oddball Task ## Overview This dataset contains 64-channel EEG recordings from young adults performing a roving oddball task with emotional face stimuli, together with behavioural ratings of the same stimuli on valence, arousal, and dominance. Participants were recruited and grouped by their score on the UCLA Loneliness Scale (Version 3) to investigate whether subjective loneliness is associated with altered automatic processing of social-emotional information, indexed by ERP components elicited by emotional facial expressions in a roving oddball stream. - **Institution:** Department of Psychology, University of Amsterdam - **Acquisition system:** BioSemi ActiveTwo (72 channels, 2048 Hz) - **Reference / ground:** CMS at Cz / DRL at Fz (BioSemi active-electrode design) - **BIDS version:** 1.9.0 - **License:** CC-BY-4.0 ## Participants `participants.tsv` lists 99 adults grouped by UCLA Loneliness Scale (UCLA-LS) score into `Lonely` (n = 36) and `Non-Lonely` (n = 63). The file also includes demographics (age, gender, handedness via the Edinburgh Handedness Inventory laterality quotient, educational attainment, self-reported ethnic background) and self-report questionnaire totals: Brief Symptom Inventory-53 (per-subscale mean item scores plus GSI/PSDI/PST), Lubben Social Network Scale-6 (family/friends/total), GAD-7, PSS-10, and PHQ-9. Variable definitions and units are documented in `participants.json`. The `exclude` column flags 7 participants (sub-22, sub-37, sub-42, sub-56, sub-65, sub-77, sub-87) recommended for exclusion from primary analyses due to data-quality or protocol issues. Excluded participants are retained in the dataset so that users may make their own inclusion decisions. Note: an empty cell in `ucla-ls` (`n/a`) indicates a participant who screened were screened into one of the groups based on their UCLA-LS score, but their actual score was not recorded due to a data-collection error. ## Tasks ### `task-RovingOddball` (EEG) A roving auditory-style oddball implemented in the visual domain with face stimuli. Trains of repeated presentations of a single face identity are interrupted by a change to a new identity, which acts as the "deviant" beginning of the next train (i.e., every deviant is also the first standard of the following train). Face stimuli were selected from the FACES database, and preprocessed using webmorphR and SHINE. Faces displayed either a **happy** or **angry** expression. Trials are framed by a fixation cross whose colour signals the response requirement on the upcoming train: - **White fixation cross (trigger 20):** no response required (`no_response`). - **Red fixation cross (trigger 30):** participant should respond on any catch trial within the upcoming train (`response`). Each face presentation carries a two-digit trigger code `XY`, where `X` indexes the (colour × emotion) combination and `Y` indexes the serial position within the train (1 = deviant, 2–8 = standards): | `X` | Colour × emotion | |----:|--------------------------| | 1 | red × angry | | 2 | white × angry | | 3 | red × happy | | 4 | white × happy | | 5 | red × neutral *(rare)* | | 6 | white × neutral *(rare)* | Additional codes appearing in `*_events.tsv`: | Code | Meaning | |-----:|------------------------------------------| | 20 | Fixation cross, white (no-response cue) | | 30 | Fixation cross, red (response cue) | | 101 | End-of-task marker | | 111 | Button press | | 200 | Feedback | The full numeric→label mapping used by the preprocessing pipeline (e.g. `41 → white/happy/deviant/1`) is defined in `code/1_organise_bids.py::EVENT_DICT` in the analysis repository. No `events.json` is shipped because trial types are retained as raw integer trigger codes in `*_events.tsv` to keep the source-of-truth unambiguous; users wishing to recode to descriptive labels should consult that mapping. ### `task-ImageRatings` (behavioural) After the EEG session, participants rated each face stimulus on three Self-Assessment Manikin (SAM) dimensions — **valence**, **arousal**, and **dominance** — on a 1–5 Likert scale. Data are stored as `sub-<ID>/beh/sub-<ID>_task-ImageRatings_beh.tsv`. Columns: `stim_file`, `emotion` (`happy`/`angry`), `dimension`, `rating`. The accompanying `task-ImageRatings_beh.json` sidecar at the dataset root documents the levels and instructions. ## Stimuli The face images used in the experiment are drawn from a restricted-access FACES database and **cannot be redistributed**. The `stimuli/` folder therefore contains 1×1 black JPEG **placeholders** that share the original filenames, so that the `stim_file` paths referenced in events/behavioural TSVs resolve to a real file on disk and satisfy the BIDS validator. See `stimuli/README` for details and filename conventions (`SHINEd_<id>_<age>_<sex>_<emotion>_<version>.jpg`). To obtain the original images, contact the dataset maintainer: https://faces.mpdl.mpg.de/imeji/ ## Derivatives `derivatives/` contains the outputs of the preprocessing pipeline used in the accompanying analyses: - `derivatives/sub-<ID>/eeg/sub-<ID>_task-RovingOddball_eeg-epo.fif.gz` — cleaned, filtered, ICA-corrected, and epoched EEG (MNE-Python `Epochs`). - `derivatives/sub-<ID>/eeg/sub-<ID>_task-RovingOddball_eeg.html` — per-subject preprocessing QC report. - `derivatives/task-RovingOddball_desc-preprocessing_qc.tsv` — dataset-level QC table with per-subject epoch counts (broken down by emotion × repetition), retained sampling frequency, and bad channels. Derivatives are provided for convenience and reproducibility of the reported analyses; users may always recompute them from the raw data. ## Recommended citation If you use this dataset, please cite: > Bathelt, J., van Dijk, C., & Otten, M. (forthcoming). *Loneliness in the Brain: Distinguishing Between Hypersensitivity and Hyperalertness.* University of Amsterdam. ## Acknowledgements We thank Famke Bruggeman, Rosalind Dingarten, Vita Karkauskaite, Cleo Rong, Sophie Serrarens, Evi Veer, and Jari Vink for their assistance with data collection. We are also grateful to the participants who generously gave their time to this research. ## Contact Joe Bathelt — Department of Psychology, University of Amsterdam: [e-mail]