{"success":true,"database":"eegdash","data":{"_id":"6953f4249276ef1ee07a3424","dataset_id":"ds005907","associated_paper_doi":null,"authors":["Ethan Campbell","James F Cavanagh"],"bids_version":"1.1.1","contact_info":["Arnaud Delorme"],"contributing_labs":null,"data_processed":false,"dataset_doi":"doi:10.18112/openneuro.ds005907.v1.0.0","datatypes":["eeg"],"demographics":{"subjects_count":53,"ages":[43,50,22,18,31,24,44,38,32,54,37,51,33,50,28,36,44,27,34,49,49,30,29,44,28,45,37,44,45,25,55,24,34,30,31,34,24,22,27,33,26,50,32,47,40,42,48,43,53,44,47,52,48],"age_min":18,"age_max":55,"age_mean":37.867924528301884,"species":null,"sex_distribution":{"f":30,"m":23},"handedness_distribution":null},"experimental_modalities":null,"external_links":{"source_url":"https://openneuro.org/datasets/ds005907","osf_url":null,"github_url":null,"paper_url":null},"funding":[],"ingestion_fingerprint":"0a59d27415cb51e9f4875d832833c7ac83e8cdeca34731ac359281d6a1282e87","license":"CC0","n_contributing_labs":null,"name":"EEG: RL Task (3-Armed Bandit) with alcohol cues in hazardous drinkers and ctls","readme":" RL task (3-armed bandit) with alcohol vs. beverage cues in N=53 Community participants.   Data collected from 2019-2021 in the CRCL at UNM.   The paper [Campbell, E., Singh, G., Claus, E.D., Witkiewitz,K., Costa, V.D., Hogeveen, J; & Cavanagh, J.F. Electrophysiological markers of aberrant cue-specific exploration in hazardous drinkers]  Should be coming out in print soonish.  Your best bet for understanding this task  would be to read that paper first.   For more info on triggers and outputs, see BEH_EXPLAIN.m file in code folder.  - James F Cavanagh 03/06/2023","recording_modality":["eeg"],"senior_author":"James F Cavanagh","sessions":[],"size_bytes":6062770483,"source":"openneuro","study_design":null,"study_domain":null,"tasks":["ThreeArmedBandit"],"timestamps":{"digested_at":"2026-04-22T12:28:59.426630+00:00","dataset_created_at":"2025-02-06T23:20:31.333Z","dataset_modified_at":"2025-02-07T00:07:46.000Z"},"total_files":53,"storage":{"backend":"s3","base":"s3://openneuro.org/ds005907","raw_key":"dataset_description.json","dep_keys":["CHANGES","README","participants.json","participants.tsv","task-ThreeArmedBandit_events.json"]},"tagger_meta":{"config_hash":"4a051be509a0e3d0","metadata_hash":"6fa5b970630864f0","model":"openai/gpt-5.2","tagged_at":"2026-01-20T18:43:13.468361+00:00"},"tags":{"pathology":["Alcohol"],"modality":["Visual"],"type":["Learning"],"confidence":{"pathology":0.8,"modality":0.7,"type":0.6},"reasoning":{"few_shot_analysis":"Most similar few-shot paradigms are:\n- \"EEG: Three armed bandit gambling task\" (Healthy, Visual, Affect): same \"3-armed bandit\" reinforcement-learning style task structure, indicating a screen-based choice task with feedback.\n- \"EEG: Probabilistic Learning with Affective Feedback\" (Healthy, Visual, Learning) and \"EEG: Reinforcement Learning in Parkinson's\" (Parkinson's, Visual, Decision-making): both show that RL/choice-with-feedback tasks in this catalog are typically mapped to either Learning or Decision-making (depending on whether the emphasis is learning from feedback vs choice policy/value).\nThese examples guide conventions: (a) 3-armed bandit tasks are treated as visual choice tasks, and (b) RL tasks are usually labeled Learning or Decision-making rather than Perception/Motor.","metadata_analysis":"Key metadata facts from the provided README:\n- Population/condition cue: \"RL task (3-armed bandit) with alcohol vs. beverage cues\" and paper title mentions \"aberrant cue-specific exploration in hazardous drinkers\".\n- Task/paradigm: \"RL task (3-armed bandit)\".\n- Cohort description: \"N=53 Community participants\" (but the explicit focus is hazardous drinking / alcohol-cue effects).\nThese statements indicate an alcohol-related recruited/stratified population focus and a reinforcement-learning decision task involving alcohol vs non-alcohol cues (likely visually presented, consistent with standard bandit tasks).","paper_abstract_analysis":"No useful paper information.","evidence_alignment_check":"Pathology:\n1) Metadata says: \"alcohol vs. beverage cues\" and study focus on \"hazardous drinkers\".\n2) Few-shot pattern suggests: bandit tasks often use Healthy cohorts (e.g., the healthy 3-armed bandit example), but pathology should follow explicit population focus.\n3) Alignment: CONFLICT on cohort typicality (Healthy in few-shot) vs explicit alcohol-use focus here.\n4) Winner: Metadata FACTS override few-shot convention → Alcohol.\n\nModality:\n1) Metadata says: \"3-armed bandit\" with \"alcohol vs. beverage cues\" (implies cue stimuli presented).\n2) Few-shot pattern suggests: the 3-armed bandit example is labeled Visual; RL tasks in examples are screen-based.\n3) Alignment: ALIGN (both indicate visual cue presentation typical for bandit tasks).\n\nType:\n1) Metadata says: \"RL task (3-armed bandit)\" and paper focus on \"cue-specific exploration\".\n2) Few-shot pattern suggests: RL tasks map to Learning (probabilistic learning) or Decision-making (reinforcement learning with choice emphasis).\n3) Alignment: PARTIAL—both candidates are consistent; metadata does not explicitly state whether learning dynamics or decision policy is primary.\n4) Winner: Learning slightly favored because the task is explicitly framed as \"RL task\" and exploration/learning from feedback is central to bandit paradigms.","decision_summary":"Top-2 candidates per category:\n\nPathology:\n- Alcohol (selected): supported by \"alcohol vs. beverage cues\" and \"hazardous drinkers\".\n- Healthy (runner-up): suggested only by \"Community participants\" and by the healthy bandit few-shot, but not consistent with the explicit hazardous drinking focus.\nSelection justification: explicit hazardous drinking/alcohol cue focus implies Alcohol recruitment/stratification.\n\nModality:\n- Visual (selected): inferred from \"3-armed bandit\" + \"alcohol vs. beverage cues\" and supported by the few-shot 3-armed bandit labeled Visual.\n- Other (runner-up): possible if cues were not sensory stimuli (unlikely for bandit tasks), but no supporting text.\nSelection justification: conventionally bandit cues are visual and few-shot match supports it.\n\nType:\n- Learning (selected): \"RL task (3-armed bandit)\" implies reinforcement learning from feedback; similar to probabilistic learning few-shot.\n- Decision-making (runner-up): \"exploration\" could be treated as decision policy/value-based behavior; similar to the PD reinforcement learning example.\nSelection justification: learning-from-feedback is the most direct mapping given the explicit \"RL task\" wording; decision-making remains plausible so confidence is moderate.\n\nConfidence basis: Pathology has direct alcohol/hazardous drinkers wording; modality and type rely on standard bandit-task conventions with limited explicit stimulus/aim detail in the provided metadata."}},"computed_title":"EEG: RL Task (3-Armed Bandit) with alcohol cues in hazardous drinkers and ctls","nchans_counts":[{"val":58,"count":13},{"val":57,"count":13},{"val":56,"count":11},{"val":55,"count":6},{"val":59,"count":4},{"val":54,"count":2},{"val":61,"count":1},{"val":52,"count":1},{"val":53,"count":1},{"val":33,"count":1}],"sfreq_counts":[{"val":500.0,"count":53}],"stats_computed_at":"2026-04-22T23:16:00.311073+00:00","total_duration_s":51018.902,"author_year":"Campbell2025","canonical_name":null}}