{"success":true,"database":"eegdash","data":{"_id":"6953f4239276ef1ee07a32e3","dataset_id":"ds003682","associated_paper_doi":null,"authors":["Toby Wise","Yunzhe Liu","Fatima Chowdhury","Raymond J. Dolan"],"bids_version":"v1.5.0","contact_info":["Toby Wise"],"contributing_labs":null,"data_processed":true,"dataset_doi":"10.18112/openneuro.ds003682.v1.0.0","datatypes":["meg"],"demographics":{"subjects_count":28,"ages":[],"age_min":null,"age_max":null,"age_mean":null,"species":null,"sex_distribution":null,"handedness_distribution":null},"experimental_modalities":null,"external_links":{"source_url":"https://openneuro.org/datasets/ds003682","osf_url":null,"github_url":null,"paper_url":null},"funding":[],"ingestion_fingerprint":"3cbea1ccffa007b78a78c8a376e2d2f0d27af88c20f32021bce0221d38cb1dd3","license":"CC0","n_contributing_labs":null,"name":"Model-based aversive learning in humans is supported by preferential task state reactivation","readme":"This dataset contains raw and processed MEG data for the paper \"Model-based aversive learning in humans is supported by preferential task state reactivation\" by Toby Wise, Yunzhe Liu, Fatima Chowdhury & Ray Dolan.\nRaw data is provided as `.fif` files, although it was acquired on a CRF system.","recording_modality":["meg"],"senior_author":"Raymond J. Dolan","sessions":["01"],"size_bytes":227225155192,"source":"openneuro","study_design":null,"study_domain":null,"tasks":["AversiveLearningReplay"],"timestamps":{"digested_at":"2026-04-22T12:25:37.829292+00:00","dataset_created_at":"2021-06-05T16:53:45.247Z","dataset_modified_at":"2021-06-05T22:59:10.000Z"},"total_files":336,"storage":{"backend":"s3","base":"s3://openneuro.org/ds003682","raw_key":"dataset_description.json","dep_keys":["CHANGES","README"]},"tagger_meta":{"config_hash":"3557b68bca409f28","metadata_hash":"7bcc4c38fa5398d9","model":"openai/gpt-5.2","tagged_at":"2026-04-07T09:32:40.872789+00:00"},"tags":{"pathology":["Healthy"],"modality":["Unknown"],"type":["Learning"],"confidence":{"pathology":0.6,"modality":0.4,"type":0.8},"reasoning":{"few_shot_analysis":"Closest few-shot conventions by task purpose are the reinforcement/probabilistic learning examples: (1) \"EEG: Probabilistic Learning with Affective Feedback\" labeled Type=Learning, where the key driver is that the paradigm is reinforcement/probabilistic learning; (2) \"EEG: Reinforcement Learning in Parkinson's\" also maps an RL/learning paradigm to a learning/decision-focused label set. For this dataset, the title explicitly frames the study as \"aversive learning\" and discusses \"task state reactivation\" in support of learning, so by the same convention the Type should be Learning. Few-shot examples do not provide a direct analog to infer stimulus Modality for aversive learning without additional metadata; therefore Modality cannot be confidently inferred from the few-shot set alone. For Pathology, few-shot examples consistently treat datasets with no named diagnosis/clinical recruitment as Healthy.","metadata_analysis":"Key metadata facts available are sparse. The dataset readme states: \"This dataset contains raw and processed MEG data for the paper \\\"Model-based aversive learning in humans is supported by preferential task state reactivation\\\"\" and \"Raw data is provided as `.fif` files\". The title itself is explicit about the construct: \"Model-based aversive learning in humans...\". The tasks field lists a single paradigm name: \"AversiveLearningReplay\". Participants information is limited to a count: \"Subjects: 28\". No metadata lines describe the sensory stimulus channel (e.g., visual/auditory/tactile) or whether an aversive US (e.g., shock) was used.","paper_abstract_analysis":"No useful paper information.","evidence_alignment_check":"Pathology — Metadata says: only \"in humans\" and \"Subjects: 28\" with no mention of patients/diagnoses; few-shot pattern suggests assigning Healthy when no clinical recruitment is stated. ALIGN (no conflict).\n\nModality — Metadata says: only task name \"AversiveLearningReplay\" and no stimulus description; few-shot pattern cannot reliably determine stimulus modality for this task family without explicit stimulus text. ALIGN in the sense that both metadata and few-shot provide insufficient evidence; thus Modality remains Unknown.\n\nType — Metadata says \"aversive learning\" (title) and task name \"AversiveLearningReplay\"; few-shot pattern maps learning/RL paradigms to Type=Learning. ALIGN (no conflict).","decision_summary":"Top-2 candidates per category with head-to-head comparison:\n\nPathology:\n- Healthy: Supported by lack of any clinical recruitment language (\"in humans\", \"Subjects: 28\") and few-shot convention to default to Healthy when no diagnosis is stated.\n- Unknown: Possible because the metadata never explicitly says \"healthy\" or \"controls\".\nDecision: Healthy wins because the dataset provides no indication of a disorder-focused recruitment, matching the catalog convention. Evidence quotes/features: \"in humans\", \"Subjects: 28\". Confidence=0.6 (contextual inference; no explicit 'healthy').\n\nModality:\n- Unknown: Supported because there is no stimulus description beyond the task name.\n- Visual: Plausible for many state-based learning tasks, but not supported by any quoted metadata.\nDecision: Unknown wins due to absence of explicit modality facts. Evidence quotes/features: only \"AversiveLearningReplay\" without stimulus description. Confidence=0.4.\n\nType:\n- Learning: Supported explicitly by \"aversive learning\" in the title and the task name \"AversiveLearningReplay\"; also matches few-shot learning/RL conventions.\n- Decision-making: Plausible for model-based tasks, but secondary to the explicit framing as learning.\nDecision: Learning wins because learning is explicitly named and appears to be the primary research purpose. Evidence quotes/features: \"Model-based aversive learning...\", \"AversiveLearningReplay\". Confidence=0.8 (2 explicit metadata cues + strong few-shot analog)."}},"nemar_citation_count":1,"computed_title":"Model-based aversive learning in humans is supported by preferential task state reactivation","nchans_counts":[{"val":414,"count":336}],"sfreq_counts":[{"val":1200.0,"count":336}],"stats_computed_at":"2026-04-22T23:16:00.222351+00:00","total_duration_s":114318.225,"canonical_name":null,"name_confidence":0.89,"name_meta":{"suggested_at":"2026-04-14T10:18:35.342Z","model":"openai/gpt-5.2 + openai/gpt-5.4-mini + deterministic_fallback"},"name_source":"author_year","author_year":"Wise2021"}}