Download from the UCL Research Data Repository and use the linked GitHub code to understand processed data structures. The behavtaskatlas adapter expects a local MATLAB combined block or CSV/TSV trial export preserving the spatialAnalysis trial fields. License restrictions should be kept visible before redistributing derived trial tables.
Dataset
Coen-Sit mouse audiovisual decisions UCL dataset
Preprocessed UCL Research Data Repository release containing behavioral responses, trial timings, inactivation metadata, sorted spikes, and histology records for mice performing an audiovisual spatial decision task.
Access notes
Data formats
- MATLAB files
- Python analysis data
- processed spike-sorting outputs
- UCL Research Data Repository archive
Expected trial-table mapping
Canonical fields the harmonizer expects this dataset to populate.
| Canonical field | Source mapping |
|---|---|
| choice | Wheel-derived left or right choice. |
| correct | Source outcome where defined; conflict-trial correctness may be analysis-specific. |
| evidence_strength | Visual contrast and auditory azimuth should remain separate task variables. |
| feedback | Reward or timeout/error feedback where available. |
| response_time | Reaction time or wheel movement onset relative to stimulus onset. |
| session_id | Behavioral or recording session identifier. |
| stimulus_side | Left or right stimulus location for unisensory and congruent trials. |
| stimulus_value | Signed audiovisual decision evidence after source-specific coding. |
| subject_id | Mouse identifier from processed behavioral data. |
| task_variables | modality condition, visual contrast, auditory azimuth, cue congruence, open-loop versus closed-loop epoch, inactivation target, laser flag, spike sorting unit ids, and histology records. |
| trial_index | Trial number within session. |
Caveats
- The license is non-commercial and no-derivatives, so redistributable derived products need review.
- Raw data are not in the public processed release and are available from the lead contact on reasonable request.
- Conflict-trial correctness follows the source correctResponse field in the default adapter; alternative policies should be modeled as separate analyses.