Dataset

Steinmetz mouse visual decision Figshare/ONE dataset

Public behavioral and Neuropixels dataset from ten mice performing a lateralized visual contrast decision task designed to separate visual stimulus, action, choice, and engagement variables.

Access notes

Download from Figshare and extract one session directory locally. The atlas Steinmetz adapter reads `trials.*.npy` ALF files from that extracted directory and keeps no-go/no-turn outcomes explicit as canonical withhold choices.

Data formats

  • ONE-compatible ALF objects
  • NumPy arrays
  • Figshare archive

Expected trial-table mapping

Canonical fields the harmonizer expects this dataset to populate.

Canonical field Source mapping
choice Wheel-derived left, right, or no-go/no-turn outcome.
correct Source feedback or reward outcome mapped to correctness.
evidence_strength Absolute contrast difference or source-specific bilateral contrast tuple.
feedback Reward or error feedback from source trial fields.
response_time Movement onset or response time relative to stimulus onset.
session_id ONE session identifier.
stimulus_side Side with positive visual evidence, or none for no-stimulus trials.
stimulus_value Signed contrast difference, preferably right contrast minus left contrast.
subject_id Mouse identifier from session metadata or subject folder.
task_variables left contrast, right contrast, go cue, stimulus onset, wheel movement, no-go condition, reward volume or feedback, passive replay flag, and neural probe/session metadata.
trial_index Trial number within the trials object.

Caveats

  • The task is not a strict 2AFC dataset because no-go/no-turn outcomes are valid.
  • Trial-level comparisons with IBL visual decision records require an explicit mapping from bilateral contrast combinations to the canonical stimulus axis.

Papers

Findings

Curves attached to this dataset on shared signed-evidence axes.

See all findings overlaid on shared axes.