First, install or update the AllenSDK, our Python based toolkit for accessing and working with Allen Institute datasets.

Data is provided in in NWB format and can be downloaded using the AllenSDK, or accessed directly via an S3 bucket (instructions provided in notebook #1 below). Regardless of which method of file download you choose, we recommend that you load and interact with the data using the tools provided in the AllenSDK, which have been designed to simplify data access and subsequent analysis. No knowledge of the NWB file format is required.

To get started, check out these jupyter notebooks:

  1. Download data using the AllenSDK or directly from our Amazon S3 bucket (download .ipynb)

  2. Identifying experiments and sessions of interest using the data manifest (download .ipynb)

  3. Aligning behavioral data to task events with the stimulus and trials tables (download .ipynb)

  4. Plot quality metrics for the ‘units’ identified in these experiments (download .ipynb)

  5. Visual Behavior Neuropixels Quickstart (download .ipynb)

  6. Analyzing LFP data (download .ipynb)

  7. Analyzing behavior-only data for one mouse’s training history (download .ipynb)

You may also find these tutorials helpful, which were made for students in the Summer Workshop for the Dynamic Brain.

If you have questions about the dataset that aren’t addressed by the whitepaper or any of our tutorials, please reach out by posting at

Visual Behavior - Neuropixels

Overview of the dataset



  • Updated stimulus presentations tables.

  • Add LFP sample rate when loading from cache. (Value is duplicated in the probes metadata table.)

  • Fixed scaling of LFP data in multiple sessions. Amplitudes were off by a factor of 2 for most ecephys sessions. Specifically changed were units table in the session amplitude, recovery_slope, and repolarization_slopelfp data, mean waveform, and current source density data in the NWB files. The scaling is also corrected in released metadata tables.

    • All ecephys sessions had this issue except those listed here: 1059678195, 1108334384, 1108531612, 1109680280, 1109889304, 1111013640, 1111216934, 1112302803, 1112515874, 1113751921, 1113957627, 1115077618, 1115356973, 1118324999, 1118512505

  • Added quality column to units table.


  • Added stimulus presentations columns that are computed on load: is_sham_change

  • Fixed sessions where mouse appears to be running backwards.

  • Fixed issues with loading probes


New Data:

  • Added Local Field Potential (LFP) data associated with individual ecephys session probes.

  • Added 3424 behavior only sessions.

Stimulus_presentations/trials tables changes:

  • Add trials_id column in stimulus_presentations table.

  • stop_time -> end_time

  • Various data type fixes.

Metadata changes:

  • New channels metadata table.

  • Various data type fixes.