Guide to choosing a Behavior dataset

July 5-23, 2021

Everyone should consider the behavior-only datasets that we have, which are very rich with many subjects and many trials.

Caltech

The Caltech dataset (youtube) has pose-tracking data from socially-interacting mice, and is well supported with code and a project template.

Credit for data curation: Ann Kennedy

Run

View

Loader notebook

Open In Colab

View the notebook

IBL

The IBL dataset (youtube) is a visual decision-making task, that is very similar to the one used in the Steinmetz dataset.

Credit for data curation: Eric DeWitt and the IBL team

Run

View

Exploration notebook

Open In Colab

View the notebook

Psychometric notebook

Open In Colab

View the notebook

Additional analyses

Open In Colab

View the notebook

Laquitaine & Gardner, Neuron, 2017

The Laquitaine & Gardner dataset contains behavioral data collected in a motion direction estimation task performed by 12 human subjects. The data include motion direction estimates, reaction times and various task variables recorded for 83,214 trials. The authors varied the strength of sensory evidence (motion coherence) and priors (set of directions) and compared the trial-estimate distribution with the predictions of Bayesian observer models.

Run

View

Loader & Exploration notebook

Open In Colab

View the notebook

Additional analyses

Open In Colab

View the notebook