our motivation is that we wanted to compare across groups (typically developing, neuropsychiatric disorder models, treatment groups) and group comparisons for traits so variable across individuals (like reaction to novel animals) required large sample sizes

Thanks! I'm a big fan of your approach! Our reconstructions are much lower quality but we think they're good enough to get high-level information (time spent interacting with stimuli, time near each cagemate, time spent running or climbing, etc) and use models to look for patterns at that level

Thanks!

It took a long time to get here from the project's origin with the help of @trackingactions.bsky.social and all the great open source developers who made it possible. www.nature.com/articles/s41...

We will be making the code and initial dataset (>2k videos with 2-4 animals each; over 3 billion timepoints) publicly available once we finish the paper revisions. It was really great working with talented young scientists @erincorbett3.bsky.social and @kimneuro.bsky.social on this project.

Something positive - I think the journal is going in a great direction. After the editors decide what studies will go under review, the authors and reviewers can collaboratively improve the work without fear of rejection until it meets a quality standard that they all agree is sufficient. Perfect.