I’m sorry. My Bluesky profile is meant to be mostly about science. But I have to speak up: my country is not okay. We are being ignored, both internationally and by our own authorities. The violence is real, and the world must see this. #Madagascar #GenZ

Pictures and videos belong to several authors of this paper, and have been used with their permission.

Authors: Me, Emily Grout, Brock, @meg-crofoot.bsky.social, @animal-sounds.bsky.social, Andy Gersick, @coatiben.bsky.social, Kay Holekamp, @wildcognition.bsky.social ition.bsky.social, Amlan Nayak, Josué Ortega, Marie Roch, Eli Strauss, @arispeshkin.bsky.social

Ilustrations by Sai P Kumaran

Linking the paper again: www.pnas.org/doi/10.1073/...
Go check it out! Huge thanks to all the awesome collaborators who made this possible! (next post for credits) (10/10)

Remarkably, predictivity decay was quantitatively similar across all studied species and individuals, hinting at *general principles* underlying animal behaviour. (9/10)

We also explored how current behaviour predicts future behaviour at various timescales—something we call "predictivity decay."
Predictivity decay quantifies how quickly we lose predictive power when forecasting an animal's behaviour due to the accumulation of stochasticity. (8/10)

Surprisingly, across states, individuals, and species, we found the same result: the longer an animal continues a behaviour, the **LESS** likely 📉 it becomes to switch away in the next instant. (7/10)

Imagine a hyena that's been walking for 10 minutes. As it keeps walking, with time does the probability that it switches its behaviour increase or decrease? What about this probability in other behavioural states, do you think it goes up or down? (6/10)

Our study animals live in different habitats & behave differently based on context and goals, so we didn't expect universal patterns in how they switch between behavioural states. However, we found them anyway! (5/10)

Through extensive fieldwork, multi-sensor collars with accelerometers, and machine learning, we arrived at behavioural sequences of multiple animals from our three different species. The behavioural sequences were several days - several weeks long. (4/10)