animals can perform any behavior in its given repertoire at any given time. representing all behaviors in a low dimensional space, results in a distributed probability density map.
there is low probability of an animal being in any one behavior (based on postural data) at any given time. This wide distribution indicates high entropy.
If an animal is only performing one behavior, the distribution is very narrow, with a high probability of being in that one low dimensional representation (a peak or hotspot in the map). This indicates that the behavioral entropy is very small.
fig 1b,c
2 dimensional representation of behaviors and human labeled watershed map of regions in the space
Can this same or similar framework transfer to neural imaging data?
- yes— see Calaim.etal2022, Schaffer.etal2023