😅😅😅

Originally shared by Eli Fennell

Machine Learning ‘Translates’ Bat Talk and Hears a Lot of Fighting

Researchers at Tel Aviv University in Israel adapted a Machine Learning algorithm originally used for human voice recognition for a different, but related, purpose: to see if it could tease meaning out of the squeals of bats, or more specifically Egyptian Fruit Bats (Rousettus aegyptiacus).

Until now, most researchers in the field supposed that all bat squeals meant the same thing, usually some variation on a warning call, while others have been eager but, due to the difficulty of the task, unable to challenge this convention.

By feeding the algorithm over 15,000 bat calls, and comparing its results afterwards with video of the behaviors of the bats making the calls, they were able to classify ~60% of bat squeals into four categories, each involving disputes: over food, over sleeping position, in response to unwanted mating advances, and what might be called ‘comfort zone’ disputes when one bat sits too close to another. They also found variations in how the same bats speak to different individuals, as is often observed in highly social mammals.

Given the great density and crowdedness of typical bat populations, this may be likened to the sort of frequent dust ups observed in groups of human roommates living in close quarters with each other. As with all such research,

though, it’s important to remember the Hard Problem of Consciousness: mapping objective behaviors onto subjective states, even within our own species let alone between species.

I can’t help but wonder if dispute may be the wrong word in some cases, rather than something like beseechment, lively discussion, or banter. What appears to be a fight about food, might instead be the bat equivalent of each bat shouting, “Pass the potato salad here, please!”

Then again, they do say that familiarity breeds contempt. Maybe bats, with their extremely crowded and intimate cohabitation patterns, are the ultimate living proof of this principle.

#BlindMeWithScience #AI #Biology

https://www.smithsonianmag.com/smart-news/researchers-translate-bat-talk-and-they-argue-lot-180961564///cdn.embedly.com/widgets/platform.js