Scientists have developed an artificial intelligence system, CapuchinAI, that uses facial recognition and real-time touchscreen testing to automate cognitive studies of capuchin monkeys in the wild, according to a July 29 announcement. The proof-of-concept for this method was published in the American Journal of Primatology by researchers from Emory University and Georgia Institute of Technology.
CapuchinAI integrates a compact, battery-powered computing system into a field-research platform. The system identifies approaching monkeys, presents individual learning tasks on a touchscreen, and automatically delivers food rewards when tasks are completed correctly. Field tests conducted in Costa Rica's Taboga Forest Reserve showed that CapuchinAI could identify individual capuchins with 97% accuracy after training on images and videos. Wild capuchins quickly adapted to the device and learned the association between touchscreens and rewards.
Federico Sánchez Vargas, first author of the paper and an Emory PhD student in anthropology, said, "This project builds on the legacy of Frans de Waal." De Waal pioneered animal cognition studies at Emory's Living Links Center for the Advanced Study of Ape and Human Evolution before passing away in 2024. Sánchez Vargas added, "Our AI method allows us to more deeply understand individuals that we already have data on through field observation. We can now automate cognitive testing of them and quantify the findings... Studying individuals in their natural environments... lets us learn how environmental influences shaped them."
The research team included Jacob Abernethy from Georgia Tech; Sai Rakshith Potluri; as well as contributions from undergraduate students who helped create training datasets using GoPro imagery tagged with monkey identities. They used open-source software YOLO (You Only Look Once) for facial recognition model development running on Raspberry Pi computers powered by lightweight batteries.
The team constructed weatherproof platforms using common materials to house webcams, touchscreens, food dispensers, and computer components—testing their prototype at Capuchinos de Taboga research facility where local capuchins interacted with it enthusiastically. Individual differences among monkeys were observed during pilot trials: some learned rapidly while others required more time or watched peers before participating.
Researchers are updating their facial-recognition model based on new video data from pilot sessions involving sixteen monkeys. Future experiments will assess various domains such as learning ability, impulse control, flexibility, memory skills—and link these findings with decades-long observational records about each animal’s life history at Taboga Forest Reserve.
