Every mark left on the cave wall is a conversation across time. Thousands of years ago, someone pressed their fingers into a smooth mineral layer and drew squiggly lines, which archaeologists call finger grooves. These gestures continue, but the people behind them remain anonymous.
This anonymity may be starting to fade. In a new paper published in Scientific reportsResearchers have unveiled an artificial intelligence system that analyzes modern finger flutes to test whether the gender of their ancient makers could one day be deduced, providing a rare way to trace identity into the deep past.
“Whether the marks were made by men or women could have real-world implications,” lead researcher Dr. Andrea Jalandoni said in a recent study. press release.
What are finger flutes?
Finger flutes are prehistoric marks made by running fingertips through clay or mineral deposits on the walls, ceilings and floors of limestone caves. They appear at archaeological sites throughout Western Europe and Australia, dating from approximately 60,000 to 12,000 years ago during the Late Middle Paleolithic to Upper Paleolithic.
It is considered by archaeologists to be one of the oldest known examples of symbolic expression, and one of the few art forms created by both Homo sapiens and Neanderthals.
“Finger flutes have the potential to reveal information about age, gender, height, and mark-making preferences,” the authors wrote in the study, describing how their machine learning framework combines physical redundancy with digital modeling.
Previous attempts To determine who made finger flutes relies on measuring the width of the fingers – a method that recent reviews have criticized as unreliable due to variation in cave surfaces and measurement error. The new AI-based approach offers a more objective way to test those ideas.
Read more: AI can translate a 5,000-year-old language and provide time and historical insights
Using artificial intelligence and virtual reality to decode finger movements
The researchers conducted two controlled trials with 96 adult participants. Each person created nine flutes twice, once in a moonmilk clay substitute to simulate cave walls, and once in virtual reality (VR) using Meta Quest 3. The images were used to train two image recognition models designed to detect geometric differences in the signs.
The virtual reality flutes did not yield reliable gender ratings, but the haptic flutes performed much better.
“Under one training condition, the models reached about 84 percent accuracy, and one model achieved a relatively strong discrimination score,” Dr. Gervase Tuxworth of Murdoch University said in the press release.
However, the models captured artifacts of the setting, not features that could be generalized across caves. However, the study shows the existence of a reproducible pipeline linking archaeological methods and artificial intelligence, a crucial step toward more rigorous digital archaeology.
Make analysis of ancient art open and reproducible
The researchers stress that their work is a proof of concept, not a definitive method. While the models performed best on physical clay samples, they also revealed the limitations of current AI methods, learning patterns associated with the experimental setting rather than traits that could apply to ancient caves.
“We have released the code and materials so that others can replicate, critique and extend the experiment,” said Dr Robert Houpt, co-author and information scientist at the Australian Research Center for Human Evolution. “This way the proof of concept becomes a reliable tool.”
The open dataset and the team’s code are available at github. They point out that the framework could also be adapted to analyze other forms of ancient marks, from petroglyphs to tool wear, and expand its use.
Read more: Artificial intelligence revives a 2,000-year-old Roman manuscript burned in the eruption of Mount Vesuvius
condition sources
Our book in discovermagazine.com We use peer-reviewed studies and high-quality sources for our articles, and our editors review for scientific accuracy and editorial standards. Review the sources used below for this article: