{"slug": "ant-teams-beat-gravity-based-puzzle-solvers", "title": "Ant teams beat gravity-based puzzle solvers", "summary": "Researchers at Weizmann Institute of Science found that larger groups of longhorn crazy ants (Paratrechina longicornis) outperform smaller teams on complex geometric puzzles, with results published in Journal of the Royal Society Interface. In field experiments using 3D-printed loads and laser-cut mazes, larger ant groups solved more intricate puzzles efficiently, while computer simulations showed that simple puzzles could be solved by gravity-based models but harder ones required ant-like distributed properties.", "body_md": "August 3, 2026\n[\nfeature\n](https://phys.org/editorials/)\n\n# Ant teams beat gravity-based puzzle solvers\n\n##### Ingrid Fadelli\n\nAuthor\n\n##### Gaby Clark\n\nScientific Editor\n\n##### Robert Egan\n\nSenior Editor\n\nAnts live in highly organized colonies and cooperate daily to tackle a wide range of problems. Their striking group behaviors have fascinated biologists for centuries and even inspired the development of various artificial intelligence (AI) systems.\n\nResearchers at Weizmann Institute of Science recently carried out a study aimed at further exploring how ant teams tackle problems of varying complexity. Their findings, [published in Journal of the Royal Society Interface](https://royalsocietypublishing.org/rsif/article/23/240/20250989/482447/Collective-ant-transport-outperforms-gravity-based?searchresult=1), suggest that larger ant groups can solve more complex problems or puzzles than smaller teams.\n\n\"The inspiration for this paper came from observing ants cooperatively transporting large food loads to their nest in the field,\" Ofer Feinerman, senior author of the paper, told Phys.org. \"When the ants reach their nest, they inevitably face a challenge—they need to maneuver the large, often irregularly shaped load through the narrow entrance. This natural geometric puzzle inspired us to try to map out the ants' puzzle-solving capabilities and see how these may scale with the size of the group.\"\n\n## Probing ant cooperation with experiments and simulations\n\nFeinerman and his colleagues wanted to determine whether the size of an ant team influences the complexity of problems that can be tackled collectively. To do this, they performed a series of field experiments involving longhorn crazy ants (Paratrechina longicornis).\n\nThe researchers first created tiny objects using 3D printing technology, precisely engineering their weight and characteristics. They then incubated the 3D-printed loads in cat food overnight, making them attractive to ants. Essentially, they tricked the ants into collecting the tiny objects and carrying them through a maze toward their nest.\n\n\"All we had to do was put the load and a laser-cut maze near the ant nest—the ants would do the rest as we filmed from above,\" Feinerman explained. \"To test different group sizes, we created scaled versions of each maze. We made sure the weight of the load, normalized by the number of ants, stayed constant across the different scales so that we were measuring coordination and cooperation and not simply making the task physically more demanding. Since this is a purely geometric puzzle, scaling it made no difference in the computational difficulty of solving it.\"\n\nThe researchers found that larger ant groups were typically able to tackle more complex puzzles than smaller groups. However, differences between small and large groups only became apparent when problems became sufficiently complex.\n\nIn other words, almost all ant groups performed well on straightforward tasks that required them to carry small loads through a simple maze. When mazes were more intricate and loads heavier, however, larger groups became far more efficient than smaller ones.\n\nAs part of their study, Feinerman and his colleagues also ran a computer simulation in which agents tried to solve the same puzzles tackled by the ants, relying on physical forces or theories. These simulations offered possible explanations for why complex problems were solved more efficiently by larger ant groups.\n\n\"The simulations showed that simple puzzles can be solved by simple gravity-based models: If one were to take the maze, with the load inside, tilt it and shake the whole thing, the load would eventually fall out,\" Feinerman said.\n\n\"However, as the puzzles grew more difficult, we had to supplement the physics simulation with more 'ant-like' properties that have to do with the ants' distributed nature. These can include gravity not acting on the center of mass, gravity acting on a different point on the object and switching every several seconds, or gravity knowing where the next opening in the maze is and acting in that direction.\"\n\n## Next steps for exploring ant problem-solving\n\nInterestingly, when the team supplemented their physics-based models with ant-inspired properties and strategies, they found that simulated agents could solve the puzzles more effectively. These findings further highlight the collective intelligence and adaptability of ant colonies.\n\n\"Would one call the models we developed cognitive, though? Probably not. They are just random and unknowing,\" Feinerman said. \"It is important to note that puzzles still do not match the ants; they match the ants only if we tune puzzle parameters differently for each maze. The ants do not require such tuning and appear to use the same rules to solve all mazes without any outside information.\"\n\nIn the future, the new insights gathered by Feinerman and his colleagues could potentially inspire the development of new AI systems and swarm robotics frameworks. Meanwhile, the researchers plan to continue investigating the complex group behaviors of ants to learn more about their behavioral and neural underpinnings.\n\n\"We are now trying to understand how ants fine-tune their behavior to be able to generalize and solve a huge variety of puzzles,\" Feinerman added. \"Specifically, we want to look at whether this impressive property comes from the brain of individual ants or from properties of the collective that we have yet to understand.\"\n\nWritten for you by our author [Ingrid Fadelli](https://sciencex.com/help/editorial-team/ingrid-fadelli/), edited by [Gaby Clark](https://sciencex.com/help/editorial-team/gaby-clark/), and fact-checked and reviewed by [Robert Egan](https://sciencex.com/help/editorial-team/robert-egan/)—this article is the result of careful human work. We rely on readers like you to keep independent science journalism alive.\nIf this reporting matters to you, please consider a [donation](https://sciencex.com/donate/?utm_source=story&utm_medium=story&utm_campaign=story) (especially monthly). You'll get an **ad-free** account as a thank-you.\n\n###### Publication details\n\nTabea Dreyer et al, Collective ant transport outperforms gravity-based solvers on complex puzzles, *Journal of the Royal Society Interface* (2026). [DOI: 10.1098/rsif.2025.0989](https://dx.doi.org/10.1098/rsif.2025.0989).\n\n**Journal information:**\n[Journal of the Royal Society Interface](https://phys.org/journals/journal-of-the-royal-society-interface/)\n[\n](http://rsif.royalsocietypublishing.org)\n\n© 2026 Science X Network\n\n**Citation**: Ant teams beat gravity-based puzzle solvers (2026, August 3) retrieved 10 August 2026 from https://phys.org/news/2026-08-ant-teams-gravity-based-puzzle.html", "url": "https://wpnews.pro/news/ant-teams-beat-gravity-based-puzzle-solvers", "canonical_source": "https://phys.org/news/2026-08-ant-teams-gravity-based-puzzle.html", "published_at": "2026-08-10 10:16:39+00:00", "updated_at": "2026-08-10 10:41:59.534925+00:00", "lang": "en", "topics": ["artificial-intelligence"], "entities": ["Weizmann Institute of Science", "Ofer Feinerman", "Journal of the Royal Society Interface", "Paratrechina longicornis"], "alternates": {"html": "https://wpnews.pro/news/ant-teams-beat-gravity-based-puzzle-solvers", "markdown": "https://wpnews.pro/news/ant-teams-beat-gravity-based-puzzle-solvers.md", "text": "https://wpnews.pro/news/ant-teams-beat-gravity-based-puzzle-solvers.txt", "jsonld": "https://wpnews.pro/news/ant-teams-beat-gravity-based-puzzle-solvers.jsonld"}}