At a beach club on the southwestern coast of Cozumel, the space under the kitchen porch had been walled in with wooden planks, and a small opening facing the beach carried a sign identifying it as the raccoons’ home. Staff had left a stainless steel water bowl outside. On 29 January 2026 an adult female pygmy raccoon put a paper towel in that bowl and abandoned the whole business when she could not get it back out.
Her presumed daughters had already proved better at it. Across eight eight-hour observation sessions between 13 and 29 January, two juvenile female siblings repeatedly took paper from a nearby waste bin, carried it to the same water bowl, briefly pushed it under, then worked it against the sand until it packed into a ball. They batted the balls across the ground, chased them, rolled them between their palms while sitting upright, and growled at each other over possession.
The observations were published on 25 August in the journal Wild by Nessie O’Neil and Michelle Szydlowski of Miami University and Noel Anselmo Rivas-Camo of the Cozumel municipality’s ecology office. To their knowledge, they write, the sequence has not been reported in any species of Procyon, the genus that contains the living raccoons.
A beach club, a bin and a water bowl
The ball-making was not what anyone came to Cozumel to study. O’Neil and Rivas-Camo were on the island between January and February 2026 doing fieldwork on how the tourism industry affects the raccoons, using direct observation, site assessment, photography, video and interviews with tourists and staff. The ball-making turned up inside that larger dataset and was pulled out afterwards.
The pygmy raccoon, Procyon pygmaeus, is critically endangered and lives nowhere except Cozumel. Most of what is known about it concerns population density, genetics, home range, activity patterns and diet. Nothing, the authors write, has been reported on object manipulation, play, or behaviour passed between individuals.
The sequence itself is short. A juvenile takes paper from the waste bin, napkins and receipts among it, and carries it to the water bowl. She holds it in her forepaws and briefly submerges it. She lays it on the sandy ground and works it with her forepaws, pressing, flipping and rolling while keeping close manual control. The paper picks up a coat of sand, becomes progressively more compact, and ends up a ball.
Where she got the paper is the part the authors dwell on. Adults at the same site were seen soliciting food from tourists and, when refused, climbing into the organic waste bin first. Only when there was nothing edible in that one did they move to the non-food bin, open discarded containers and lick sauce out of them. That exact sequence was documented three times. The juveniles skipped the order entirely and went straight to the non-food bin.
That apparent selectivity matters because the existing research on raccoon intelligence is mostly about food and possible tool use. Raccoons open puzzle boxes, generate several different solutions to the same task, and get faster at food-extraction problems with practice, and a 2024 study of wild raccoons found individual variation in how they innovate. What the raccoon literature had not described, the authors say, was an animal deliberately modifying an object and then using it for something other than eating.
The sister who watched
On 19 January one juvenile was making a ball while the presumed mother lounged nearby, occasionally eating tortilla chips handed to her by tourists or staff. The second juvenile came over and watched closely. Then she went to the same bin, brought paper back to the water bowl, and tried to work it while it was still dry.
She could not finish. Her sister took the material out of her hands and completed the wetting and rolling herself, while the second juvenile rolled the first sister’s finished ball between her own palms.
The two juveniles were identified as siblings on two grounds. They were small relative to an accompanying adult female with prominent mammary tissue, presumed to be their mother, and the three of them moved as a unit. Both were female, one confirmed by veterinary examination and the other assessed visually by the researchers.
Across the 64 hours the researchers counted 14 ball-making sequences. Eight produced a finished ball and six ended incomplete or unsuccessful, which the paper puts at a success rate of 57.1 percent. The table caption names only the two juveniles, but both the table and the appendix log carry the mother’s single attempt among the 14. The social side of it was documented in at least four observations across three days. The second juvenile’s own first four attempts, on 19 and 22 January, were all unsuccessful or incomplete.
What the paper describes as the presumed mother’s one apparent attempt failed for a mundane reason: she could not retrieve the paper towel she had put in the water. The authors add that her success might have been influenced by her earlier consumption of one ounce of a margarita, which in their words “appeared to lower her dexterity”. It is a throwaway line in the paper and a fairly grim one about what happens to habituated wildlife at a beach bar.
Nothing was seen eaten
The distinction the paper leans on is that none of this was feeding. No ingestion of the paper or of food residue on it was observed. There was one instance of brief mouthing. The authors read it as exploratory rather than as eating, on the grounds that oral exploration is a common component of object play across taxa and particularly in juveniles.
The finished balls were used in play in six separate documented instances. In one the presumed mother joined her offspring. In another she played with a ball on her own. In a third, two juvenile dwarf coatis, Nasua narica nelsoni, joined the two raccoons.
Stone handling in macaques is a parallel the paper draws: a socially transmitted form of object play in which animals strike stones together, pass them between their hands and drop them into holes, with no obvious payoff. That behaviour is treated in the literature as cultural, and the raccoon balls, the authors suggest, may belong in the same conversation.
The comparison site
A behaviour repeated at one location does not by itself show that anything is being transmitted, so the team went looking for a comparison inside their existing dataset. Site 3 is a beach club at the northwestern end of the island, roughly 30 kilometres away. It had juvenile raccoons of similar age, the same access to paper waste and fresh water, and the same high level of habituation to people.
No ball-making was recorded there. Some object play involving souvenirs was seen, but most juvenile play at Site 3 was play-fighting.
The distance is doing real work. Pygmy raccoon average home ranges are roughly 97 to 123 hectares, and genetic work has split the island’s population into a northern and a southern group with limited connectivity between them. On those numbers, the authors call regular movement between the two sites highly unlikely. That is what pushes them towards individual innovation followed by social exposure.
Back at Site 1, on 29 January, about five days after her sister had been removed from the site for rehabilitative care, the second juvenile produced two balls on her own, one from a hamburger wrapper and one from a receipt. She was recorded bringing the wrapper into contact with water, rolling it against sand and a rock, then batting the finished ball across the sand and chasing it. The Discussion describes the same gap as about a week rather than about five days; the shorter figure is the one printed in the Results and the one used here.
The limits of an accident
Nothing about this was planned, and the paper says so repeatedly. There was no ethogram, no experimental design, no provisioning of materials and no attempt to influence the animals. The whole thing is described as a preliminary, qualitative, naturalistic observation, reconstructed afterwards from field notes, photographs and video.
The paper’s own title states the consequence: it reports preliminary evidence of material culture, not material culture. Fourteen attempts by one family at one site is a thin record; the authors say the limited number of attempts precludes statistical assessment of the second juvenile’s improvement, and they say plainly that the observations cannot by themselves establish a stable cultural tradition.
There is also a limit on the comparison. Site 3 was chosen after the fact from data collected for a different purpose. An absence of ball-making there is real, but it is an absence in a dataset that was not looking for the behaviour either.
And the conservation reading cuts both ways. The same beach that supplied the paper, the water bowl and the audience may also have supplied the harm. The first juvenile seen performing the complete sequence later died in rehabilitative care, likely from malnutrition and exposure to human-derived toxic substances. Two of her siblings, one from each of two litters, had been lost in the weeks before the January observations, apparently the same way.
That is why the paper arrives at a question it was never built to answer. Is the habit one family’s invention, one that could vanish with the family that made it, or is it already moving along the beach?