Alleged early hominin burials may just be accidents, according to a new study.

Every human culture has a special way of laying its dead to rest. Some cremate the remains, some lay them beneath the open sky, and others place them in the ground. Regardless of its form, that final ritual implies an understanding of our own mortality, one of the things that seems to clearly set humans apart from other animals. Along with art and jewelry, deliberate burial is one of the few ways that we can trace the evolution of human thought using the archaeological record.

But it's hard to objectively determine what's a deliberate burial and what's an accidental collection of bones. Now, scientists have attempted to hand off the task to an impartial judge: a machine-learning algorithm. Its analysis indicates that possible signs of burial in other hominins are more likely to be the result of chance.

Grave or not?

Archaeologists are very interested in figuring out when humans started burying our dead. At the moment, the best candidates for the oldest-known burials of modern humans come from Skhul and Qafzeh Caves in Israel, where people appear to have been interred with ochre and other items around 100,000 years ago.

But some researchers have even suggested that the first people to hold something like a funeral may not have been us at all. At several sites in Europe, archaeologists have found Neanderthal fossils in what might be 50,000-year-old graves rather than chance accumulations of bones, although none of the burials has been accepted without debate among archaeologists.

But laying the dead to rest looks very different from one culture to another, so it can be hard to know for sure whether we're looking at a grave or an accidental grouping of bones. "We are, however, talking about different species of humans, so it's certainly possible that culturally mediated corpse disposal manifested differently among them than it does among modern humans," anthropologist Charles Egeland of the University of North Carolina at Greensboro told Ars Technica.

For instance, modern humans often visit graves of friends and loved ones, so archaeologists often find traces of the living, such as stone tools, animal bones, or other debris, near burials. But at some of the sites that have been suggested as possible hominin graves, there's no sign of such activity. That could mean that other hominins just didn't interact with their dead in the same way that we do—even among modern humans, there's a lot of variation among cultures, after all—or it could mean these sites aren't really graves.

"It would be nice to have the bodies interred in some sort of artificial structure that looks like a burial, but, again, this sort of behavior may not have been part of the cultural traditions of these species of humans," said Egeland.

Egeland and his colleagues decided to let a computer try to tell, based on the particular bones found at each site. The team fed information on some possible early hominin burials to a machine-learning algorithm—a program that lets a computer "learn" to identify patterns in complex sets of data using statistical rules—along with data from several other sites where they knew what fate had befallen the bones. Those sites included prehistoric human burials, undisturbed modern corpses, places where prehistoric bones had accumulated naturally, modern human corpses scavenged by wildlife, modern baboons eaten by leopards, and modern baboons that had died of natural causes in a cave.

Testing an extraordinary claim

The sites on the list of possible burials include Skhul Cave, along with a couple of much older and even more hotly debated prospects. In Spain, a limestone cave called Sima de los Huesos ("Pit of the Bones" in Spanish) held the fossil remains of 28 individuals. Based on their size and features, they are probably the remains of either very early Neanderthals or their common ancestor with modern humans, dating to between 300,000 and 600,000 years old.

Separately, the discoverers of Homo naledi say the collection of fossils in South Africa’s remote, nearly inaccessible Dinaledi Chamber, part of the Rising Star cave system, may also be the results of deliberate burial.

Both collections of bones were found deep in the recesses of fairly remote caves, which don't seem like the sort of place one ends up by accident; they seem more like the sort of place one might go to leave the dead far from the world of the living. And nearly all the remains at both sites are from hominins of the same species; they're not mixed with wild animals like you might expect if scavengers had dragged the bones into the caves to snack in relative peace. In fact, at both Sima de los Huesos and Dinaledi Chamber, archaeologists claim that the bones are mostly undamaged by the sharp teeth and powerful jaws of carnivores, which they claim means that the remains were buried there on purpose shortly after death and not disturbed by hungry scavengers.

Egeland's algorithms sorted the sites pretty reliably into distinct groups. One of these was made up of known prehistoric burials and undisturbed modern corpses, which contained higher proportions of more-or-less complete skeletons. Another was made up of the scavenged human corpses, the baboons that died in a cave, and the baboons that were eaten by leopards. The second group had much less even distribution of bones; some parts of the skeleton were noticeably less common than others.

The bones that made the difference seemed to be the bones of the hand and wrist (phalanges, metacarpals, and carpals), the lower arm (radius and ulna), ankles (tarsals), and parts of the leg (femur and fibula, but oddly not the tibia). Egeland and his colleagues say that's probably because these bones are more appealing and accessible to predators, because they're either small—like the hand bones and tarsals—or not too dense to break easily with teeth and jaws.

The algorithm apparently thought the fossil collections from Sima de los Huesos and Dinaledi Chamber looked more like scavenged or naturally accumulated remains than like deliberate burials.

Homo naledi co-discoverer John Hawks of the University of Wisconsin- Madison is skeptical of the algorithm's classifications, however, especially since it groups Skhūl cave with predator kills, not burials. "This study places Skhūl together with known cases of leopard predation. That tells you that the method doesn't work for distinguishing burial from carnivore activity," he told Ars Technica.

Egeland his colleagues say their work doesn't dismiss the idea that hominins buried their dead on purpose in the Pit of Bones or Dinaledi Chamber, but the study also shows there's also not enough evidence to prove the burial claims—and an extraordinary claim requires very solid evidence.

Sorting the bones

What would definite proof of ritual burial look like? According to Egeland, it looks like undisturbed remains.

"I think, at a minimum, there would need to be conclusive evidence that no other biological (e.g., carnivores) or geological (e.g., water movement) agents modified the skeletons after their initial disposal. This way, we could be reasonably sure that the patterns we see are the result of human behavior," he told Ars. While carnivores could scavenge a burial, of course, a completely undisturbed skeleton would offer pretty compelling evidence that it had been placed out of reach of scavengers on purpose.

Especially for Homo naledi, the debate about whether Dinaledi Chamber is a grave or something else centers, in part, on whether the bones there have been chewed on by carnivores. Hawks and others who interpret the cave as a place where hominins deliberately laid their dead to rest, say the chamber is too deep and too hard to access for the bones to be the discarded remains of some predator's meals. Egeland and his colleagues say that's possible, however.

"The inaccessibility of the Dinaledi Chamber, assuming that the current opening was the only access point in antiquity, would have made it difficult, but not impossible, for carnivores to have been involved," said Egeland. "Small carnivores may have been able to access the chamber to scavenge the remains, and larger carnivores, like leopards, could have dragged the corpses down into the chamber."

Thanks to conditions in the cave, the surfaces of many of the bones aren't well-enough preserved to tell for sure. But while Egeland and his colleagues say that relatively poor preservation could make it hard to detect evidence that carnivores tampered with the skeletons, Hawks says it actually supports his team's claim that Homo naledi's remains are nearly untouched by carnivores.

"My guess is that what carnivores do to bone is not so different from what moisture and time do to bone underground—they both eliminate spongy bone parts, which are the ends of long bones, ribs, and vertebrae," Hawks said. "In other words, I wouldn't look at vertebrae and ribs to decide if a bone assemblage resulted from hominin or carnivore activity. You have to look at all the data, and in the Rising Star sites, there's simply no evidence of carnivore activity."

The thing to remember is that, even if carnivores gnawed Homo naledi's bones, it doesn't mean the hominin didn't bury its dead. Graves can be scavenged, after all. It's just that scavenging makes recognizing a deliberate burial tens of thousands of years later even harder.

But...

Of course, those aren't the only explanations.

"Another possibility is that the hominins themselves ventured into the chamber only to become trapped," said Egeland. "We find this to be an attractive hypothesis given that the skeletal representation of the Dinaledi remains matches closely with what we see in an assemblage of modern baboons that died naturally within a cave in South Africa."

And a machine-learning algorithm is only as good as the dataset it's given. Excavations are still underway at the Rising Star cave, which includes Dinaledi Chamber, so it's hard to be sure the dataset Egeland and his team gave their machine-learning algorithm actually reflects the proportion of different skeletal parts at the whole site.

"Most of the Dinaledi bones are still in the site, and we cannot know whether the small area we dug is representative of the whole thing," Hawks told Ars Technica. "We've just excavated a new partial skeleton this year, with some ribs apparently in place. And the Neo skeleton, from the Lesedi Chamber, which [Egeland et al.] didn't include in their study, has most of its vertebrae and many ribs but lacks some of the long bones, which may still be in the site." Egeland and his colleagues acknowledged that point in their paper, as well.

Ultimately, the findings mean that, so far, these sites don't provide a conclusive demonstration that early Neanderthals or Homo naledi buried their dead, but they also don't rule out the possibility.