Humans and Neanderthals overlapped geographically tens of thousands of years ago, and the two groups did occasionally mate—but scientists have had limited evidence about who paired with whom and how those encounters played out. In an effort to narrow that uncertainty, researchers analyzed ancient DNA and reported a pattern consistent with more female modern humans mating with male Neanderthals than with male modern humans mating with Neanderthal females. The work was published Thursday in the journal Science, according to the Associated Press.

The new analysis also builds on a broader finding that Neanderthal DNA survives in most modern humans outside sub-Saharan Africa. Scientists have linked those Neanderthal segments to traits including genes that may help people fight some diseases and also genes that can make them more susceptible to others, the report said.

A key clue for the researchers involved a sex-specific mismatch in where Neanderthal DNA appears in the human genome. The Associated Press reported that Neanderthal DNA is not distributed evenly across chromosomes, including a lack of Neanderthal DNA on the human X chromosome compared with levels on non-sex chromosomes. Scientists previously offered hypotheses such as whether the X-linked Neanderthal variants were not beneficial—or were harmful—so they might have been filtered out over time by evolution. Others suggested the difference might reflect how the species intermingled.

To investigate the question, Alexander Platt and colleagues instead looked at Neanderthal genetic data and at the human DNA interspersed during a “mating event” dated by the study to 250,000 years ago. In comparing the genes, Platt and co-authors reported what the report described as a mirror-image pattern: the human “fingerprint” is more visible in Neanderthal X chromosomes at the point of genetic intermixing than would be expected based on what is seen in modern humans today. Platt explained in the report that sex chromosomes are inherited differently, with genetic females having two X chromosomes and genetic males having one X and one Y. On average, two out of every three X chromosomes in a population come from mothers, the report said.

The underlying logic is that if female modern humans more often mated with Neanderthal males, the descendants’ genetic contributions would tend to leave the pattern the researchers found—more human DNA showing up in Neanderthal X chromosomes and less Neanderthal DNA showing up on the human X chromosome. In the Associated Press report, Platt called the explanation for the mirror pattern most likely to be mating behavior, rather than chromosome-level selection alone.

Independent experts who were not involved said the work moved the evidence toward a clearer answer, while still emphasizing that the mating dynamics cannot be fully reconstructed from DNA alone. Joshua Akey, an evolutionary genomics researcher at Princeton University, said in the report that he thought the study “has taken some really important steps in filling missing pieces to the puzzle.” Akey was quoted alongside the study in the Associated Press story.

Population genetics expert Xinjun Zhang, University of Michigan, said researchers may not ever reach a definitive account of how the encounters occurred because scientists cannot travel back in time to observe mating directly. Zhang told the Associated Press that he doesn’t know if researchers will “ever get a definitive answer to how this happened, since we can’t travel back in time,” when commenting on the analysis.

The study also acknowledged limits on what the genetic pattern can prove. Zhang said it’s possible that children of human males and Neanderthal females simply did not survive as well, the Associated Press report said. But Platt argued that the genetic pattern is better explained by how modern humans and Neanderthals interacted—rather than a “strictly Darwinian survival of the fittest” framing—saying it was “really the result of how we interact with each other, and what our culture and society and behavior is like,” according to the report.