The 400 members of the Aka tribe, indigenous to the Congo River basin in Africa, are some of the last remaining hunter-gatherers in the world. Collectively, they’re also pretty into smoking marijuana, which might help them keep their bodies rid of harmful intestinal parasites, according to a recent study published in the American Journal of Human Biology. To the researchers, this may offer a different explanation for human drug use: an unconscious drive to rid ourselves of parasites.

When surveyed, 70 percent of Aka males said that they had recently smoked marijuana, which the researchers confirmed by checking the men’s urine for THCA, a biomarker that indicates recent marijuana consumption (6 percent of Aka women said they had smoked recently, but the percentage was so small that the researchers decided to restrict their study to males). The researchers checked the men’s stool to see how many parasitic worms were living in their bodies. They found that, though 95 percent of the men had parasites, the ones that had recently smoked pot had significantly fewer worms. A year after they were treated with commercial anthelmintics (to kill the worms), the pot-smokers were re-infected much more slowly.

Though the Aka don’t consider marijuana to be medicinal, study author Ed Hagen of the Washington State University suggests that they might be using marijuana to unconsciously keep the worms at bay, as the percent of the population smoking it is higher than the global average and marijuana has been shown to kill toxins in petri dishes. Hagen warns that these conclusions are indirect–though the findings are statistically significant, there is no directly proof that the pot is keeping the parasites at bay and not another factor. But if the study authors are right, their work could start casting doubt on the prevailing theory that recreational drug use is just for pleasure. “In the same way we have a taste for salt, we might have a taste for psychoactive plant toxins, because these things kill parasites,” he said in a statement.