Kanzi the bonobo could play pretend — a trait thought unique to humans

For the first time, scientists have experimentally shown bonobos (Pan paniscus), our closest living relatives along with chimpanzees (Pan troglodytes), can engage in pretend play — something previously assumed to be unique to humans.

Just like 2-year-old children can, Kanzi, a unique bonobo who could understand English, kept track of imaginary juice and grapes during pretend tea parties, according to a study published Thursday (Feb. 5) in the journal Science.

Although Kanzi, who died in March 2025, did not initiate the pretend play, his ability to follow along shows he had some of the mental building blocks needed to imagine pretend objects and scenarios.

"We were really in awe at this finding," study co-author Christopher Krupenye, an assistant professor of psychological and brain sciences at Johns Hopkins University, told Live Science. "What we're seeing in this case is that … something that seems to be fundamentally human and is emerging early in our human development is also shared with our closest relatives," he said.

This suggests the human capacity to imagine objects that aren't really there could have evolved before humans and bonobos split from our last common ancestor over 6 million years ago, Krupenye said.

Imagined realities

Previous anecdotal evidence has hinted that captive and wild great apes engage in pretend play. For example, a wild 3-year-old chimpanzee in Guinea was observed playing with a discarded human-made leaf cushion by placing it on his head. A captive bonobo also "picked" and "ate" blueberries from a photograph of real blueberries.

But because the anecdotal examples could be explained by alternative explanations, such as the apes believing the pretend objects were actually real, Krupenye and his colleague Amalia Bastos, a comparative psychologist at the University of St Andrews in the U.K., wanted to bring the question of "can animals really pretend?" into a controlled, experimental setting.

Because Kanzi could understand and respond to English, he was the obvious first animal to study, Krupenye said.

First, Kanzi was trained to point to the vessel containing juice. He was shown two see-through bottles, one containing juice and the other empty, and he was asked to point out where the juice was. If he answered correctly, he was rewarded with some of the juice. Kanzi got a perfect score over the 18 repeats of this training phase.

In the test trials, an experimenter placed two transparent empty cups side-by-side on a table in front of Kanzi. Next, they pretended to pour juice from an empty jug into each cup, then poured the pretend juice from one of the cups back into the jug. Kanzi was then asked to point to the location of the cup with the juice, but he was never told if he was correct, and wasn't rewarded.

Kanzi correctly identified the location of the pretend juice 68% of the time, which suggested he could keep track of the imaginary liquid.

But the possibility remained that he simply thought the empty cup actually contained real juice. To check if this was the case, the team ran a second experiment where they placed a juice-filled cup and an empty cup on a table. They pretended to pour juice into the empty cup and then held the empty jug over the full cup without doing the pouring motion.

Krupenye said that if Kanzi really thought there was juice in both cups, he would have chosen them at equal frequency. But, when asked which cup he wanted, Kanzi selected the cup containing real juice 77.8% of the time, suggesting he could clearly distinguish real from imaginary juice.

"That sort of gave us confidence that we were really looking at some ability to track imaginary or pretend objects," Krupenye said.

Bastos said she was still a little skeptical at this point — Kanzi's ability to point out where the pretend juice was could have been a fluke. So the team repeated the same procedure but with a pretend grape. Kanzi correctly identified the location of the imaginary grape in 68.9% of trials.

"By the time we finished experiment three, I was very confident that what we saw was what we saw," Bastos said.

The research is limited as only one bonobo was tested, but is nonetheless the first clear evidence that great apes can engage in pretend play, Laura Simone Lewis, an evolutionary anthropologist and psychologist at the University of California, Santa Barbara, who was not involved in the research, told Live Science in an email.

"This is a huge development for our field, because it provides direct evidence to support the anecdotal reports from the wild that our great ape cousins can use their imaginations for all sorts of activities, including pretend play," she said.

This research demonstrates that Kanzi could understand shared pretense created by humans, but not that he could produce pretend scenarios himself.

"I think it would be a big leap to say that, because of this, in some sense we're seeing something comparable to what we see in 2-year-old children, where you typically routinely see pretense production including things like drinking from empty cups and so forth," Paul Harris, a psychologist at Harvard University who was not involved in the study, told Live Science.

Krupenye and Bastos hope that pretend play can now be explored in other great apes. "If the anecdotes are right, it should be the case that other apes also share this capacity," Krupenye said.

Primates Quiz: What do you know about our closest relatives?