The bonobo, already famous for understanding spoken English, was being offered something stranger than treats: imaginary juice and invisible grapes. What he did next is now reshaping what scientists think separates human minds from those of other apes.
Another line between humans and apes just blurred
For decades, psychologists treated pretend play as a hallmark of human childhood. Toddlers hosting tea parties with invisible tea and biscuits were seen as doing something apes could not. A new study in the journal Science challenges that assumption.
Researchers have shown that Kanzi, a language‑trained bonobo, could track the location of entirely imagined food and drink in controlled experiments. He did not see any real liquid change containers. He never tasted the pretend grape. Yet he behaved as if those things existed in specific places.
Kanzi followed shared make‑believe scenarios created by humans, keeping track of objects that existed only in imagination.
The work suggests that at least some great apes can represent pretend objects in their minds — a skill long claimed as uniquely human and central to how children develop language, stories and social rules.
Who was Kanzi, the ‘talking’ bonobo?
Kanzi was no ordinary ape. Raised in research settings in the US, he learned to understand hundreds of spoken English words and to communicate using lexigrams — symbols on a keyboard that stand for words.
- Species: bonobo (Pan paniscus)
- Famous for: understanding spoken English and using lexigrams
- Age at death: 44, in March 2025
- Research centre: Ape Initiative, Iowa
Because he could reliably follow spoken instructions, Kanzi offered a rare opportunity: scientists could run tests with verbal directions very similar to those used with young children. That allowed a much cleaner comparison between human and ape abilities.
From anecdote to experiment
Researchers have long reported hints of pretend play in apes. Wild chimpanzees have been seen “wearing” discarded human items, such as using a leaf cushion as a hat. Captive bonobos have acted as if pictures of fruit were edible, miming picking and eating from photographs.
Those cases were intriguing, but they left too much room for doubt. Were the apes truly pretending, or did they partly believe the objects were real? To move beyond speculation, comparative psychologist Amalia Bastos and cognitive scientist Christopher Krupenye designed a set of experiments to test pretend play more rigorously.
➡️ Italian torta nua: the recipe for this cloud cake with custard, a pure delight to share
➡️ This melting parsnip–potato gratin even reconciles kids with winter vegetables
➡️ Grim photo captures polar bear mom and cubs resting in mud in summer heat
➡️ This milk-braised pork roast is so tender you can cut it with a spoon
The central question: can a nonhuman ape knowingly track an object that both ape and human understand is purely imaginary?
How do you serve imaginary juice to a bonobo?
Step 1: Teaching the real‑juice game
Before any make‑believe could happen, Kanzi had to show he understood a simple reality-based task. He was presented with two transparent bottles: one filled with real juice, one empty.
When asked where the juice was, he pointed to the correct bottle every time, across 18 training trials, and received a sip as a reward. This confirmed that he understood the basic question and cared about the outcome.
Step 2: The invisible liquid test
In the main experiment, the bottles disappeared. Instead, experimenters placed two clear, empty cups in front of Kanzi. Then came the twist: they pretended to pour juice from an empty jug into both cups.
After that, they mimed pouring “juice” from one of the cups back into the jug. Nothing physical changed in front of him; the cups stayed empty.
At the end of the routine, a researcher asked Kanzi to point to the cup that had the juice. He received no reward and no feedback, so he could not learn across trials which answer was “right”.
Kanzi chose the correct cup 68% of the time — well above chance. This suggested he was mentally tracking the pretend liquid as it “moved” between containers, much like a human toddler in a tea party game.
| Experiment | Pretend object | Correct choices | What it tested |
|---|---|---|---|
| 1 | Juice (imaginary) | 68% | Tracking liquid that never existed |
| 2 | Real vs imagined juice | 77.8% for real juice | Can he tell real from pretend? |
| 3 | Grape (imaginary) | 68.9% | Generalising pretend tracking to solid food |
Separating belief from make‑believe
One sceptical response was obvious: maybe Kanzi thought there really was juice in the pretend cup. To address that, the team ran a second, more subtle test.
They placed one cup with real juice and one empty cup in front of him. Then they pretended to pour juice into the empty cup, while holding the jug motionless over the full cup.
If Kanzi believed the act of pretending created real juice in the empty cup, he should choose both cups equally often. Instead, he selected the genuinely full cup almost 78% of the time.
Kanzi treated the two situations differently: he acted as though imaginary juice existed for the sake of the game, while still preferring the only cup that held real liquid.
This pattern suggests he could distinguish between reality and pretense, yet still track the imagined object when that was what the humans were asking about.
From juice to grapes: was it just a fluke?
Bastos remained cautious. A single successful experiment could be written off as a statistical accident or some unnoticed cue in the setup. So the team repeated the whole design with a new imaginary target: a grape.
Again, they mimed moving the grape between cups. Again, Kanzi had to indicate where the non‑existent fruit ended up. He did so correctly in just under 69% of trials, a very similar performance to the juice task.
For the researchers, the replication with a different pretend object strengthened the case that Kanzi was not simply guessing. He seemed able to mentally represent absent, fictional items and follow their “journeys” through an invented scenario.
What this suggests about ape minds
Outside experts are treating the study as a significant advance. The work backs up years of scattered observations suggesting that great apes can imagine, not just perceive.
The results also nudge back the timeline for when certain mental abilities likely evolved. If a bonobo can join humans in shared pretense, that capacity may have been present in the last common ancestor of humans, chimpanzees and bonobos more than 6 million years ago.
At the same time, the findings come with clear limits. Only one individual was tested, and Kanzi did not invent the games himself. The humans framed the context, introduced the pretend objects and guided the play.
The study shows that a bonobo can understand and participate in human-led pretend play; it does not show that bonobos spontaneously stage elaborate fantasies like human toddlers.
Why pretend play matters for language and culture
Pretend play is not just a cute stage of childhood. For human children, it links closely to language learning, storytelling and social understanding. When a child “drinks” from an empty cup, they are practicing an early form of fiction: acting as if something were true when everyone knows it is not.
Cognitive scientists see this as part of a broader capacity called “mental representation” — the ability to hold ideas in mind that do not match the here-and-now. That includes remembering the past, planning the future and imagining impossible events.
If great apes can share pretense, even on a smaller scale, it suggests that the roots of human storytelling, religion and collective imagination may be deeper in our family tree than once thought.
Key terms that help make sense of the study
Pretend play: Acting as if something is real when all participants understand that it is not. Examples include imaginary tea parties, toy doctors’ kits or capes that “turn” children into superheroes.
Shared pretense: When two or more individuals take part in the same make‑believe scenario, each tracking the same set of fictional rules. Kanzi’s experiments required this: he and the humans had to coordinate around where the imaginary juice was supposed to be.
Mental representation: The mind’s ability to hold and manipulate information about things that are absent, abstract or hypothetical. This includes fictional objects like Kanzi’s invisible grape.
What might future bonobo ‘games’ look like?
The obvious next step is to test more apes. Chimpanzees, gorillas and additional bonobos in different facilities could run through similar tasks, ideally without language training, to see how widespread the ability is.
Researchers are also interested in whether apes can ever initiate pretense themselves. For example, would a bonobo start offering an empty cup to another as if it contained drink, or pretend to feed a toy? That kind of spontaneous make‑believe would bring ape behaviour closer to what is seen in human two‑year‑olds.
There are practical angles too. Better understanding of ape cognition can inform enrichment in zoos and sanctuaries. Games that tap into imagination, not just problem‑solving for food, could improve welfare by providing more complex social and mental challenges.
For anyone working with young children, the study also highlights how powerful pretend games are as a window into the mind. When a child insists that an empty plate is full of cake, they are not just playing; they are wielding the same mental tools that, in a lab in Iowa, once let a bonobo point confidently at a cup of juice that never existed at all.
