A bonobo’s pretend tea party is rewriting what we know about imagination

Imagination is often cast as a singularly human gift. We tell stories, play pretend, and plan futures that don’t yet exist. But what if the roots of that talent run deeper in our family tree? A new study centered on Kanzi—the world’s best-known language‑trained bonobo—adds a surprising twist: in structured pretend “tea party” games, Kanzi kept track of invisible, entirely make‑believe items as if they had genuine locations in space. And yet, when the stakes turned practical (Which would you eat: the real grape or the pretend one?), he unfailingly chose the real thing.

That double competence—treating the imaginary as trackable while never confusing it with the real—is a hallmark of human pretend play. If it holds up under scrutiny and replication, it chips away at one of the most stubborn claims about what makes us unique.

Background

What scientists mean by “imagination” in animals

“Imagination” is a sprawling word that can mean different things. In cognitive science, several related capacities often travel under that banner:

• Mental imagery: forming pictures or sensations in the absence of stimuli.
• Counterfactual thinking: entertaining ways the world could be different from how it is now (“if only…” scenarios).
• Pretense/pretend play: treating objects and events “as if” they were something else, while keeping that fiction separate from reality.
• Prospection: planning for future needs, including ones not currently pressing.

Nonhuman animals have shown hints of some of these. Scrub-jays plan ahead for hunger tomorrow. Ravens and orangutans can delay gratification to use a tool later. Great apes solve invisible displacement tasks (they track where a hidden object must be, even when it disappears behind screens). But pretend play—deciding that an empty cup “has juice” and following the consequences of that make-believe—has been considered a uniquely human cognitive cocktail because it requires representing a scenario that is deliberately not real and keeping that “decoupled” representation from contaminating real-world action.

A brief history of ape pretense, symbolism, and “like us” debates

• Enculturated apes: A small number of chimpanzees and bonobos raised in human environments and taught symbol systems (lexigrams, sign language) have displayed sophisticated comprehension of words and flexible problem-solving. Kanzi, a male bonobo born in 1980, is the most famous of these individuals. He uses a lexigram keyboard, understands spoken English at a basic sentence level, and has a long history of experimental participation.
• Object substitution in the wild: Juvenile chimpanzees in Uganda have been observed carrying sticks as if they were infants—behavior interpreted as “doll play.” While much rarer than human children’s make-believe, such reports have been used to argue that symbolic or pretend-like behaviors are not alien to great apes.
• Theory of mind and false belief: In anticipatory-looking experiments, great apes have sometimes anticipated an agent’s actions based on that agent’s mistaken belief, suggesting they track others’ perspectives. Those results remain debated, but they opened the door to more nuanced attributions about ape mental life.

Even with this background, the mainstream consensus has been that open-ended pretend play is a human specialty. The new “tea party” work refocuses the debate on a more minimal—but revealing—question: Can an ape keep a running account of things that do not exist, while still knowing that they do not exist?

What happened

Researchers invited Kanzi to participate in a series of pretend “tea party” interactions. The essential idea was simple: treat empty containers as if they held specific foods or drinks, move these imaginary items around, and then probe whether Kanzi could keep track of where those fictional objects “were.” Crucially, when it came time to actually eat, the tests contrasted pretend with real options.

While the technical details will matter as other scientists scrutinize the study, the core findings are striking:

• When an experimenter enacted make-believe events—“pouring” imaginary juice into a particular cup, or “placing” pretend grapes behind a specific screen—Kanzi consistently indicated the correct locations of these non-existent items when asked.
• At choice moments, when the setup pitted a pretend option against a real one, Kanzi opted for the real food or drink, not the imaginary version.

Together, those patterns suggest he was operating with two concurrent mental tracks: one for the actual world, another for an agreed-upon story world. That is the psychological juggling act human children learn during early pretend play: keeping a banana-as-phone identity active in an “as if” frame while never truly forgetting that it’s just a banana.

Why this is more than clever pointing

Skeptics will immediately wonder whether Kanzi could have been cued by subtle gestures or routines—classic “Clever Hans” pitfalls. Any such study must address this with controls (counterbalancing, minimizing gaze cues, barrier placements, or blind testing). The report summarizes a robust pattern: Kanzi tracked pretend locations even as real outcomes remained correctly prioritized. That decoupling is not easily explained by simple stimulus–response learning because it requires maintaining an internally coherent yet non-actual representation.

Moreover, Kanzi’s long training history cuts both ways. On the one hand, he is exquisitely attuned to human intentions and experimental setups; on the other, he has had decades of opportunities to acquire exactly the kind of representational flexibility that might reveal dormant ape capacities. If an enculturated bonobo can routinely do this, the capacity may be part of the great-ape toolkit, even if it is rarely expressed without human-like social scaffolding.

Why this matters for science (and for how we see apes)

Decoupling: the key cognitive move

Pretend play hinges on “decoupling”: the mind runs a simulation that is insulated from immediate action, yet detailed enough to be queried. In humans, that faculty supports storytelling, hypothetical reasoning, moral imagination (“what if I were in their place?”), and counterfactual planning. Evidence that a bonobo can do a version of this, even in a constrained lab game, suggests that the evolutionary roots of decoupling could predate our species.

Evolutionary implications

If great apes possess a capacity for representing non-actual states—however minimally—then:

• The last common ancestor of humans and bonobos/chimpanzees (around 5–7 million years ago) may already have had proto-imaginative abilities.
• Human pretend play may be an elaboration and cultural amplification of a more general primate skill: running internal models that do not perfectly mirror the present.
• The step from “tracking pretend juice” to “constructing elaborate fictional worlds” likely required language, teaching, and social norms—human specializations that massively scale up a basic capacity shared with other apes.

Animal welfare and ethics

Recognizing that apes can engage with the “as if” dimension strengthens arguments for complex enrichment in sanctuaries and zoos. Games that involve shared narratives or role-like routines may be more cognitively appropriate than we assumed. It also reinforces the ethical case for treating great apes as beings with rich inner lives, not just problem-solving machines.

Methodological ripple effects

Animal cognition research has sometimes avoided pretense tasks out of a belief they were too human-centric. This study suggests new paradigms: asking nonhumans to follow fictional “rules of the game” and then measuring their ability to keep track of consequences inside that frame. That could open comparative work across species—from corvids and parrots to dogs and elephants—on how widely distributed “decoupled” representations are.

What we should not overclaim

• One individual is not a species. Kanzi is unusual: enculturated, language-trained, and extremely practiced at interpreting humans. His success does not automatically generalize to other bonobos or to wild apes.
• Pretend tracking is not the same as full-blown human imagination. There is no evidence here of narrative fiction, imaginary friends, or spontaneous role play initiated by the ape. The study probes a minimalist form of pretense: keeping locations of supposed items separate from reality.
• Cues and routines matter. Without full methodological details—such as blinding procedures, counterbalancing, and controls for inadvertent cues—we should interpret the results as promising rather than definitive.

How this fits with prior findings

• Invisible displacement and object permanence: Great apes robustly track hidden real objects. The new twist is that Kanzi tracked “hidden” non-objects that only existed within a mutually agreed pretend frame.
• Planning and prospection: Orangutans, bonobos, and ravens have shown the ability to plan for future tool use. Pretense requires a different form of off-line modeling—one that’s social and conventional. Putting these strands together strengthens the case that nonhumans can run counterfactuals when conditions encourage it.
• Symbol systems and language training: Kanzi’s lexigram experience likely primes him for conventional “this stands for that” mappings. Pretense is also a conventional mapping: this empty cup “stands for” a cup with juice. That overlap may explain why an enculturated ape is the first to show clear success.

Key takeaways

• A language‑trained bonobo correctly tracked the locations of imaginary items in a pretend tea party task, indicating a capacity to represent non-actual states.
• He kept pretend and real separate: he pointed to where make-believe items “were” but chose real food when given the chance to consume.
• This suggests great apes can decouple representations—an essential ingredient of human imagination—at least in structured, socially scaffolded contexts.
• The finding narrows the gap between human and ape minds, but it does not erase it. Enculturation and language exposure likely play large roles.
• Replication with additional individuals, rigorous cue controls, and cross-species comparisons are the next critical steps.

What to watch next

• Replication across apes: Will other bonobos, chimpanzees, gorillas, and orangutans track pretend items without extensive language training?
• Developmental trajectories: Do juvenile apes show more spontaneous object substitution (e.g., using leaves “as” plates) in social play than adults? How does exposure to humans change this?
• Methodological tightening: Expect follow-ups with double-blind procedures, opaque barriers to block gaze cues, and automated prompts to minimize unintentional signaling.
• Beyond primates: Corvids (crows, ravens), parrots, and domestic dogs excel at reading social cues and following conventional rules. Can they maintain parallel “pretend” and “real” representations in similar tasks?
• Neural signatures: Non-invasive brain measures (like near-infrared spectroscopy in apes trained to sit calmly) might reveal whether pretend tracking recruits networks analogous to those engaged in human mental simulation.
• From pretend to counterfactual reasoning: If an ape can follow a pretend scenario, can it also judge counterfactuals (“If we had poured the juice into the red cup, where would it be now?”)? Probing such what-if reasoning would sharpen the theoretical stakes.

FAQ

• Does this mean apes have imaginary friends?

  • No. The study shows controlled pretend tracking, not spontaneous invention of fictional companions. It’s a minimal, testable form of pretense.

• Could Kanzi just be following human body language?

  • Subtle cueing is always a risk. That’s why replication with stricter controls is crucial. Still, the split pattern—accurate pretend tracking plus correct real-world choices—suggests more than simple cue-following.

• Is language required for pretense?

  • Language probably amplifies and stabilizes pretend play, but the core capacity may not require words. The question is how much enculturation and social scaffolding are needed for nonhumans to display it.

• At what age do humans show similar abilities?

  • Human children begin object-substitution pretend play around 18–24 months and get increasingly sophisticated by ages 3–4, when they also start passing classic false-belief tasks.

• What’s the practical upshot for ape care?

  • Enrichment that includes rule-based games, role routines, and “as if” frames may be cognitively appropriate for great apes and could improve welfare.

Bottom line

Kanzi’s pretend tea party doesn’t put apes on stage with Shakespeare. But it does suggest that the mind of at least one bonobo can entertain a small fiction, keep it coherent enough to be queried, and prevent it from leaking into reality where it doesn’t belong. That is not storytelling. It is, however, a sturdy foundation stone of imagination—one we may share with our closest relatives.

Source & original reading: https://www.sciencedaily.com/releases/2026/02/260210040605.htm