Imagination has long been treated as a hard cognitive line separating humans from the rest of the animal kingdom. Pretend tea parties, imaginary swords, and make-believe worlds are often held up as quintessentially human achievements—evidence that our minds can detach from reality and explore alternatives that exist only in fantasy.
However, a new peer-reviewed study published in Science suggests that this line may be far blurrier than previously assumed.
In a series of controlled experiments, researchers found that Kanzi, a language-trained bonobo, could reliably track the location of pretend objects—such as imaginary juice poured between empty cups—despite knowing full well that the objects were not real.
The findings provide some of the strongest experimental evidence yet that at least one nonhuman primate can form what cognitive scientists call “secondary representations”: mental models of imagined states that exist alongside, but separate from, reality itself.
“It really is game-changing that their mental lives go beyond the here and now,” co-author and professor in the Department of Psychological and Brain Sciences at Johns Hopkins, Dr. Christopher Krupenye, said in a press release. “Imagination has long been seen as a critical element of what it is to be human, but the idea that it may not be exclusive to our species is really transformative.”
The ability to imagine sits at the foundation of many advanced cognitive skills. Secondary representations allow humans to engage in pretense, imagine possible futures, reason about counterfactuals, and even attribute beliefs to others. Until now, most scientists have argued that these capacities—and especially the ability to keep imagined states quarantined from real ones—are uniquely human.
However, this new study directly challenges that assumption.
The star of this discovery is Kanzi, a bonobo who was 43 years old at the time of the experiments—a close relative of chimpanzees and one of humanity’s nearest living evolutionary cousins. Bonobos split from the human lineage roughly six to nine million years ago and are known for their high intelligence, complex social behavior, and sophisticated communication skills, though they remain far less familiar to the public than chimps or gorillas.
Kanzi has spent decades working with scientists and is internationally known for his ability to understand spoken English and communicate using a lexigram keyboard. While he has long figured into debates about language and cognition in nonhuman apes, the new experiments move beyond anecdotes.
Researchers designed a task inspired by classic developmental psychology experiments used with human toddlers. In one version, an experimenter placed two empty, transparent cups on a table and pretended to pour juice from an empty pitcher into both cups.
Then, in full view of Kanzi, the experimenter pretended to pour the contents of one cup back into the pitcher. Finally, Kanzi was asked a simple question: “Where’s the juice?”
From a purely physical standpoint, the question makes no sense. Both cups are empty, and Kanzi can see that they are empty. If he had relied solely on a representation of reality, his choices should have been random. Yet, across dozens of unrewarded trials, Kanzi consistently pointed to the cup that, within the pretend scenario, still “contained” the imaginary juice.
“This is exactly what Kanzi did,” the researchers write, noting that he selected the correct cup in 68% of probe trials, well above chance, despite receiving no reward for his choice.
Researchers anticipated and ruled out simpler explanations that have complicated earlier claims about animals engaging in “make-believe.” One possibility was that Kanzi mistakenly believed the pretend juice was real.
To test this, the team ran a control experiment in which Kanzi had to choose between a cup containing real juice and a cup that had only been pretend-filled. If Kanzi were confusing pretend with reality, he should have chosen randomly. Instead, he overwhelmingly selected the cup with real juice.
“Kanzi selected the cup containing real juice above chance, in 14 out of 18 trials (77.8% correct; BF = 4.51), demonstrating that he could distinguish real and pretend juice and that his success in the original experiment did not reflect a mistaken belief that the manipulations involved real juice,” the researchers write.
Another alternative explanation was that Kanzi was simply following perceptual cues—such as choosing the cup most recently handled by the experimenter. However, researchers deliberately designed the task so that the most manipulated cup was actually the wrong answer. If stimulus enhancement were driving Kanzi’s behavior, he should have consistently failed. He did the opposite.
In a third experiment, the team replaced juice with grapes and cups with jars, conceptually replicating the task in a new context. Once again, Kanzi tracked the location of a pretend object that did not exist in the real world, performing above chance from the very first trial.
Taken together, the results strongly suggest that Kanzi was not guessing, not imitating, and not confusing fiction with reality. Instead, he appeared to be maintaining two parallel models of the situation at once. One, a primary representation of the real world, in which all containers were empty, and a secondary representation of a pretend world, in which an imaginary object persisted and could be moved.
“It’s extremely striking and very exciting that the data seem to suggest that apes, in their minds, can conceive of things that are not there,” co-author and lecturer at Scotland’s University of St. Andrews, Dr. Amalia Bastos, said. “Kanzi is able to generate an idea of this pretend object and at the same time know it’s not real.”
Secondary representations are widely regarded as a cognitive prerequisite for mental time travel, planning, causal reasoning, and theory of mind—the ability to reason about what others believe, even when those beliefs are false.
Demonstrating that a bonobo can generate such representations in a pretense context increases the likelihood that similar mechanisms may underlie other sophisticated behaviors previously documented in apes, such as belief tracking and future-oriented planning.
The findings also reopen a long-running evolutionary question: when did the capacity for imagination emerge?
Bonobos and humans last shared a common ancestor roughly 6 to 9 million years ago. If Kanzi’s performance reflects a capacity that is not uniquely the product of human culture or language, then the cognitive roots of imagination may stretch far deeper into the primate lineage than once thought.
At the same time, the authors are careful not to overgeneralize. Kanzi was an enculturated ape with extensive experience interacting with humans and using symbolic systems. It remains an open question whether wild bonobos—or other nonhuman animals—would show similar abilities without such a background.
The researchers outline several possibilities: enculturation may simply make existing capacities easier to detect; symbolic training may strengthen abilities that are already present; or, more provocatively, symbolic training may fundamentally reshape cognitive architecture.
Unfortunately, follow-up testing with Kanzi is no longer possible. He passed away in March 2025 at the age of 44, bringing to a close a singular scientific career spanning decades and helping shape modern debates about ape cognition.
His death means researchers cannot directly probe how stable or generalizable his abilities were over time, or how they might have responded to new experimental designs. As a result, the findings stand as compelling but necessarily limited evidence drawn from a uniquely experienced individual—one whose long history of human interaction and symbolic communication makes both his capabilities and their broader implications especially challenging to disentangle.
Researchers say future studies will need to test non-enculturated animals and explore alternative experimental designs, such as violation-of-expectation paradigms, to determine how widespread these abilities really are.
For now, the findings offer striking evidence that imagination, long considered a defining human trait, may not be ours alone. In at least one bonobo mind, the researchers suggest, the ability to hold fantasy alongside reality is measurable, repeatable, and experimentally robust.
“Imagination is one of those things that in humans gives us a rich mental life. And if some roots of imagination are shared with apes, that should make people question their assumption that other animals are just living robotic lifestyles constrained to the present,” Dr. Krupenye said. “We should be compelled by these findings to care for these creatures with rich and beautiful minds and ensure they continue to exist.”
Tim McMillan is a retired law enforcement executive, investigative reporter and co-founder of The Debrief. His writing typically focuses on defense, national security, the Intelligence Community and topics related to psychology. You can follow Tim on Twitter: @LtTimMcMillan. Tim can be reached by email: tim@thedebrief.org or through encrypted email: LtTimMcMillan@protonmail.com