For nearly two decades, a rotating cast of chimpanzees, bonobos, gorillas, and orangutans at Leipzig Zoo sat through hundreds of cognitive experiments: hiding games that tested whether they could track objects out of sight, puzzles that required pulling a rope in tandem with a partner, tasks that asked them to read a human experimenter’s intentions from a glance. The raw data from all of those sessions was locked inside individual lab computers, parceled out across 150 separate journal articles. In April 2026, that changed. A team led by comparative psychologist Alejandro Sánchez-Amaro of the University of Stirling released the EVApeCognition dataset, an open-access archive of 262 experimental datasets covering 18 years of testing on more than 80 great apes, all available for free download by any researcher on the planet.
“This is the first time the scientific community has access to such a comprehensive collection of great ape cognition data,” Sánchez-Amaro said in a statement accompanying the release. The dataset, described in a peer-reviewed Data Descriptor published in Scientific Data, compiles trial-level records from experiments conducted between 2004 and 2021 at the Wolfgang Köhler Primate Research Center, one of the world’s most prolific facilities for studying nonhuman intelligence.
Why a single archive matters
Ape cognition research has long suffered from a structural problem: tiny sample sizes. A typical published study might test eight or ten animals, and the raw data from each trial almost never accompanies the journal article. Other scientists who want to reanalyze the results or combine them with their own work have had to email authors individually and hope for a response. The result, as Sánchez-Amaro and his collaborators have argued, is that the field has struggled to draw reliable conclusions about species-level cognitive abilities or track how individual apes develop over time.
The EVApeCognition dataset attacks both problems at once. Because the same apes at Leipzig participated in experiments year after year, the archive functions as a longitudinal record. A researcher can now trace how a single chimpanzee performed on memory tasks at age seven, then again at age twelve, then on a causal reasoning task at age fifteen. That kind of within-individual tracking across cognitive domains has been essentially impossible until now.
The project drew contributions from what the team describes as nearly 100 collaborating institutions across Europe, North America, and Asia. Each experiment is stored as a CSV file paired with YAML metadata that documents the design, coding scheme, and outcome measures. Ape profile files and glossaries allow outside researchers to reconstruct who was tested, under what conditions, and what each variable means. The full archive is hosted on Zenodo, with a public GitHub repository for ongoing documentation updates.
What the data can and cannot tell us
It is worth being precise about what was actually released. The Scientific Data paper is a Data Descriptor: it documents what was collected and how others can reuse it. It does not advance a specific evolutionary argument or claim that humans differ from other apes in any particular cognitive module. No published analysis has yet reported findings from the full aggregated archive. The institutional press materials from the University of Stirling frame the release as offering “new insights on human intelligence evolution,” but that language reflects anticipated impact, not results already in hand.
The dataset also carries inherent limitations that researchers will need to grapple with. Every data point comes from captive apes housed at a single facility with specific enrichment regimes, social groupings, and daily human contact. Whether cognitive patterns observed at Leipzig Zoo generalize to wild populations, or even to apes in other zoos, is an open question. Captive animals may develop skills tailored to experimental settings that have little parallel in the forest.
There is also the question of what counts as “intelligence.” The 262 experiments cover classic domains: object permanence, causal reasoning, social learning, cooperation, and tool use, among others. But they inevitably reflect the theoretical priorities of the researchers who designed them over two decades. Emotional regulation, long-term planning in naturalistic contexts, and culturally transmitted tool traditions may be underrepresented. The archive is a rich but partial window onto what these animals can do, not a complete inventory of great ape minds.
The project team’s claim that this is the “world’s largest” great ape cognition collection is plausible given the scale of work concentrated at Leipzig, but no independent comparison against every other existing dataset has been published. Collaborative efforts like the ManyPrimates project have also been pooling primate data across sites, though with a different structure and scope. Readers should anchor their assessment in the documented numbers: 262 datasets, 150 publications, more than 80 individual apes, 18 years.
Why the timing matters for AI and cognition research
The release arrives at a moment when debates about the nature of intelligence have spilled well beyond comparative psychology. As of spring 2026, researchers building large language models and other AI systems are actively drawing on cognitive benchmarks originally designed for human and nonhuman primates to evaluate machine performance on tasks like causal reasoning, theory of mind, and cooperative problem-solving. An open, standardized archive of how great apes actually perform on such tasks gives AI researchers an empirical reference point that was previously unavailable at scale. At the same time, the dataset offers cognitive scientists a way to stress-test whether the capacities AI systems appear to replicate are genuinely shared with biological minds or merely superficially similar. The intersection is not speculative: several of the 262 experiments in the archive cover domains, such as inferring others’ intentions and reasoning about hidden causes, that are now central to benchmarking AI cognition. Having granular, trial-level ape data in the open could sharpen the comparisons that both fields are already trying to make.
The questions the Leipzig archive opens for evolutionary science
For comparative psychologists, the practical payoff is immediate. Any lab with basic statistical software can begin secondary analysis today without negotiating individual data-sharing agreements. Cross-study comparisons that were previously logistically impossible are now a matter of filtering metadata tags and writing code. A team interested in spatial memory can pull every relevant experiment from the archive and build a pooled analysis with statistical power that no single zoo study could achieve alone.
For the broader debate about human origins, the dataset shifts the terms of argument. Competing hypotheses that have circulated for years can now be tested against a shared empirical baseline rather than argued through narrative reviews of small, incomparable studies. Did capacities like understanding others’ intentions or using causal cues to solve physical problems develop gradually across great ape lineages, or did they appear as a sharp break in the human branch? Are individual differences in ape performance stable across time and tasks, suggesting something like a general intelligence factor, or are abilities highly domain-specific? The raw material to begin answering those questions is now sitting on a public server.
What happens next depends on who picks it up. The dataset’s value will multiply if other primate research centers and field sites follow Leipzig’s lead and release comparable open archives. Sánchez-Amaro has described the project as a model for how the field should operate going forward. Whether that vision takes hold will say as much about the sociology of science as it does about the minds of apes.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
