Citation
Esther Herrmann, Josep Call, María Victoria Hernández-Lloreda, Brian Hare, Michael Tomasello, "Humans Have Evolved Specialized Skills of Social Cognition: The Cultural Intelligence Hypothesis," Science, New Series, Vol. 317, No. 5843 (September 7, 2007), pp. 1360-1366. DOI: 10.1126/science.1146282. JSTOR: https://www.jstor.org/stable/20037753.
Overview
A landmark comparative-cognition paper proposing — and providing direct experimental evidence for — the cultural intelligence hypothesis: humans diverge cognitively from their nearest primate relatives not primarily because of higher general intelligence, but because of a species-specific, early-emerging package of social-cognitive skills that enable participation in and exchange of knowledge within cultural groups. The paper is one of the most-cited empirical foundations for the claim that human cognition is distinctively cultural — and, by extension, that knowledge management practices rest on an evolved substrate.
The Hypothesis
Two competing accounts of human cognitive uniqueness:
- General intelligence hypothesis. Humans are just "smarter" than other apes across the board — better at reasoning, memory, problem-solving generally. Differences across domains are a matter of degree, not kind.
- Cultural intelligence hypothesis (the paper's proposal). Humans have roughly comparable physical cognition to great apes but diverge sharply in social cognition — specifically the skills for learning from, communicating with, and coordinating with other minds. These skills emerge early in development and are the foundation for participating in culture.
The paper tests these against each other directly.
Method — The Primate Cognition Test Battery (PCTB)
A 16-task, 2-domain, 6-scale battery designed by Tomasello and Call, covering a representative range of cognitive skills:
Subjects (N = 243)
- 106 chimpanzees (Pan troglodytes), 53M/53F, ages 3-21 (mean 10), at Ngamba Island (Uganda) and Tchimpounga (Republic of Congo) sanctuaries
- 32 orangutans (Pongo pygmaeus), 17M/15F, ages 3-10 (mean 6), at Orangutan Care Center and Quarantine (Pasir Panjang, Kalimantan, Indonesia)
- 105 human children (Homo sapiens), 52M/53F, aged 2.5 years ± 2 months, medium-sized German city — all using language for ~1 year, before literacy and formal schooling
Test structure
| Domain | Scale | Tasks |
|---|---|---|
| Physical | Space | spatial memory, object permanence, rotation, transposition |
| Physical | Quantities | relative numbers, addition numbers |
| Physical | Causality | noise, shape, tool use, tool properties |
| Social | Social learning | social learning |
| Social | Communication | comprehension, pointing cups, attentional state |
| Social | Theory of mind | gaze following, intentions |
Administered individually over 3-5 hours across multiple days. Temperament control battery given in parallel. Inter-observer agreement 98-99%.
Key Findings (Quantitative)
Top-level result (proportion correct)
| Domain | Human | Chimpanzee | Orangutan |
|---|---|---|---|
| Physical | 0.68 | 0.68 | 0.59 |
| Social | 0.74 | 0.36 | 0.33 |
- On the physical domain, humans and chimpanzees did not differ; both outperformed orangutans (P < 0.001). Effect size η² = 0.14.
- On the social domain, humans outperformed both ape species by a large margin; apes did not differ from each other. Effect size η² = 0.72 — one of the largest effect sizes reported for a cross-species cognition comparison.
Within-scale patterns
- Social learning scale: human 0.86, chimp 0.10, orangutan 0.07. This is where the human-ape gap is largest — nearly complete for both ape species.
- Communication: human 0.72, chimp 0.57, orangutan 0.55; pointing-cups task specifically shows no species difference at all (0.72/0.74/0.73).
- Theory of mind: human 0.65, chimp 0.40, orangutan 0.36.
- Three tasks where chimps beat humans: transposition (0.70 vs 0.57), addition numbers (0.69 vs 0.64), tool use (0.74 vs 0.23). The tool-use finding is striking — 2.5-year-old children largely fail the one active-tool task the battery includes.
Interpretation
The pattern cleanly supports cultural intelligence over general intelligence: humans and chimps match on physical cognition but humans dominate social cognition. The authors note the social-cognition advantage emerges early in ontogeny, before literacy and formal schooling — consistent with the "bootstrap" hypothesis that specialized social-cultural skills are what enables humans to absorb the accumulated knowledge of their cultural group.
Authors' Alternative Hypothesis (flagged in the Discussion)
The authors themselves acknowledge an interesting alternative: since children also outperformed apes on causality tasks not involving active tool manipulation, "what may be distinctive is the ability to understand unobserved causal forces in general, including (as a special case) the mental states of others as causes of behavior." They add that even so, this general causal-reasoning ability may have evolved first for social/mental-state understanding and only later generalized.
Caveats Discussed by the Authors
- The battery was constructed and administered by humans. Previous studies with the same social-domain tasks using human vs conspecific interactants show no significant effect, and temperament did not correlate with social-domain performance — addressing the obvious concern.
- Different test batteries could yield different results. The social tasks are more open to controversy than the physical tasks.
- Dog comparison (Hare et al. 2002): domestic dogs underperform chimps on physical cognition but outperform them on social cognition — a striking supporting data point since dogs were not evolutionarily selected for cultural cognition but were selected for social interaction with humans.
Key Concepts
- Cultural intelligence hypothesis — the paper's central claim
- General intelligence hypothesis — the alternative it argues against
- Social cognition — cognitive skills for understanding and interacting with other minds: social learning, communication, theory of mind
- Physical cognition — cognitive skills for understanding objects, space, causality, quantity
- Ontogeny — development over individual lifespan (the paper emphasizes early-emerging skills, pre-schooling, pre-literacy)
- Social learning — acquiring knowledge through observation of and interaction with others, vs. individual exploration
- Cumulative culture — the accumulation of knowledge across generations via social learning; enabled by the social-cognitive skills this paper identifies
Why This Matters for PKM
The cultural intelligence hypothesis is the evolutionary substrate that makes PKM possible and necessary. Several consequences:
- PKM is not just a modern accident. The human capacity for — and drive toward — accumulating, sharing, and organizing knowledge is built in. It is what we are for, cognitively. Note-taking, commonplace books, Zettelkasten, AI-assisted vaults: all are cultural technologies on top of a biological substrate.
- Social learning is the foundation of knowledge work. We are not built to figure things out alone; we are built to learn from others. Networked reading, public learning, community participation are not optional extras in PKM — they are close to its core function.
- Cultural accumulation > individual cleverness. The paper's finding that children don't outperform chimps on physical cognition, only on social, suggests that human intellectual achievement is overwhelmingly a function of what the group can accumulate together, not what any individual can figure out alone. Compounding knowledge is the mechanism; cultural intelligence is the predisposition.
- The AI collaborator as cultural partner. Contemporary agentic knowledge management plugs into evolved cultural-cognition circuits. We may be prone to treating AI systems as social-learning partners (teachers, collaborators) even when that framing is imprecise — both a strength (natural engagement) and a risk (sycophancy, overreliance).
- Why we're bad at solo knowledge work. If our cognition is evolved for cultural participation, the solo-thinker model (write alone, think alone, produce alone) fights against our actual cognitive architecture. Practices like pair programming, writing groups, and third places for thinking succeed because they restore the social substrate.
Credibility and Caveats
- Tier: high. Peer-reviewed Science paper, N=243 (large for comparative cognition), senior author Michael Tomasello is one of the most prominent living figures in the field.
- Full-text access. Obtained via Max Planck Institute for Evolutionary Anthropology mirror (https://www.eva.mpg.de/documents/AAAS/Herrmann_Humans_Science_2007_1554784.pdf). Quantitative data above verified against the primary paper.
- Since-2007 literature. The hypothesis has been developed (Tomasello's A Natural History of Human Thinking 2014, Becoming Human 2019), partially competed-with (Heyes's Cognitive Gadgets 2018 argues these skills are culturally transmitted rather than genetically specified), and nuanced (subsequent work finds ape social cognition richer than 2007 suggested, narrowing but not closing the gap). This summary captures the 2007 claim; the 2026 state of debate is richer.
Concepts and Entities
Concepts
- Cultural intelligence hypothesis
- General intelligence hypothesis (the competitor)
- Social cognition (domain)
- Physical cognition (domain)
- Theory of mind
- Social learning
- Cumulative culture
- Ontogeny of cognitive skills
Entities (People, Institutions, Species)
- Esther Herrmann (Max Planck Institute for Evolutionary Anthropology, Leipzig)
- Josep Call (Max Planck Institute, now University of St. Andrews)
- María Victoria Hernández-Lloreda
- Brian Hare (Duke University)
- Michael Tomasello (Max Planck Institute; Duke University) — senior author, major comparative-cognition figure
- Pan troglodytes (chimpanzees) — test species
- Pongo (orangutans) — test species
- Homo sapiens children, 2.5 years — test species