Evidence from Over 30 Studies

Evolved Adaptations to War in Humans

Cognitive psychology experiments with ~12,000 participants suggest that human minds are equipped with specialized mechanisms for detecting, enumerating, and assessing coalitional threats—consistent with the hypothesis that small-scale war was a recurrent feature of human evolution.

Bartusevičius, Aminihajibashi, Goetz, Skoog & Hagen · Peace Research Institute Oslo · Open Research Europe, 2026

Was War a Recurrent Feature of Life During Human Evolution?

Whether war characterized human prehistory is one of the most debated questions in the social and biological sciences. Archaeological evidence—skeletal trauma, projectile-embedded bones, fortifications, and cave art depicting violence—documents intergroup coalitional aggression going back thousands of years, though the clearest signs appear only around the end of the last Ice Age, and older traces remain sparse and contested. Ethnographic studies of small-scale societies and observations of intergroup aggression in chimpanzees add further evidence.

Yet the debate remains unresolved: some scholars argue humans have deep evolutionary roots in coalitional aggression, while others view war as a recent cultural invention. This project takes another approach—instead of relying solely on bones and artifacts, it uses cognitive psychology experiments to probe for evolved psychological mechanisms in contemporary humans.

The logic is straightforward: if small-scale war was a recurrent adaptive problem during human evolution, then natural selection should have shaped our minds with efficient, complex, and specialized mechanisms for navigating it.

Archaeological Evidence

Skeletal trauma, projectile injuries, cranial fractures, and multiple burials dating back thousands of years

Primate Parallels

Intergroup coalitional aggression in chimpanzees suggests deep evolutionary roots

Small-Scale Societies

Records from forager and horticultural societies document intergroup raiding and warfare

Cognitive Experiments

Testing for psychological adaptations in contemporary humans

What Would Adaptations to War Look Like?

If ancestral humans regularly faced surprise encounters with potentially hostile outgroups, their survival depended on solving three critical tasks—rapidly and under extreme pressure.

Detect

Task I: Coalition Detection

Automatically attend to groups of coordinated males—the ancestral cue for a potential coalitional attack.

1 2 3 4 ?

Count

Task II: Enumeration

Rapidly and accurately determine how many opponents there are—a critical input for deciding whether to fight, flee, or stand ground.

Assess

Task III: Formidability Assessment

Compute the coalition's fighting capacity based on their number, individual size, your own strength, and available allies.

Imagine foraging with a small group when you suddenly encounter unfamiliar men moving in coordinated formation. In seconds, your mind must detect them, count them, and assess whether they are more formidable than your group—then decide whether to flee, stand ground, or fight. These are the adaptive problems the research investigates.

How We Tested It

0 Preregistered Studies
0 US Participants
0 Exploratory Pilots
0 Hypotheses Tested

The Stimuli

Schematic 3D-rendered human silhouettes were used rather than photographs to isolate the key variables—sex of individuals, postural coordination, and group size—while minimizing cultural confounds.

Coordinated Male-Like Group

Aligned posture, broad shoulders, narrow hips—the key coalition cue

Rotated Male-Like Group

Same figures, different orientations—the non-coalition control

Coordinated Female-Like Group

Aligned posture but with female-typical body characteristics

Three Experimental Paradigms

Studies 1 & 2

Dot-Probe Task

Do people automatically attend to threatening coalitions?

Two images flash briefly on screen. A probe appears where one image was. Faster responses to probes replacing coalition images reveal automatic attentional bias—measured in milliseconds.

Automatic Attention
Studies 3 & 4

Enumeration Task

Do people count threatening coalitions more accurately?

Arrays of figures flash for 400ms. Participants estimate how many they saw. Comparing accuracy across target categories reveals whether coalitions that evolutionarily posed the threat of an attack receive privileged enumeration.

Counting Accuracy
Studies 5 & 6

Counting Speed Task

Do people count threatening coalitions more quickly?

The same arrays remain visible until counted. Response latency (in milliseconds) reveals whether coalitions that evolutionarily posed the threat of an attack are enumerated faster than other targets.

Counting Speed
Study 7

Formidability Rating

How do people assess the fighting capacity of threatening coalitions?

Participants rate the fighting capacity of coalitions varying in size (4–10) and formidability of individual members. Ratings reveal systematic assessment mechanisms.

Threat Assessment

What the Experiments Revealed

Finding 1

Automatic Attention to Coalitions

Participants automatically attended to coordinated male-like groups—stimuli that evolutionarily cued threat of a coalitional attack. This bias was specific: coordination among female-like figures did not produce the same effect.

Coordinated Males vs Rotated Males
4.0 ms
p = 0.049
Large Coord. Male Groups vs Small Coord. Male Groups
6.7 ms
p = 0.001
Large Coord. Male Groups vs Small Rotated Male Groups
10.5 ms
p < 0.001
Coordinated Females vs Rotated Females
0.9 ms
p = 0.669 (n.s.)

Attentional bias in milliseconds (Studies 1 & 2). Larger values indicate stronger automatic attention toward the first-listed stimulus. Coordinated males consistently drew attention; coordinated females did not.

Finding 2

More Accurate Counting of Coalitions

People estimated the size of coordinated male-like groups with less error than rotated male-like groups, coordinated female-like groups, or simple dot arrays—despite dots being visually simpler.

Probability of Correct Response

Coord. Males
76.6%
Rotated Males
75.6%
Coord. Females
75.1%
Black Dots
77.6%

Males were also more accurate than females: 77.8% vs 74.3% correct responses for coordinated male-like targets (p < 0.001).

Estimation accuracy from Studies 3 & 4 (Study 4 shown). Coordinated male-like targets were enumerated with lower error than both rotated males and coordinated females, suggesting a category-specific effect beyond mere visual similarity.

Finding 3

Faster Counting of Coalitions

Coordinated male-like targets were counted faster than rotated male-like targets, and male participants counted coalitions significantly faster than female participants.

Counting Latency (milliseconds)

Coord. Males
1,957 ms
Rotated Males
1,977 ms
Black Dots
1,936 ms
1,863 ms Male participants
193 ms faster p < 0.001
2,056 ms Female participants

Counting latencies from Study 5. Coalition speed advantage was 20ms over rotated males (p < 0.001). The sex difference (193ms) was also highly significant.

Finding 4

Systematic Formidability Assessment

Larger coalitions were rated as more formidable. The presence of a physically larger individual boosted ratings—but this effect diminished as overall coalition size grew, suggesting a nuanced assessment mechanism.

Predicted Formidability Rating by Coalition Size

108642
45678910
Number of individuals in the coalition
Without large individual
With one larger individual
+0.613 rating points per additional member p < 0.001
+0.088 boost from one larger individual p = 0.037
−0.066 interaction: large-individual effect fades with group size p < 0.001

Formidability ratings from Study 7 (1–10 scale). The lines converge as coalition size increases—individual size matters less when facing a large group.

Finding 5

Complementary Sex Differences

Males and females showed distinct patterns across tasks. Males were more accurate and faster at enumerating coalitions. Females perceived coalitions as more formidable and were more likely to overestimate their size—a potentially adaptive cautionary bias.

♂ Male Participants

  • More accurate enumeration (77.8% vs 74.3%)
  • Faster counting (193ms advantage)
  • Lower formidability ratings
  • Lower overestimation rates (9.2–9.4%)

♀ Female Participants

  • Higher formidability ratings (b = −0.308, p = 0.002)
  • Greater overestimation (11.6–12.3%)
  • Slower enumeration
  • Sex difference in ratings decreased with coalition size

These complementary patterns are consistent with evolutionary logic: ancestral males, more likely to engage in intergroup coalitional aggression, may have evolved more precise assessment mechanisms; ancestral females may have evolved more cautious threat appraisals, erring on the side of overestimation.

Hypothesis Scorecard

11 of 14 hypotheses supported across 7 preregistered studies

H1 Supported Automatic attention to coalitions
H2 Supported Accurate enumeration
H3 Partly Rapid enumeration
H4 Supported Coalition size → formidability
H5 Supported Individual size → formidability
H6 Supported Size effect diminishes with N
H7 Not Supported Own formidability effect
H8 Not Supported Coalitional support effect
H9 Supported Males count more accurately
H10 Supported Males count more rapidly
H11 Supported Females overestimate more
H12 Supported Females rate as more formidable
H13 Supported Less formidable → overestimate
H14 Not Supported Low coalitional support → overestimate

What It Means

The convergence of findings across detection, enumeration, and assessment tasks points to a coherent pattern: human minds appear to be equipped with specialized mechanisms for navigating coalitional aggression. Alternative explanations—general attention to humans, similarity-based grouping, or culturally learned associations—fail to account for the full pattern of results.

The hypotheses were derived a priori through a systematic adaptationist framework and preregistered before data collection. The fact that schematic, context-free stimuli produced these effects in online experiments suggests that the underlying mechanisms would operate even more strongly in naturalistic, high-stakes encounters.

Effect sizes were modest by design: the researchers intentionally used conservative, stripped-down stimuli. Yet the effects were consistent across independent samples totaling ~12,000 US residents, and the pattern was coherent across three fundamentally different experimental paradigms.

What Comes Next

The project is now conducting Study 8—replicating the core experiments across 30 countries to test whether these patterns are universal or vary with cultural histories of intergroup violence. If the effects do not significantly differ across diverse cultural contexts, this would provide strong evidence for evolved, species-typical adaptations shaped by ancestral coalitional aggression.