# Critiques of Dunbar's Number > [!summary] > Recent statistical reanalyses and theoretical challenges have questioned whether a single cognitive limit on human group size can be derived from primate data, revealing enormous confidence intervals and alternative explanations for group size variation. While [[Dunbar's Number]] has been widely influential for three decades, it faces substantial challenges from researchers who question both its statistical foundations and theoretical assumptions. These critiques do not necessarily disprove the existence of cognitive constraints on social networks, but they challenge whether we can specify a precise number like 150. ## Statistical and Methodological Critiques ### The 2021 Swedish Reanalysis In 2021, researchers at Stockholm University (Lindenfors, Wartel, and Lind) published a comprehensive reanalysis of Dunbar's original primate data using updated datasets and different phylogenetic methods. Their findings were striking: - Bayesian methods predicted average group sizes between 69-109 - Generalized least-squares (GLS) methods predicted 16-42 - 95% confidence intervals were enormous: 4-520 (Bayesian) and 2-336 (GLS) The researchers concluded: "A cognitive limit on human group size cannot be derived in this manner." The confidence intervals are so wide that specifying any single number becomes meaningless. When the plausible range spans from 2 to 520 people, saying "the number is 150" provides little useful information. ### Methodological Disputes Dunbar defended his work by pointing out that different regression methods yield different results. He used reduced major axis (RMA) regression, which predicted 158 for humans. The Stockholm study used ordinary least squares regression (LSR), which predicted 71. The choice of statistical method dramatically affects the outcome. This raises a deeper question: if the result depends so heavily on which statistical technique you use, how robust is the underlying finding? Critics argue this reveals fundamental instability in the relationship being measured. ### The Problem of Extrapolation The core methodological concern is whether it's valid to extrapolate from non-human primates to humans at all. Human brains don't simply scale up primate brains—we have unique cognitive capacities, particularly language and complex culture, that may fundamentally change the constraints on social network size. ## Theoretical Challenges ### Human Brain Uniqueness Critics argue that "Dunbar's assumption that the evolution of human brain physiology corresponds with a limit in our capacity to maintain relationships ignores the cultural mechanisms, practices, and social structures that humans develop to counter potential deficiencies." Humans use: - Written records to extend memory - Social institutions to organize relationships - Cultural norms to simplify complex social coordination - Language as a vastly more efficient form of "social grooming" than physical grooming These uniquely human capacities might allow us to maintain larger networks than brain size alone would predict. ### The Diet Hypothesis Some researchers have found that primate neocortex and brain size correlate better with diet than with sociality. Fruit-eating primates need larger brains to remember where and when fruiting trees are productive across large territories. Leaf-eating primates, with more abundant food, have smaller brains relative to body size. If diet explains more variance in brain size than social group size, then the foundation of the [[Social Brain Hypothesis]] becomes questionable. ### Other Ecological Factors Critics note that primate sociality is explained by multiple factors beyond brain size: - What they eat (dietary complexity) - Who their predators are (safety in numbers) - Habitat type (forest vs. savanna) - Mating systems (monogamous vs. polygamous) Social group size might be an adaptation to these ecological pressures rather than a constraint imposed by cognitive capacity. ## Empirical Challenges ### Wide Variation in Human Group Sizes When researchers have examined actual human social organization, they find enormous variation: - A 2010 study suggested people actually know between 472 and 611 people - Hunter-gatherer band sizes (where people spend most time) average around 30, not 150 - Online social networks show mean sizes around 150-200 but with huge individual variation - Different cultures show different patterns, with possible bias toward Western, educated, industrialized societies in the supporting data ### The Hunter-Gatherer Evidence Anthropologist Frank Marlowe suggested that hunter-gatherers spend most of their time in "local bands" of only about 30 people. The 150-person "clan" is a more abstract unit that rarely assembles as a whole. If humans evolved in small bands, why would we have cognitive adaptations for managing 150 relationships? ### Digital Age Findings Research on social media has produced mixed results: - Some studies show that Facebook friend counts cluster around 150 - Others find that people maintain meaningful contact with only 5-15 people online, regardless of total friend count - Social media may function primarily to slow relationship decay rather than to expand network size These findings support the idea of layered networks (see [[Nested Social Layers]]) but don't clearly validate 150 as a hard limit. ## Dunbar's Defense Dunbar has responded to critics by arguing that: 1. **Methodological rigor:** The RMA regression method he used is appropriate for the kind of biological data involved, where both variables have measurement error. 2. **Accumulating evidence:** Beyond the original primate regression, 30 years of research has found the 150 pattern in telephone networks, Christmas card lists, military units, village sizes across history, church congregations, and online gaming communities. 3. **Quality vs. quantity:** Critics often conflate "people you know" with "people you can maintain stable relationships with." Dunbar's number applies only to quality relationships, not to the much larger number of acquaintances. 4. **Layered structure:** The nested pattern of 5, 15, 50, 150 has been replicated across many studies and contexts, lending credibility to the underlying theory even if the exact numbers vary. ## What Remains Uncertain The debate reveals several unresolved questions (see [[Cognitive Limits on Group Size - Open Questions]]): - Is there a cognitive limit on human social network size? - If so, is it best characterized as a single number or a range? - How much do cultural and technological factors modify biological constraints? - Do different types of relationships (kin vs. friends, strong ties vs. weak ties) have different cognitive ceilings? - How much individual variation exists, and what explains it? ## Implications for Practice Even if the critiques are valid, they don't necessarily invalidate the practical insights: - Organizations consistently experience problems when they exceed 150-200 members without subdividing - Humans clearly have some limits on how many relationships they can maintain - Nested social structures appear robust across cultures - Time constraints alone create practical limits regardless of cognitive capacity The question becomes not "Is Dunbar's number exactly 150?" but rather "What are the actual constraints on human social networks, and how should we design institutions that work with rather than against them?" Regarding why institutions of 2000-3000 don't naturally subdivide into units of 250: perhaps the real insight is that human communities need subdivision at multiple scales. Without intentional design creating teams (5-15), departments (30-50), and divisions (100-200), large institutions become socially incoherent regardless of whether the precise limit is 150, 250, or some other number. ## See Also - [[Dunbar's Number]] - [[Social Brain Hypothesis]] - [[Nested Social Layers]] - [[Cognitive Limits on Group Size - Open Questions]]