← all systems · see it in The System
Social group
In the social sciences, a social group is defined as two or more people who interact with one another, share similar characteristics, and collectively have a sense of unity.
Properties published scienceour proposal
Sourced fields, and our own readings.
| born at | for each group, at its first meetings (Sherif planned 5-6 days for in-group formation); for the class contested (C-3989) |
| born precision | days (per group); unknown (class) |
| boundary | Inside: the members while they interact as a unit and the structure, norms, identity and bonds that make them one (roles, status, cohesion, shared category), the processes by which the group acts on members (conformity, facilitation, loafing, polarisation, depersonalisation) and the group's acts as a unit (decisions, performance, intergroup behaviour, splitting and merging). Outside: the persons themselves (sys:human; in animals, the individual animals), other groups (out-groups), the society and networks the group sits in, the ecology (food, predators, tasks), and the media that carry it online. A collection without interaction or shared identity (people at a bus stop) is outside: 'more than a simple collection or aggregate of individuals' (src:wp:social-group). |
| nested level | Between the person and the society: 'groups represent a level of association mediating between the individual and whole societies' (src:flamino-2021). Built from members, dyads and cliques; part of a society's nested groups and networks (sys:human-society L3); carried on members' acts and talk (runs on Human L7) and, online, on social media. |
| scale | social |
Measures (32) published science
| Asch conformity error rate (original) | 36.8 % of critical trials (123 subjects) |
| Asch error rate (replication, no incentive) | 33 % of critical trials (202 valid cases, both conditions) |
| Asch error rate (replication, incentivised) | 25 % of critical trials |
| conformity on political opinions | 38 percentage points (average difference by group's answer) |
| average error rate in strict Asch replications (Bond and Smith) | 25 % (44 strict replications) |
| effort per member, shouting with one other | 82 % of solo effort |
| effort per member, shouting with five others | 74 % of solo effort |
| social loafing meta-analysis | 78 studies |
| competition gain in paced cycling records | 5.15 seconds per mile (up to 25 miles) |
| ostracism effect size | d > |1.4| standardised mean difference (120 Cyberball studies, N = 11,869) |
| risk of death in 6 years after retirement | 2 / 5 / 12 % (two groups kept / one lost / both lost) |
| collective intelligence data | 22 studies; 5,279 individuals; 1,356 groups count |
| average variance extracted by the CI factor | 19.6 % |
| predicted human group size (Bayesian; GLS) | 69-109; 16-42 individuals (95% intervals 4-520; 2-336) |
| network layer sizes (Dunbar Graph) | 1.5, 5, 15, 50, 150, 500, 1500, 5000 individuals per layer |
| Facebook friends (two UK samples) | 155.2; 182.8 mean friends |
| support clique (Sample 1) | 4.1 mean individuals |
| scaling ratio between preferred group sizes | about 3 ratio (3, 9, 27, ...) |
| grooming time in the most social primates | up to 20 % of the day |
| baboon group size (Amboseli) | 47 (SD 21.2) baboons (153 group-years) |
| face-to-face groups in NetSense | 436 groups; average 3.9-4.1 members; lasting 1-4 semesters groups / members / semesters |
| amygdala study sample | 58 adults (age 19-83) |
| deliberation experiment | 2,793 participants at 339 tables in 19 sites count |
| money awarded for sameness against difference (no groups) | 43.1 % more |
| Robbers Cave design: in-group formation stage | 5-6 days |
| groupthink studies reviewed by Park | 16 empirical studies (to 1990) |
| time for in-group preference to form | a matter of minutes |
| variance explained by the presence of others | 0.3-3 % of variance (241 studies, nearly 24,000 subjects) |
| variance explained by the group factor (Bates & Gupta study 1) | 39.8 % |
| hidden-profile meta-analysis | 65 studies; 3,189 groups; 8 times less likely to solve count / odds |
| children judging a group 'real' (friends / task / look-alike / tram stop) | 85.4 / 81.3 / 85.4 / 33.3 % of children (N = 137) |
| intergroup contact meta-analysis | 713 samples from 515 studies count |
Open contentions (13) open contention
Where sources or readings disagree. Both stay on the map.
- C-413 Where do thoughts come from?
booked: No theory is proven, so the map lists the main ones side by side and doesn't pick one. It shows where thinking is handled in the brain, not where thoughts start.
the other reading: - C-790 Does natural selection act on whole groups?
booked: The map notes that group selection, if it happens, acts on groups, and that how much it happens is still debated.
the other reading: - C-2736 How many relationships can a person keep?
booked: The map keeps the competing readings; none is ruled.
the other reading: - C-3437 Is there a natural limit of about 150 on a person's social circle?
booked: The map keeps the competing readings; none is ruled.
the other reading: - C-3922 Do groups have a general 'collective intelligence', or just the intelligence of their members?
booked: The map keeps the competing readings; none is ruled.
the other reading: see booked - C-3929 Do people in crowds lose themselves, or act on a shared identity?
booked: The map keeps the competing readings; none is ruled.
the other reading: crowd behaviour driven by deindividuation (the encyclopedia's framing; Le Bon among the theorists); shared social identity that guides and empowers crowd action (elaborated social identity model, Stott et al.; depersonalisation is not loss of self) - C-3944 Do like-minded groups always become more extreme when they talk?
booked: The map keeps the competing readings; none is ruled.
the other reading: the 'law of small-group polarization' (Sunstein, as cited); deliberation's persuasion in moderated, diverse groups with balanced material (Esterling et al.) - C-3945 Does 'groupthink' hold up when it is tested?
booked: The map keeps the competing readings; none is ruled.
the other reading: cohesion the main antecedent (Janis); no main effect of cohesion in 16 studies, homogeneity and insulation supported (Park 1990 as reported); Turner & Pratkanis 1998 on the evidence (abstract only) - C-3955 Do animals in bigger groups really spend less time watching for predators?
booked: The map keeps the competing readings; none is ruled.
the other reading: vigilance falls with group size, a benefit of grouping (Pulliam 1973, over fifty studies as reviewed); most studies fail to show it unambiguously once confounds are controlled (Elgar 1989) - C-3957 Why do people favour 'their' group even when the groups are meaningless?
booked: The map keeps the competing readings; none is ruled.
the other reading: social identity and positive distinctiveness (Tajfel and Turner); a generic norm of social competition (Tajfel's first account); interindividual comparison treating difference worse than sameness, with no group at all (Deschrijver & Ramsey 2025) - C-3967 What made primate brains big: living in groups, or finding food?
booked: The map keeps the competing readings; none is ruled.
the other reading: group size (social brain: Dunbar, Dunbar & Shultz; path analyses); ecology, little evidence for group size (Powell et al. 2017; DeCasien et al. 2017 as cited); both diet and sociality (Grabowski et al. 2022) - C-3987 Why do people perform differently when others are watching?
booked: The map keeps the competing readings; none is ruled.
the other reading: mere presence raises arousal and strengthens dominant responses (Zajonc); fear of evaluation (Henchy and Glass); Triplett's 'latent energy'; Bond & Titus 1983 (241 studies, abstract): effects small (0.3-3% of variance) and unrelated to evaluation apprehension; Strube 2005 (abstract): Triplett's own data showed very little facilitation - C-3988 What makes a set of individuals a group: interaction, a shared identity, or simply staying close?
booked: The map keeps the competing readings; none is ruled.
the other reading: behavioural ecology's call to study the dynamics rather than static 'social groups' (Madsen & de Silva 2024); groupness as a graded quantity in people (entitativity: Lickel et al. 2000 as summarised by Plötner et al. 2016, with four empirically derived types), paralleling behavioural ecology's degree of fission-fusion
Cross-scale patterns seen here (23) our proposal
Shapes that recur when the same kind of problem is solved at different scales, and how each shows in Social group. Each links to every other system that shows it.
- same role, different builds (structural)
One job, keeping members bonded, is done by grooming and the endorphin system in most primates and by language (and shared identity) in humans. - symmetry breaking (name the cut) (structural)
The minimal group: a random, symmetric split (a coin toss) into A and B, after which allocations are no longer symmetric: members favour their own side. The cut is named (the category label) and it sets what is built after it. - the whole sets the local reference (structural)
Each member reads its own judgement against the group: a third of answers follow a wrong majority on lines a child could judge, and the group used as a reference group is the standard members evaluate themselves by. The salient category sets who is prototypical and so who influences whom. - trade one property for another (structural)
Group size trades protection and between-group strength against within-group competition: larger groups can win food from smaller ones but deplete patches faster for their own members; the human network layers are read as a trade-off between the time costs of relationships and their benefits. Specie - who opens the door (structural)
Membership doors are set by the receiving group ahead of time: closed groups require a reference from a current member or an initiation; open groups leave the door open. The receiver opens the door (a port). - borrowed-then-held (partial)
Structure supplied from outside, then held: Sherif's staff introduce goals with common appeal that need interdependent action; the groups then hold their own hierarchy and norms (supported in the 1949 and 1953 camps). A newcomer's compliance (borrowed) turns into internalization (held). - causal ceiling (partial)
In fluid (fission-fusion) groups no member can witness all interactions: they must remember members they do not see and infer relationships from partial observations. The limit is set by the group's structure, not by members' means. - coordination-before-mapping (partial)
Fission-fusion groups modelled with agents that follow simple rules (move randomly, flock with neighbours, follow a male) and need no sophisticated cognition reproduce grouping patterns; whether real animals do it this way is the model's question. - namespace completeness (the map predicts its gaps) (partial)
The social-brain account predicted a human group size of about 150 from the primate regression, and its authors report later confirmation in 23 studies and a whole series of layer sizes (1.5, 5, 15, 50, 150, 500, 1500, 5000); the re-analysis says no number can be derived this way. A structure that c - nested ladder (partial)
Influence deepens in steps, each a stronger form of the last: compliance (outward), identification (imitation), internalization (the group's beliefs become one's own). Categories also nest by inclusiveness. - outside-to-inside (partial)
A partition drawn outside the person (an experimenter's random A/B assignment) comes first and is then carried inside as a self-category that guides allocations; Turner's definition is a category membership internalised into the self-concept. - same toolkit, independent builds (partial)
Fission-fusion dynamics arose in separate lineages (chimpanzees and spider monkeys, elephants, hyenas, dolphins), and the species with high fission-fusion share better inhibitory control, more than phylogeny or diet predicts. Aureli et al. caution that the link may be correlation. - templating (partial)
The same layer sizes are reported to recur face to face, on telephones and social media, in online games, and in primate, carnivore and cetacean groups (Dunbar's account); people's Facebook friends match offline network sizes. A module reused across media and species, as claimed. - The carry (partial)
Overflow passed on and the wait: past a size limit a gathering stops working as a group and becomes a crowd; primate groups that outgrow food or grooming time split, but fission is rare and a collective-action problem, so groups sit above their optimum until enough members leave together. - The Payment to Entropy (partial)
Pooling pays a toll: when members' efforts are combined, part of each member's effort never arrives (82% of solo shouting with one other, 74% with five); Karau & Williams find the loss robust across 78 studies. Form: a motivation toll on a pooled hand-off. Per-member efficiency about 0.82 to 0.74 as - Umwelt: each system carves its own slice of one shared medium (partial)
Each group, and each member, carves its slice of the same social world: which category is salient depends on accessibility and fit; members of cohesive groups are 'bombarded' with social information while members of fluid groups gather it patchily. - barrier-gates-complexity (surface)
Closed groups with rigid boundaries and initiations are the most cohesive; the without-barrier case (open groups) is described but cohesion is not shown to depend on the boundary. - holding / entrainment (surface)
Fission-fusion groups fuse at sleeping sites and split by day, so the group sets members' daily rhythm of being together; but the shared driver (night, predators, sleeping sites) may be what sets it (the dark twin). - lightning (surface)
A riot's sudden empowerment: a shared identity forms, rivalries are overridden and action spreads to new targets; and in-groups that favour themselves form within minutes. Firings that change what comes after; the riot from an abstract only. - partner-becomes-part (surface)
A newcomer is converted into a member whose beliefs are the group's (internalization); no inheritance, so only a surface likeness. - red-queen (surface)
Grouping prey and their predators: the timing of attacks selects prey's aggregation strategies, 'to which predators must respond'; a model, no stalemate shown. - runaway from an unstable balance (surface)
Reinforcing loops push groups away from their starting mix (opinion homogeneity rising each semester; a riot's empowerment loop); no balance point is shown. - same shape, opposite sign (surface)
One process, evaluation of a member's own output by others, with its sign reversed by the setting: present when outputs are individual (facilitation, evaluation apprehension) and absent when they are pooled (loafing, low evaluation potential). The joining of the two literatures is the walker's; each
Valence: what it trades, and with whom (11) published science
Each exchange with a neighbour, read from both sides: who gains, who pays, and what carries it.
- Predators (Cambrian and later) via sys:sensory-environment · Social group gain
- Amygdala via sys:sensory-environment · Social group gain
- Human (Homo sapiens) via sys:sensory-environment · Social group gain
- Natural selection via sys:sensory-environment · Social group mixed
- Human society (from local community up to societies) via sys:sensory-environment · Social group gain
- Social network via sys:sensory-environment · Social group gain
- Crowd via sys:sensory-environment · Social group mixed
- Social norm via sys:sensory-environment · Social group gain
- Food via sys:sensory-environment · Social group mixed
- Social media via sys:internet · Social group gain
- Out-group via sys:sensory-environment · Social group mixed
Partners and parts published science
Built from / builds (7)
- Human (Homo sapiens)
- Social norm
- Groups contesting food: larger groups can outcompete smaller ones
- Members scrambling for food inside a group
- Friendship
- Status and role structure
- Shared social identity
Sits inside / holds (3)
expressed by (11)
- Templating / Standardization
- Symbiosis/Co-Creation (Collaborative Specialization)
- Internal Coordination/Regulation
- Hypothesis/Goal Selection
- Social Learning / Public Information Use
- Resource Denial/Attrition
- Proactive Defense
- Physical/Virtual Barrier
- Validation of Identity/Intention
- Quality Control & Error Correction
- Scheduled Build-and-Dismantle (On-Demand Structure)
floor filled by (9)
- Amygdala
- Human (Homo sapiens)
- Human brain
- Human society (from local community up to societies)
- Social network
- Social norm
- Social media
- Status and role structure
- Shared social identity
Kinds (16)
- Primate social group
- Family (the human family as a system)
- Fish school (shoal)
- Crowd
- Peer group
- Reference group
- Social spider colony
- Selfish herd (predator-driven aggregation)
- Fission-fusion society
- Primary group
- Secondary (task) group
- Category
- In-group
- Out-group
- Minimal group (laboratory)
- Team (work group)
Notes published science
- Seventy years on, the line test still works: in a 2021 replication in Switzerland, people gave a plainly wrong answer about which line was longest 33% of the time when five others said it first, almost exactly Asch's 36.8%; paying for correct answers only brought it down to 25%, and the same pressure moved political opinions by 38 points (src:franzen-mader-2023).
- Keeping your groups goes with living longer: English retirees who kept both of their group memberships had a 2% risk of dying in the next six years, those who lost one had 5%, and those who lost both had 12%; the authors found the difference comparable in size to that of physical exercise. It is one cohort followed over time, so it shows an association, not proof that losing groups causes death (src:steffens-2016).
- The famous 'us against them' result may not need an 'us': in seven preregistered experiments with over 1,400 people, participants still gave less money to a single person who merely made a different guess, liked a different painting or got a different coin flip, with no groups at all, awarding 43.1% more for sameness than for difference (src:deschrijver-ramsey-2025).
Layers
The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.
social group: from individuals meeting in shared settings to a unit that shapes its members and acts on the world (reader's view)
- L1: Individuals together in a shared setting: face-to-face co-presence at given places and times; in animals, spatial association
- L2: Noticing and categorising: perceived similarity and common category membership, made salient by accessibility and fit
- L3: Joining, staying and leaving: attraction and homophily, recruitment and initiation, open or closed boundaries; members with unpopular opinions move
- L4: Structure: interdependent status and role relations, division of labour, dominance, leadership
- L5: Influence on members: conformity, facilitation and loafing, polarisation, depersonalisation; the social brain in members
- L6: Shared identity, norms and cohesion: social identity, in-group and out-group, the group's own norms and values
- L7: The group acting as a unit: collective decisions and performance, intergroup competition and collective action, splitting and merging; effects back on members' health
the boardroom (shared floors)
- L4: Structure: interdependent status and role relations, division of labour, dominance and leadership; the collaboration process
- L5: Influence on members: conformity, facilitation and loafing, polarisation, depersonalisation; in members, the social brain (amygdala and its cortical partners; in primates the endorphin bond)
- L6: Shared identity, norms and cohesion: social identity, in-group and out-group, the group's norms and values, the bonds to the group as a whole
- L7: The group acting as a unit: decisions, collective performance, intergroup competition and collective action, splitting and merging
way in: Emotion: belonging and exclusion
- L1: Being included or left out: the ball passed or withheld, being excluded and ignored
- L2: The reflexive sting of ostracism, large across 120 studies
- L3: The value and emotional significance attached to membership
way in: Hearing: talk, answers and discussion
- L1: Talk at the table: members' stated views in a discussion group
- L2: Hearing others' answers before giving one's own
- L3: Social comparison and informational influence route what is heard
way in: Screens: texting, calls and social media
- L1: Texting, online chat and social media between members
- L2: Online friends lists the same size as offline networks
- L3: Routing back to face-to-face: real relationships need occasional meetings
way in: Sight and hearing of other groups: comparison and competition
- L1: Other groups in functional relation: competing for the same goals
- L2: Comparison with relevant out-groups: categories defined and evaluated comparatively
- L3: Stereotypes of the out-group come into use under competition (Sherif's hypothesis)
way in: Sight: others present, their conduct and position
- L1: Others co-present in a shared setting: co-actors doing the same task, audiences watching
- L2: Members see one another; a co-actor's bodily presence registers
- L3: Arousal and evaluation route what is seen: the fear of being evaluated
way in: The surroundings: food, predators and tasks
- L1: Food distribution and predation pressure (animal groups); in human groups, the task
- L2: Members meet the food: encounter rates fall as search fields overlap
- L3: Movement rules route members toward others when predators threaten (the selfish herd)
way in: Thought: categories and self-categorisation
- L1: Categories supplied to members, even trivial or random ones (a coin toss, a painting preference)
- L3: Self-categorisation: seeing oneself as interchangeable with in-group members (depersonalisation), at the level made salient by accessibility and fit
way in: Touch: grooming and bodily contact
- L1: Grooming between group members (most primates)
- L2: The endorphin system takes up the contact and builds the bond
- L3: Bonds kept up only as time allows: grooming time routes how many relationships are held
way out: Acts on other groups: allocation, discrimination, competition
- L1: Contest for food between groups: larger groups can win
- L5: Through members' drive for positive distinctiveness
- L7: Maximum differentiation: favouring the in-group's relative gain
way out: Admitting, excluding, splitting and merging
- L1: New party compositions in space; subgroup size tracks food competition
- L5: Fission as a collective action problem: a subset must leave together
- L7: Fission and fusion: merging to sleep, splitting to forage; permanent fissions rare
way out: Members' acts in concert: collective work, collective action, a party's travel (movement)
- L1: Acts in the world: races ridden faster in company, ropes pulled less hard together
- L5: Through members' effort, lowered when inputs are pooled (social loafing)
- L7: Collective action, empowered by a shared identity
way out: Members' bodies: autonomic, endocrine and immune effects of belonging (autonomic and endocrine)
- L1: Health and survival: risk of death after retirement by memberships kept
- L5: Through immune, endocrine and cardiovascular function and allostatic load (in primates, group size and glucocorticoids)
- L7: Belonging, purpose and meaning given by memberships
way out: Members' talk: answers given, decisions voiced, norms stated (speech)
- L1: Views stated aloud at the table, persuading or polarising
- L5: Members voice the majority's answer, in judgement and in political opinion
- L7: The group's decision, sometimes by concurrence-seeking (groupthink)
Sources (53) published science
Wikipedia: Group polarization (sections: lead, risky shift, theories) (behind its links)
Pettigrew TF, Tropp LR (2006) A meta-analytic test of intergroup contact theory. JPSP 90(5):751-783. doi:10.1037/0022-3514.90.5.751 (behind its links)
Franzen A, Mader S (2023) The power of social influence: A replication and extension of the Asch experiment. PLOS ONE 18(11):e0294325. doi:10.1371/journal.pone.0294325 (behind its links)
Triplett N (1898) The dynamogenic factors in pacemaking and competition. American Journal of Psychology 9:507-533 (read in full) (behind its links)
Lindenfors P., Wartel A., Lind J. (2021) 'Dunbar's number' deconstructed. Biol Lett 17:20210158. doi:10.1098/rsbl.2021.0158 (behind its links)
Dunbar 2016, Do online social media cut through the constraints that limit the size of offline social networks?, R Soc Open Sci 3:150292 (doi:10.1098/rsos.150292) (behind its links)
Zhou W.-X., Sornette D., Hill R.A., Dunbar R.I.M. (2005) Discrete hierarchical organization of social group sizes. Proc R Soc B 272:439-444. doi:10.1098/rspb.2004.2970 (behind its links)
Wikipedia: Groupthink (sections: lead, history, symptoms and causes, empirical findings and meta-analysis, Bay of Pigs) (behind its links)
Wikipedia: In-group and out-group (lead) (behind its links)
Wikipedia: Fission-fusion society (lead) (behind its links)
Wikipedia: Social facilitation (sections: lead, Triplett, Zajonc, evaluation apprehension) (behind its links)
Wikipedia: Minimal group paradigm (read in full) (behind its links)
Wikipedia: Types of social groups (reached via 'Reference group'; sections: lead, primary and secondary groups, reference groups) (behind its links)
Wikipedia: Group cohesiveness (sections: lead, definitions, cohesion and performance) (behind its links)
Wikipedia: Group dynamics (sections: history, Lewin, Tuckman) (behind its links)
Powell LE, Isler K, Barton RA (2017) Re-evaluating the link between brain size and behavioural ecology in primates. Proc R Soc B 284:20171765. doi:10.1098/rspb.2017.1765 (behind its links)
Crowd psychology (Wikipedia, rev 1379448339, 2026-10-10, via Wikimedia Enterprise) (behind its links)
Dunbar RIM, Shultz S (2007) Understanding primate brain evolution. Phil Trans R Soc B 362:649-658. doi:10.1098/rstb.2006.2001 (behind its links)
Postmes T, Spears R (1998) Deindividuation and antinormative behavior: a meta-analysis. Psychological Bulletin 123(3):238-259. doi:10.1037/0033-2909.123.3.238 (behind its links)