Skip to content

Kin selection

A costly gene can still spread when it makes its carrier help relatives who likely share that gene, as long as the relatedness-weighted benefit to them (rB) outweighs the cost to the actor (C).

Core Idea

Kin selection is the evolutionary mechanism by which a costly allele can spread in a population because it causes its carrier to raise the reproductive success of relatives who carry copies of the same allele by descent, such that the allele's aggregate inclusive-fitness gain outweighs the direct fitness cost to the actor. The condition is formalised in Hamilton's rule: the allele is favoured by selection when rB > C, where r is the coefficient of genetic relatedness between actor and recipient — the probability, above chance, that a gene copy in the actor is identical by descent to a gene copy in the recipient — B is the fitness benefit to the recipient, and C is the fitness cost to the actor, both measured in units of offspring equivalents.

Three components make the mechanism run. First, relatedness by descent provides the statistical scaffolding: siblings share, on average, one-half of their alleles by descent, full cousins one-eighth, and so on through pedigree structure, so any allele that is identical by descent in the actor has a calculable probability of being present in any given relative. Second, the action must be directed preferentially toward relatives rather than dispersed at random across the population — achieved through kin recognition, philopatry, population viscosity, or any other mechanism that keeps the actor's social neighbourhood genealogically non-random. Without this, benefits leak to non-carriers and the r term drops toward zero, collapsing the advantage. Third, the relevant arithmetic is allelic, not phenotypic: a behaviour that depresses the actor's own reproductive output can still spread if the product rB for all recipients summed exceeds C, because selection acts on the allele's total representation across all copies in the generation, not on any individual carrier's fate.

The reframing that carries the weight is the shift from individual fitness to inclusive fitness — the actor's own offspring production plus the weighted sum of all effects the actor's behaviour has on relatives' offspring production, each weighted by r. Hamilton formalised this in two papers in 1964; the term "kin selection" was coined by Maynard Smith. Worker sterility in eusocial Hymenoptera — bees, wasps, and ants — was the original motivating puzzle: under haplodiploidy, sisters in a colony share three-quarters of alleles by descent, making sister-raising reproductively more efficient for the workers' shared alleles than producing their own offspring under many ecological conditions. Belding's ground squirrels, where alarm-call frequency correlates with the relatedness of the surrounding audience, and clonal microbial populations, where relatedness approaches one and cooperative public goods are maintained without defection for that reason, are subsequent canonical cases. Plant root-competition suppression near siblings is a more recent extension. Each case is governed by the same rB > C calculus, applied to the appropriate relatedness coefficient for the genetic system in question.

Structural Signature

Sig role-phrases:

  • the social allele — a heritable variant whose expression in its carrier alters the reproductive success of other organisms, at some direct cost to the carrier
  • the relatedness coefficient — r, the above-chance probability that a gene copy in the actor is identical by descent to one in the recipient, fixed by the genetic system and pedigree
  • the recipient benefit and actor cost — B and C, the fitness gain to the helped relative and the fitness loss to the actor, both in offspring-equivalent units
  • the discrimination channel — kin recognition, philopatry, or population viscosity that aims the benefit non-randomly at carriers, keeping r above zero
  • the inclusive-fitness ledger — accounting shifted from the individual vehicle to the allele's total representation across all copies in the generation
  • the Hamilton threshold — the allele is favoured exactly when the summed rB over recipients exceeds C, the inequality whose sign determines the trait's fate
  • the differential-reproduction outcome — the costly allele spreads when the arithmetic clears, because selection acts on its aggregate copy count rather than on any one carrier's offspring

What It Is Not

  • Not "for the good of the individual organism." The reframing is the whole point: a sterile worker or an exposed alarm-caller can lose on direct fitness and the behaviour still spreads, because selection acts on the allele's total representation across all copies, not on any one carrier's offspring. Reading kin selection as something that benefits the actor's own reproduction inverts the mechanism — the vehicle may be sacrificed precisely so the replicator wins.
  • Not group selection. Both produce cooperation, which is why they are perennially conflated, but kin selection grounds the advantage in identity-by-descent and a calculable r between actor and recipient, whereas group-selectionist accounts leave relatedness implicit and locate selection on the group as the unit. What makes a behaviour kin-selected is that benefits flow non-randomly to carriers of the same allele, measured by pedigree — not that they accrue to a co-residing collective.
  • Not reciprocal altruism. Helping that is repaid through expected future return runs on memory, recognition, and enforcement, with no requirement that helper and helped share genes; kin selection runs on shared genes, with no requirement of repayment. When measured r is near zero but cooperation persists, that is a signal the mechanism is reciprocity (or manipulation, or by-product mutualism), not kin selection misfiring.
  • Not a claim that altruism always evolves among relatives. Hamilton's rule is an inequality, not a guarantee: the behaviour is favoured only when summed rB exceeds C. High relatedness with a small benefit or a large cost leaves the arithmetic short, and the trait is selected against — relatedness raises the ceiling on what altruism can clear, it does not mandate that any given act clears it.
  • Not dependent on conscious kinship or deliberate recognition. The discrimination that aims help at carriers can be achieved by philopatry, limited dispersal, or population viscosity alone — the actor need not identify, recognise, or intend to help kin. Treating kin selection as requiring genealogical awareness mistakes one possible discrimination channel (active kin recognition) for the mechanism, which only needs benefits to land non-randomly on relatives by whatever means.

Scope of Application

Kin selection lives across the social-evolution subfields of evolutionary biology; its reach is within that domain, wherever a heritable trait alters the reproductive success of genealogical relatives. The cross-domain analogues (human nepotism, cultural in-group favouritism) belong to the broader inclusive-fitness pattern, not here.

  • Social-insect biology — the founding application. Worker sterility and reproductive division of labour in haplodiploid Hymenoptera (bees, wasps, ants) are explained by sisters sharing three-quarters of alleles by descent, which raises rB enough that helping the queen produce sisters beats producing own offspring.
  • Behavioural ecology of vertebrates — alarm calling (Belding's ground squirrels), allomothering, helpers-at-the-nest in cooperatively breeding birds and mammals, and reduced within-kin aggression are all read as inclusive-fitness payoffs once audience or recipient relatedness is measured.
  • Sociomicrobiology / microbial ecology — maintenance of costly public goods (siderophores, biofilm matrix, quorum-sensed secretions) in clonal populations where r approaches one within a clone, and the breakdown of cooperation when cheats lower local relatedness.
  • Plant ecology — sibling recognition expressed as restraint of root competition among related neighbours, an rB > C calculus applied to resource-allocation rather than behaviour.
  • Theoretical / mathematical evolution — Hamilton's rule, its Price-equation generalisation (Queller, Frank), and the kin-vs-group-selection equivalence debates, where the relatedness coefficient and inclusive-fitness accounting are the formal objects.
  • Genomic conflict and intragenomic selection — selfish genetic elements, genomic imprinting, and meiotic drive analysed through relatedness asymmetries between maternally and paternally derived alleles, the inclusive-fitness ledger applied below the level of the whole organism.

Clarity

Naming kin selection dissolves the apparent paradox that pre-Hamiltonian biology could not state cleanly: how does self-sacrificing behaviour — sterile workers, alarm calls that draw a predator's attention to the caller — survive a natural selection that should ruthlessly cull anything depressing an individual's own reproduction? The confusion lived in an unexamined assumption that the individual organism is the thing whose fitness selection maximises. Re-anchoring the accounting on the allele, and replacing individual fitness with inclusive fitness, makes altruism stop looking like a counterexample to Darwinism and start looking like one of its predictions: the behaviour is favoured precisely when rB > C, so what looked like a hole in the theory becomes a region of its solution space.

The reframing also sharpens distinctions the older vocabulary blurred. It separates the vehicle (the individual, who may lose) from the replicator (the allele, which can win across all its copies), so "good for the organism" and "favoured by selection" are no longer treated as one thing. It pries kin selection apart from group selection — both produce cooperation, but only kin selection grounds it in identity-by-descent and a calculable r, where group-selectionist talk left relatedness implicit. And it converts a vague question — "is this behaviour altruistic?" — into three measurable ones a field worker can actually pursue: what is the relatedness r between actor and recipients, what is the benefit B to them, what is the cost C to the actor? The sharper question is no longer whether altruism can evolve but whether the arithmetic clears in this case — and, when it does not, whether the behaviour is being misdescribed, the relatedness mismeasured, or some non-kin mechanism (reciprocity, manipulation, by-product mutualism) is doing the work that r was assumed to do.

Manages Complexity

The cases kin selection covers — worker sterility in haplodiploid Hymenoptera, alarm calling in ground squirrels, public-goods cooperation in clonal microbes, root-competition restraint among sibling plants — are each ecologically and genetically idiosyncratic, with their own pedigrees, recognition cues, and fitness ledgers. Hamilton's rule compresses that sprawl to a single inequality, rB > C, evaluated over just three quantities: the relatedness coefficient, the benefit to recipients, and the cost to the actor. Instead of re-deriving, for each new behaviour, whether self-sacrifice could survive selection, the biologist reads the qualitative outcome — the trait spreads or it does not — off the sign of rB − C, with the genetic system fixing r (one-half for outbred siblings, three-quarters for haplodiploid sisters, near one in a clone) and field measurement supplying B and C. The high-dimensional question "which altruistic behaviours can evolve, and why this one here" collapses to tracking three scalars and one threshold, and the same calculus re-applies unchanged across genetic systems by swapping only the relatedness coefficient. When the arithmetic fails to clear, the rule also tells the analyst where to look next — a mismeasured r, a misdescribed cost, or a non-kin mechanism doing the work — so even its disconfirmations are channelled into a small, structured set of follow-up checks rather than an open-ended search.

Abstract Reasoning

Hamilton's rule turns the inequality into a working inference engine, and the characteristic moves all run by treating r, B, and C as the only quantities that matter and reading conclusions off their relationship.

The interventionist move is the rule's signature: to predict whether an altruistic trait will spread, decline, or sit at equilibrium, the biologist does not model the whole social system but asks whether rB exceeds C and reasons from a shift in any one term. Raise the relatedness of the social neighbourhood — by enforcing philopatry, sharpening kin recognition, increasing population viscosity — and the same behaviour that was selected against can become favoured, because the r term rises and rB crosses C; the predicted effect is a higher equilibrium frequency of the altruistic allele. Lower relatedness — by mixing the population, disrupting kin structure, introducing immigrants — and cooperation should erode toward the cheat-favouring outcome even with the benefit and cost unchanged. The move reasons FROM a manipulation of one scalar (whom the behaviour is aimed at, how costly it is, how much it helps) TO a directional prediction about the trait's fate, and it is what makes kin selection prescriptive rather than merely descriptive: it tells the experimentalist which knob to turn to flip the sign of selection.

The diagnostic move runs the rule backwards from an observed behaviour to a hidden quantity. Confronted with a costly act of helping that nonetheless persists, the analyst infers that one of the three terms must make the arithmetic clear, and treats the behaviour as a measurement of whichever term is unknown: if relatedness and cost are estimable, persistent helping implies the benefit B is large enough that rB > C; if benefit and cost are known, the behaviour licenses an inference about the relatedness structure — that recipients are more related to the actor than they appear, or that the population is more viscous than a random-mixing model assumed. Belding's ground squirrels are read this way: the fact that alarm calls track the relatedness of the surrounding audience is taken as evidence that the calling allele is under inclusive-fitness selection, the audience composition serving as a read-out of the r the caller is responding to. The move reasons FROM a surface behavioural pattern (who helps whom, how often, in what social context) TO the value or sign of an underlying genetic-arithmetic term.

A distinctive failure-attribution move is built into the same arithmetic and is one of the concept's most-used inferences. When a behaviour that looks altruistic does not satisfy rB > C on the measured numbers, the rule channels the discrepancy into a short, structured set of suspects rather than an open-ended search: either r was mismeasured (the genealogy is wrong, or the relevant relatedness is the genotype's local identity-by-descent, not the pedigree average), or B and C were misassigned (the act is less costly or more beneficial than it appeared, or the currency is wrong), or — the decisive branch — kin selection is not the operative mechanism and some non-kin process (reciprocal altruism, manipulation, by-product mutualism, enforcement) is producing the cooperation that r was assumed to explain. The move reasons FROM a failed inequality TO a ranked list of where the account broke, and it is precisely how the field distinguishes genuine kin-selected altruism from cooperation that merely resembles it.

The boundary-drawing move fixes when the calculus applies at all and with which relatedness coefficient. The rule is in force only where the action is directed non-randomly toward carriers — where kin recognition, philopatry, or viscosity keeps benefits from leaking to non-carriers; absent that discrimination, r collapses toward zero, rB falls below C for any costly act, and the behaviour is outside the regime the rule explains. The coefficient itself is set by the genetic system, and selecting it correctly is a reasoning move with teeth: outbred diploid siblings take r = ½, haplodiploid sisters take r = ¾ (which raises rB enough that sister-raising can beat own-reproduction and is the inference that cracked the worker-sterility puzzle), a clone takes r ≈ 1. Reasoning FROM the genetic architecture and the spatial-social structure TO the right value of r and to whether the inequality is even the correct tool is what keeps the calculus from being misapplied to cooperation among unrelated individuals, where it would predict nothing.

Finally, the shift from individual to inclusive fitness licenses a re-attribution move on apparent counterexamples: a behaviour that depresses the actor's own offspring production is not evidence against adaptation but a candidate for inclusive-fitness explanation, and the analyst reasons FROM "this lowers the vehicle's direct fitness" TO "check whether it raises the replicator's total representation across relatives" — converting what older individual-level reasoning would file as maladaptive or paradoxical into a region of the theory's solution space where the arithmetic is expected to clear.

Knowledge Transfer

Within evolutionary biology kin selection transfers as mechanism, fully and without translation: Hamilton's rule is one inequality applied across genetic systems by swapping only the relatedness coefficient, so the diagnostics, the interventions, and the vocabulary carry intact from one substrate to the next. In behavioural ecology it explains alarm calling, allomothering, reduced within-kin aggression, and helper-at-the-nest cooperation in vertebrates; in social-insect biology it cracks worker sterility in haplodiploid Hymenoptera by raising sister-relatedness to three-quarters; in microbial ecology and sociomicrobiology it governs the maintenance of costly public goods (siderophores, biofilm matrix, quorum-sensed cooperation) in clonal populations where r approaches one; in plant ecology it frames sibling root-competition restraint. Across all of these the same three measured quantities — r, B, C — and the same failure-attribution checklist (mismeasured relatedness, misassigned cost/benefit, or a non-kin mechanism such as reciprocity or manipulation doing the work) apply unchanged. What travels is not an analogy between these cases but a single calculus instantiated in each, because every one of them literally has alleles, identity-by-descent, and differential reproduction. The portability is the mechanism, and its limit within the domain is just the regime condition: where the action is not directed non-randomly toward carriers (no kin recognition, no philopatry, no population viscosity), r collapses toward zero and the rule correctly predicts that no costly altruism evolves.

Beyond evolutionary biology the transfer changes character, and honesty requires distinguishing two things the surface conflates. Cross-domain invocations of "kin selection" — human nepotism, ethnic favouritism, organisational in-group bias, memes spreading through related groups — are analogy (case A): they rename the components (allele → cultural trait or shared identity, relatedness-by-descent → felt kinship or group membership, offspring equivalents → some looser payoff) and borrow the shape of the rB > C story, but drop the machinery that gives the original its predictive sharpness. The relatedness coefficient that makes Hamilton's rule a calculation rather than a slogan is identity-by-descent in a pedigree; replace it with "cultural relatedness" or "in-group identity" and r is no longer measurable from genealogy, the currency B/C is no longer offspring equivalents, and the inequality stops licensing the directional predictions and failure-attributions that are the whole point. These analogues are contested precisely because the substrate substitution smuggles in the very assumptions the genetic version got for free, and they need their own machinery — reciprocity, reputation, cultural-transmission models, strategic game theory — to do the explanatory work, at which point it is that machinery, not kin selection, supplying the content.

There is, however, a more general pattern that does travel (case B), and naming it correctly keeps the analogy honest rather than mistaken. Underneath kin selection sits a thinner, substrate-neutral structure — a costly behaviour persists when its aggregate benefit to other bearers of the same replicating trait, discounted by the probability they bear it, exceeds the cost to the actor — which is the parent prime kin selection instantiates (the inclusive_fitness reframing of accounting from the vehicle to the replicator, itself a specialization of selection on identity-correlated cooperation). That general pattern genuinely recurs across cooperating-replicator systems, and it is what the cross-domain lesson should carry: when human or cultural cooperation tracks a shared-identity coefficient, the transferable insight is the abstract "favour bearers of your replicating trait in proportion to the chance they share it," not the named genetic concept "kin selection" with its haplodiploidy, its identity-by-descent, and its offspring-equivalent currency — all of which stay home-bound. The discipline, then, is to let the parent travel and leave the domain accent behind; calling the human case "kin selection" overclaims the mechanism, while recognizing it as a co-instance of the same abstract inclusive-fitness logic claims exactly what is warranted. (See Structural Core vs. Domain Accent for which terms lift and which do not.)

Examples

Canonical

W. D. Hamilton's two 1964 papers formalized the rule rB > C, and J. B. S. Haldane's earlier quip supplies the cleanest arithmetic: he reportedly said he would lay down his life for two brothers or eight cousins. A full sibling shares r = ½ of alleles by descent, so saving two brothers yields rB = ½ × 2 = 1, exactly offsetting the actor's cost of one life (C = 1) — break-even. A first cousin shares r = ⅛, so eight cousins give ⅛ × 8 = 1, the same threshold. Hamilton's own motivating case was worker sterility in haplodiploid Hymenoptera: because a female's father is haploid, full sisters share r = ¾, higher than the r = ½ a worker would have with her own daughters — so raising sisters can propagate a worker's alleles more efficiently than reproducing herself, which is why sterile helping evolves.

Mapped back: The self-sacrificing variant is the social allele; ½, ⅛, and ¾ are the relatedness coefficient fixed by the genetic system; lives saved versus life spent are the recipient benefit and actor cost B and C; "two brothers or eight cousins" is the Hamilton threshold where summed rB equals C. The haplodiploid ¾ is the inclusive-fitness ledger clearing where individual-fitness accounting could not.

Applied / In Practice

Paul Sherman's field study of Belding's ground squirrels (Urocitellus beldingi) in the Sierra Nevada, reported in 1977, is a classic diagnostic deployment. Sherman recorded which squirrels gave alarm calls when a predator approached — an act that draws the predator's attention to the caller. He found that females called far more often than males, and that call frequency rose with the number of genetic relatives nearby. The pattern tracks the species' social structure: females are philopatric, remaining near their birth burrows surrounded by kin, while males disperse. Callers were, in effect, warning relatives, so the costly call raises the inclusive fitness of shared alleles even at risk to the individual.

Mapped back: Female philopatry is the discrimination channel that aims the benefit non-randomly at carriers, keeping r above zero; the alarm call is the costly social allele's expression. Sherman ran the diagnostic move in reverse — reading call frequency as a measurement of the relatedness coefficient of the surrounding audience — confirming that the calling behavior sits where the Hamilton threshold is cleared.

Structural Tensions

T1: rB > C precision versus measuring the terms (a calculable rule whose inputs resist calculation). Hamilton's rule is celebrated for converting "is this altruistic?" into three measurable quantities, and r is genuinely calculable from a pedigree. But B and C — fitness benefit and cost in offspring-equivalent units — are notoriously hard to measure in the field: they require quantifying lifetime reproductive consequences of a single act against a counterfactual that never happened, often across long timescales and confounded ecological variables. The tension is that the rule's whole appeal is turning a slogan into an arithmetic, yet in practice one term is clean while the other two are estimated with wide error, so the inequality is frequently evaluated on numbers soft enough that its sign is uncertain. The precision is real in the coefficient and aspirational in the payoffs, and the rule's authority can outrun the quality of the B and C that feed it. Diagnostic: Are B and C here measured in genuine offspring-equivalents against a real counterfactual, or is the rule's arithmetic borrowing false precision from a well-estimated r over poorly-estimated payoffs?

T2: Failure-attribution discipline versus unfalsifiability (a structured checklist that can also rescue the rule from any data). When measured numbers fail rB > C for an apparently altruistic act, the rule channels the discrepancy into a short suspect list: mismeasured r, misassigned B/C, or a non-kin mechanism. This is a genuine methodological strength — disconfirmations become structured follow-ups. But the same checklist is a route to unfalsifiability: any failed inequality can be "explained" by positing that r was really higher (cryptic viscosity), the cost really lower, or the currency wrong, so the rule can be preserved against almost any observation by adjusting an unmeasured term. The tension is that the discipline which makes failures productive is the same flexibility that lets the theory absorb refutation, and distinguishing a legitimate remeasurement from a face-saving rescue requires independent evidence the failed test did not supply. Diagnostic: Is the reattribution of a failed inequality (higher hidden r, wrong currency) backed by independent measurement, or is a term being adjusted post hoc solely to make the rule clear?

T3: Vehicle-replicator separation versus the organism as real agent (the reframing that explains altruism can over-atomize the individual). Re-anchoring accounting on the allele rather than the organism is the move that dissolves the altruism paradox — the vehicle can lose while the replicator wins. But pushed hard, the gene's-eye ledger can under-describe the organism as an integrated agent whose whole genome shares an interest in its reproduction, and can invite treating every behavior as a battleground of allelic accounting even where the individual is the coherent unit selection acts on. The tension is that the vehicle-replicator distinction is indispensable for the cases it was built for (sterile workers, intragenomic conflict) yet can mislead when applied indiscriminately, since most of the time the organism's alleles are aligned and "good for the organism" and "favoured by selection" do coincide. The reframing is a scalpel that, used as a universal solvent, dissolves a real unit of biological organization. Diagnostic: Does this behavior genuinely pit the replicator's interest against the vehicle's (warranting gene-level accounting), or are aligned alleles being needlessly atomized where the whole organism is the operative unit?

T4: Discrimination channel breadth versus what counts as kin selection (viscosity without recognition can blur the mechanism's edge). The rule needs only that benefits land non-randomly on carriers, achievable by philopatry or population viscosity alone — no kin recognition or intent required. This breadth is correct and important (it frees the mechanism from requiring genealogical awareness). But it also softens the boundary with other explanations: in a viscous population, limited dispersal simultaneously raises relatedness and increases local competition among kin (which can cancel the inclusive-fitness benefit), and mere spatial structure can produce cooperation-like patterns that are hard to cleanly attribute to kin selection versus population-structure or group-level accounting. The tension is that the most inclusive reading of the discrimination channel (any non-random benefit delivery) makes the mechanism widely applicable but also hardest to distinguish from its neighbors precisely where recognition is absent. Breadth of the channel trades against sharpness of the diagnosis. Diagnostic: Is the non-random benefit delivery here producing a net rB that survives the increased kin competition viscosity also creates, or is spatial structure being credited to kin selection without netting out local competition?

T5: Autonomy versus reduction (a genetic mechanism or an instance of inclusive-fitness / cooperation among identity-correlated replicators). "Kin selection" is a specific evolutionary construct with home-bound cargo — identity-by-descent, the pedigree-calculated r, haplodiploidy's ¾, offspring-equivalent currency, the failure-attribution checklist — and within evolutionary biology it travels intact as mechanism across social insects, vertebrate behavioral ecology, sociomicrobiology, and plant ecology, which are genuine co-instances because each literally has alleles, identity-by-descent, and differential reproduction. But its portable core is the thinner parent inclusive_fitness: a costly behavior persists when its aggregate benefit to other bearers of the same replicating trait, discounted by the probability they bear it, exceeds the cost to the actor. That parent recurs across cooperating-replicator systems. Crucially, the human/cultural invocations (nepotism, in-group favoritism) are analogy, not mechanism — replacing pedigree r with "felt kinship" forfeits the measurability and directional predictions that are the whole point, and those cases need their own machinery (reciprocity, reputation, cultural transmission). Diagnostic: Resolve toward the parent (inclusive_fitness, favor bearers of your replicating trait in proportion to the chance they share it) when carrying the lesson to cultural or human cooperation; toward kin selection's identity-by-descent-and-Hamilton's-rule machinery only where actual genes and pedigree relatedness are present.

Structural–Framed Character

Kin selection sits toward the structural end of the spectrum but stops short of the pole — mixed-structural, in the same family as the island rule, island biogeography theory, and the Janzen-Connell hypothesis: a genuine, evaluatively neutral evolutionary mechanism carried in biological vocabulary. Four of the five criteria read structural. Evaluative_weight is nil: Hamilton's rule states when a costly allele spreads; "altruism," "cost," and "benefit" are technical fitness terms carrying no moral valence, and the mechanism renders no verdict — indeed its whole achievement is to show that what looked like self-sacrifice is a straightforward prediction of selection, not a virtue. Institutional_origin is none: the rB > C condition is a fact of how identity-by-descent and differential reproduction interact, not an artifact of any survey or agency — Hamilton formalized (and Maynard Smith named) a mechanism that bees, squirrels, and clonal microbes have run for as long as the genetic systems existed. It is not human-practice-bound: remove every biologist and haplodiploid workers still raise sisters, Belding's squirrels still call more amid kin, clonal microbes still maintain public goods — the mechanism runs on alleles and pedigrees, not on a judging observer. And within its proper range cross-substrate reuse is recognition, not import: from social insects to vertebrates to microbes to plants the same Hamilton's-rule calculus is recognized intact (genuine co-instances, since each literally has alleles and identity-by-descent), swapping only the relatedness coefficient.

What keeps it off the structural pole is vocab_travels, which it fails, together with the substrate-lock behind it. The operative vocabulary — relatedness coefficient by descent, haplodiploidy, offspring-equivalents, inclusive-fitness ledger, philopatry — is irreducibly evolutionary-genetic and floats free of no non-genetic substrate the way "favour bearers of your replicating trait in proportion to the chance they share it" does; beyond biology the human/cultural invocations (nepotism, in-group favouritism) are contested analogy that forfeit the measurable pedigree r and offspring currency, so the transfer there is not mechanism. The portable structural skeleton is inclusive fitness: a costly behaviour persists when its aggregate benefit to other bearers of the same replicating trait, discounted by the probability they bear it, exceeds the cost to the actor. That skeleton is genuinely substrate-portable across cooperating-replicator systems, but it is exactly what kin selection instantiates from its umbrella inclusive_fitness, not what makes "kin selection" itself travel: the cross-domain reach belongs to that inclusive-fitness parent, while the identity-by-descent-and-Hamilton's-rule machinery stays home. Its character: structural in skeleton — a real, evaluatively neutral, recognized-in-nature replicator-accounting mechanism — but stated in a relatedness-by-descent vocabulary that pins it to the genetic substrate, leaving it mixed-structural rather than a free-floating prime.

Structural Core vs. Domain Accent

This section decides why kin selection is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity in one place.

What is skeletal (could lift toward a cross-domain prime). Strip the genetics and a thin relational structure survives: a costly behaviour persists when its aggregate benefit to other bearers of the same replicating trait, discounted by the probability they bear it, exceeds the cost to the actor — accounting run on the replicator's total representation, not the individual vehicle. The portable pieces are abstract — a replicating trait, a shared-bearing coefficient, a benefit to co-bearers and a cost to the actor, and a threshold on their discounted sum. That skeleton is genuinely substrate-portable across cooperating-replicator systems — which is exactly why the entry instantiates the catalog's inclusive_fitness (the vehicle-to-replicator reframing, itself a specialization of selection on identity-correlated cooperation). That recurrence is mechanism, but it is the core kin selection shares, not what makes it distinctive.

What is domain-bound. Everything that makes the mechanism kin selection in particular is evolutionary-genetic furniture. The shared-bearing coefficient is specifically r, the relatedness by identity-by-descent fixed by pedigree and genetic system — ½ for outbred siblings, ¾ for haplodiploid sisters, ≈1 in a clone; the currency of B and C is offspring-equivalents; the discrimination channel is kin recognition, philopatry, or population viscosity; and the threshold is Hamilton's rule, rB > C, with a structured failure-attribution checklist (mismeasured r, misassigned B/C, or a non-kin mechanism). The decisive test: replace pedigree r with "felt kinship" or "cultural relatedness" and the coefficient is no longer measurable from genealogy, the currency is no longer offspring-equivalents, and the inequality stops licensing the directional predictions and failure-attributions that are the whole point — the human/cultural invocations (nepotism, in-group favouritism) become contested analogy that need their own machinery (reciprocity, reputation, cultural transmission) to do the explanatory work. The identity-by-descent-and-Hamilton's-rule apparatus is the accent that stays home.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose transfer is recognition of the same mechanism, not analogy. Kin selection's transfer is bimodal. Within evolutionary biology it transfers as mechanism intact — social-insect biology, vertebrate behavioural ecology, sociomicrobiology, plant ecology, mathematical evolution, genomic conflict — because each literally has alleles, identity-by-descent, and differential reproduction, so the one rB > C calculus applies by swapping only the relatedness coefficient, and the diagnostics and failure-attribution checklist carry unchanged; these are genuine co-instances. Beyond biology the named mechanism does not transfer as mechanism: human nepotism and cultural in-group favouritism borrow the shape while forfeiting the measurable pedigree r and offspring currency, so calling them "kin selection" overclaims. And when the bare structural lesson is needed cross-domain — favour bearers of your replicating trait in proportion to the chance they share it — it is already carried, in more general form, by the inclusive_fitness parent kin selection instantiates, of which the human case is at most a co-instance. The cross-domain reach belongs to that inclusive-fitness parent; "kin selection," as named, packs the identity-by-descent, haplodiploidy, and Hamilton's-rule machinery that should stay home in the genetic substrate.

Relationships to Other Abstractions

Local relationship map for Kin selectionParents appear above the current abstraction, mutual partners to the right, and children below. Node labels state whether each abstraction is prime or domain-specific; colors identify relation types.Kin selectionDOMAINDomain-specific abstraction: Hamilton's Rule — is part ofHamilton's RuleDOMAINPrime abstraction: Natural Selection — is a kind ofNaturalSelectionPRIME

Current abstraction Kin selection Domain-specific

Parents (2) — more general patterns this builds on

  • Kin selection is a kind of Natural Selection Prime

    Kin selection is natural selection specialized to heritable social effects directed non-randomly toward genetic relatives.

  • Kin selection is part of Hamilton's Rule Domain-specific

    Kin selection contains Hamilton’s rule as its operational spread criterion.

Hierarchy paths (3) — routes to 2 parentless roots

Not to Be Confused With

  • Group selection. The perennially conflated rival: both produce cooperation, but group selection locates the unit of selection on the co-residing collective and leaves relatedness implicit, whereas kin selection grounds the advantage in identity-by-descent and a calculable r between actor and recipient. What makes a behaviour kin-selected is that benefits flow non-randomly to carriers of the same allele, measured by pedigree — not that they accrue to a group. (Formally the two can be shown equivalent under Price-equation bookkeeping, but the accounting objects differ.) Tell: is the advantage attributed to the group as a unit (group selection), or to benefits landing on same-allele carriers by descent (kin selection)?

  • Reciprocal altruism. Helping repaid through expected future return — running on memory, recognition, and enforcement, with no requirement that helper and helped share genes. Kin selection runs on shared genes with no requirement of repayment. When measured r is near zero yet cooperation persists, that signals reciprocity (or manipulation), not kin selection. Tell: is the cooperation sustained by expected future repayment between possibly-unrelated partners (reciprocity), or by shared alleles regardless of any payback (kin selection)?

  • Green-beard effect. A distinct assortment mechanism: a single gene (or tightly linked cluster) both produces a recognizable marker and directs help toward other bearers of that same marker — cooperation keyed to identity of the specific allele, not to genealogical relatedness. Green-beards can favour help between organisms of low pedigree relatedness, which kin selection cannot. Tell: is help directed by a self-recognizing marker allele regardless of overall kinship (green-beard), or by pedigree relatedness across the whole genome (kin selection)?

  • Inclusive fitness. The parent accounting framework — an actor's own reproduction plus the r-weighted effects on all others' reproduction — not a synonym for kin selection. Kin selection is the evolutionary process (differential allele spread) that inclusive-fitness accounting explains; inclusive fitness is the ledger. Helping non-relatives via green-beards or intragenomic conflict are inclusive-fitness effects that are not kin selection in the pedigree sense. Tell: is the referent the accounting reframing from vehicle to replicator (inclusive fitness), or the specific process of a costly allele spreading by aiding genealogical kin (kin selection)?

  • Hamilton's rule (rB > C). The formal criterion, not the mechanism itself: the inequality whose sign decides whether the trait is favoured. Kin selection is the process the rule adjudicates; treating "Hamilton's rule" as identical to "kin selection" conflates the test with the phenomenon (and the rule, suitably generalized via the Price equation, also scores non-kin and group cases). Tell: is the referent the inequality that determines the trait's fate (Hamilton's rule), or the evolutionary process of aiding kin it evaluates (kin selection)?

  • Nepotism / cultural in-group favouritism (the cross-domain analogy). Human or cultural favouring of "one's own" — genuine phenomena, but analogy to kin selection, not the mechanism: they replace pedigree r with felt kinship or group identity and offspring-equivalents with looser payoffs, forfeiting the measurability and directional predictions that make Hamilton's rule a calculation. These need their own machinery (reciprocity, reputation, cultural transmission) and are at most co-instances of the broader inclusive_fitness parent. Tell: is there actual genetic identity-by-descent and offspring-equivalent currency (kin selection proper), or a shared-identity coefficient standing in for it (a cultural analogy — carry the parent, not "kin selection")?

Neighborhood in Abstraction Space

Kin selection sits in a crowded region of the domain-specific corpus (9th percentile for distinctiveness): several abstractions share nearly its structure, so a description that fits it tends to fit its neighbors too.

Family — Population Genetics & Kin Selection (10 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12