Proc. Natl. Acad. Sci. USA Vol. 77, No. 7, pp. 4382-4386, July 1980 Population Biology Translation of epigenetic rules of individual behavior into ethnographic patterns (human sociobiology/social development/gene-culture coevolution) CHARLES J. LUMSDEN AND EDWARD 0. WILSON Museum of Comparative Zoology Laboratories, Harvard University, Cambridge, Massachusetts 02138 Contributed by Edward 0. Wilson, April 28, 1980 ABSTRACI The pivotal process in gene-culture coevolution Pure is envisaged to be the evolution of behavioral epigenesis. From genetic premises based on the known properties of enculturation and usage diffusion within societies, a probabilistic model is con- to._ structed to estimate the degree to which rules governing indi- E Gene-culture vidual development canalize ethnographic curves (the proba- bility density distributions of societies engaged in varying ._I'0 patterns of usage). The results indicate that under most con- Cd ceivable conditions the translation from individual epigenesis 4a to social pattern is amplified, to an extent that differences in la bias too faint to be detected in ordinary developmental studies .0 can generate conspicuous variation in the ethnographic curves. .- Examples are cited of sufficiently biased epigenesis in human 0 behavior. co 0 Pure cultural During socialization in a cultural species, individuals transmit an array of behaviors and artifacts, which for convenience will -1v be termed "culturgens" [L. cultur(a), culture + L. gen(o), produce]. This unit can be consistently delineated by applying the set-theoretic definition of artifact employed in quantitative 0 12 3 4 5 6 7 8 9 101112 1314 1516 17 18 archeological studies (1): A relatively homogeneous set of ar- Culturgens tifacts or behaviors that either share without exception one or FIG. 1. Three conceivable classes of programs of informational a selected set of important attribute states or at least share a transmission in a social species. consistently recurrent range of attribute states within a given polythetic set. Enculturation can theoretically consist of one or It is apparent from the data of cross-cultural studies that the other of three forms (Fig. 1): (i) Pure genetic transmission, many epigenetic rules find expression in the ultimate patterns in which all members are genetically constrained to learn one of social behavior. Examples especially noted by previous in- culturgen within a given category of alternative culturgens (thus vestigators include features of verbal color classification (14) it is possible to have a purely genetic culture); (ii) pure cultural and brother-sister incest avoidance (15, 16). We propose a transmission, in which no innate predisposition exists favoring general theoretical construction that focuses on behavioral one culturgen over another; or (iii) gene-culture transmission, epigenesis as the product of genetic prescription, which in turn in which one culturgen is favored because of innate bias in the is altered by natural selection acting through the epigenetically teaching or acquisition processes. determined cultural patterns. The key questions of gene-culture Transmission is subject to a sequence of epigenetic rules, coevolution then become the degree to which epigenesis im- which are the genetically determined peripheral sensory filters, pacts cultural evolution and the degree to which natural se- interneuron coding processes, and more centrally located lection acting on culture affects the genes committed to be- procedures of biased learning that affect the probability of havioral epigenesis. The first step, which we explore in this acquiring one culturgen as opposed to another. Epigenetic rules article, is to identify the cognitive epigenetic rules, which that produce the gene-culture form of transmission are wide- hitherto have been the concern of psychology, and to link them spread, if not general, in human behavior. They have been to the ethnographic patterns, which have been the concern of demonstrated in newborn taste preferences (2, 3), discretization cultural anthropology. Such a procedure, if successful, could of hue perception (4), phoneme production (5), preferred level render ethnographic patterns predictable from first principles of visual pattern complexity (6), preferred visual design (7), and allow a more precise evaluation of the flexibility of human general facial recognition (8), basic facial expressions (9), social behavior. mother-infant bonding (10), infant carrying (11), fear of strangers and other generalized fears and phobias (9, 12), METHODS brother-sister incest avoidance (13), and others. The new coevolutionary schema that we propose is faithful to known of transmission within societies. The publication costs of this article were defrayed in part by page the principles culturgen charge payment. This article must therefore be hereby marked "ad- In 80% or more of societies, enculturation is conducted not just vertisement" in accordance with 18 U. S. C. §1734 solely to indicate by the nuclear family, a common feature of some industrialized this fact. Western societies, but by a much broader array of relatives and 4382 Downloaded by guest on October 1, 2021 Population Biology: Lumsden and Wilson Proc. Natl. Acad. Sci. USA 77 (1980) 4383. Model Qualitative choice strctures u11(() Analytic description Threshold U12 020+G LI 12 > f response U21 all b2 f bCf4C -1 , +1 -1 0 +1 Soturotable 01402 --.blb2 U12 0 tnh l f | 2 'trendwatching' ° °U12 U2lLI 00-a -- bl-b2 .1 u21 b,..b2 tanhC Nonsaturitable u12 = aO+02C watching i UA u21 U2 a3e Preference for partial bujt 0(sinbOr2U12-= a+a02si~ not total U12 U21 consensusL . I1 i ai-02 b:b~2 b,+b2 sin (w'(1+C)] High sensitivit 0LI 12 00+01 ( 02(1e to minority U12 u21 1 opinlion U21 bo b £Ile b2( FIG. 2. The conceivable choice structures uij() selected for analysis. parent-surrogates. This circumstance insures a relatively uni- means of exponential probability densities: form exposure to culturgens and transmission of dominant values (17). The spreading patterns of about 20 kinds of suc- Di(t) =-e-t/Tl, t > 0 [la] cessful culturgens that have been studied by social scientists are well described by logistic and Gompertz equations. In the latter case, which 1 provides the closer fit, the general form of the D2(t) -.t/72Tt20., [lb] equation can be derived from elementary assumptions con- T2 cerning imitation, evaluation, and culturgen substitution as rates process are v0, = functions of the levels of general usage in the society (18). The transition of the decision where vj riuq, i ?6 j. Consider a system in which two alternative culturgens, cl and In the two-culturgen case, where ni members possess cl and C2, are the options of interest. This is a common circumstance n2 members possess c2, the a in ethnography and modern decision making, and even com- probability P(nl,n2,t) that steady-state society of N members possesses a particular com- plex arrays of alternative culturgens often can be clustered into bination of + n2 = N at rate binary sets for purposes of analysis (1, 19, 20). Individual nI culturgens changes the members of the society choose between cl and C2 at decision atP(n,,n2,t) -- points; they can decide to retain their active culturgen (1 1 = (ni + 1) v12(nlI + 1,n2-1) * P(ni + 1,n2-1,t) or 2 - 2) or to change (1 2 or '2 -p 1). In many cases of practical intent, such decision and learning processes are ade- + (n2 + 1) V21(nl-1,n2 + 1) * P(ni-1,n2 + 1,t) quately described by a Markov process as a first approximation, - [n * Vi2(nfi,n2) + n2V21(n1,n2)]P(nj,n2,t) [2] as noted by others who have modeled group behavior (21). We for 0 < ni < N. The probability density distribution suggest that a similar approach is a useful starting point in the of analysis of epigenesis-ethnography translation. In particular, culturgen frequencies at any given time can be called the ethnographic curve of the population. the epigenetic rules can be expressed in abstract form as the It is more efficient to write the culturgen proportions as a transition probabilities u0j from culturgen i to culturgen j at each decision point. single variable ' The mean lifetime between sequential decision points - isr, . (n2 nl)IN [3] for a ci-user and T2 for a c2-user, and the rate constants are ri = 1/rT and r2 = 1/T2. The parameters Tr and T2 are then the which ranges from -I to +1 as n2 ranges from 0 to N. The Downloaded by guest on October 1, 2021 4384 Population Biology: Lumsden and Wilson Proc. Natt. Acad. Sci. USA 77 (1980) l.r RESULTS 0.8 a, =0.495 Structure of the Assimilation Functions. The categories of a2 = 0.0 assimilation that we have evaluated are summarized in Fig. 2. .W 0.6 A,- Few data exist that can be used to derive real curves. The most 0.4hI elementary straight-line function (vij insensitive to t) and step functions will result in relatively uniform, easily predictable, 0.2 species-specific traits. They are most likely to occur during early infancy, when the most robust epigenetic rules direct behavior -1.0 -0.5 0.0 +0.5 +1.0 to certain limited and virtually inevitable choices. Apparent 0.10 _ al = 0.495 examples include certain broad categories of taste preferences a2 = 1.2 (2, 3), the major categories of color perception and vocabulary 0.08 - a= 0.505 (4), and basic facial expressions (9). The monotonic, "trend- a 0.06 watching" cases are also likely to occur. The existence of curves of this general form are implied by substantial empirical data 0.04 on intragroup cultural diffusion in Western industrial societies 0.02 (18, 24) and data on small-group behavior (25, 26). i i Amplification to the Ethnographic Patterns. When dvqj/dt 1.0 -0.5 0.0 +0.5 +1.0 = 0 on (-1,1), in other words when each group member op- erates independently of the others, interior maxima or minima 1.0r a - 0.495 in the ethnographic curve P(s) are determined by e2V(Q), where 0.8_ 2.0 a,a2-= 0.505 V(t) is the integral in Eq.