ArticlePDF Available

The Psychological Foundations of Culture

January 1992

January 1992

Authors:

Jerome H. Barkow

Dalhousie University

Leda Cosmides

University of California, Santa Barbara

Content uploaded by Jerome H. Barkow

Content may be subject to copyright.

The Psychological Foundations of Culture

JOHN TOOBY AND LEDA COSMIDES

INTRODUCTION: THE UNITY OF SCIENCE

One of the strengths of scientiﬁc inquiry is

that it can progress with any mixture of em-

piricism, intuition, and formal theory that

suits the convenience of the investigator.

Many sciences develop for a time as ex-

ercises in description and empirical gener-

alization. Only later do they acquire rea-

soned connections within themselves and

with other branches of knowledge. Many

things were scientiﬁcally known of human

anatomy and the motions of the planets be-

fore they were scientiﬁcally explained.

-GEORGE WILLIAMS, Adaptation and Nat-

ural Selection

Disciplines such as astronomy, chemistry, physics,

geology, and biology have developed a robust com-

bination of logical coherence, causal description, ex-

planatory power, and testability, and have become

examples of how reliable and deeply satisfying hu-

man knowledge can become. Their extraordinary

ﬂorescence throughout this century has resulted in

far more than just individual progress within each

ﬁeld. These disciplines are becoming integrated into

an increasingly seamless system of interconnected

knowledge and remain nominally separated more

out of educational convenience and institutional in-

ertia than because of any genuine ruptures in the un-

derlying unity of the achieved knowledge. In fact,

this development is only an acceleration of the pro-

cess of conceptual uniﬁcation that has been build-

ing in science since the Renaissance. For exam-

ple, Galileo and Newton broke down the then rigid

(and now forgotten) division between the celestial

and the terrestrial-two domains that formerly had

been considered metaphysically separate-showing

that the same processes and principles applied to

both. Lyell broke down the distinction between the

static present and the formative past, between the

creative processes operating in the present and the

geological processes that had operated across deep

time to sculpt the earth. Maxwell uncovered the ele-

gant principles that uniﬁed the many disparate elec-

trical and magnetic phenomena into a single system.

And, one by one, the many gulfs separating life

from nonlife were bridged and then closed: Har-

vey and others found that the macrostructure of the

body turned out to operate according to comprehen-

sible mechanical principles. Wohler’s synthesis of

urea showed that the chemistries of the living and

the nonliving were not forever separated by the oc-

cult operation of special vitalistic forces. In Wohler’s

wake, the unraveling of the molecular biology of

the gene and its regulation of cellular processes has

shown how many of the immensely complex and

functionally intricate mechanisms that constitute life

are realized in molecular machinery: the élan vital

turned out to be nothing other than this microscopic

functional intricacy. Most critically, Darwin showed

how even the intricately articulated functional or-

ganization of living systems (then only observable

at the macroscopic level) could be explained as the

product of intelligible natural causes operating over

the expanse of deep time. In so doing, he conceptu-

ally united the living and the nonliving into a single

system of principled causation, and the entire diver-

sity of plant, animal, and microbial species into a sin-

gle tree of descent. Darwin took an equally radical

step toward uniting the mental and physical worlds,

by showing how the mental world-whatever it might

be composed of arguably owed its complex organi-

zation to the same process of natural selection that

explained the physical organization of living things.

Psychology became united with the biological and

hence evolutionary sciences.

The rise of computers and, in their wake, modern

cognitive science, completed the conceptual uniﬁca-

tion of the mental and physical worlds by showing

how physical systems can embody information and

meaning. The design and construction of artiﬁcial

computational systems is only a few decades old, but

already such systems can parallel in a modest way

cognitive processes-such as reason, memory, knowl-

edge, skill, judgment, choice, purpose, problem-

solving, foresight, and language-that had supposedly

made mind a metaphysical realm forever separated

from the physical realm, and humans metaphysically

disconnected from the causal network that linked to-

gether the rest of the universe. These intellectual

advances transported the living, the mental, and the

human-three domains that had previously been dis-

connected from the body of science and mystiﬁed

because of this disconnection-into the scientiﬁcally

analyzable landscape of causation.

One useful way to organize this knowledge is as

a principled history of the universe. Starting with

some characterizable initial condition (like the Big

Bang), each successive state of the system is de-

scribed, along with the principles that govern the

transitions from state to state. To the extent that

our scientiﬁc model is well developed, we should be

able to account for the types of entities that emerge

(pulsars, tectonic plates, ribosomes, vision, incest

avoidance) and their distribution and location in the

causal matrix. Such a history-in its broadest outlines-

is well on its way to being constructed, from an ini-

tial quantum state, to the formation and distribu-

tion of particles during the early expansion, to the

cooling and formation of atoms, the formation of

galaxies, stellar evolution, the synthesis of heavier

nuclei, and, of parochial interest to us, the local his-

tory of the solar system. This includes the formation

of the sun and planets; the geochemistry of prebi-

otic earth; the generation of complex organic com-

pounds; the emergence of the initial ancestral repro-

ducing chemical system; the evolution of the genetic

code and prokaryotic design; the emergence of eu-

karyotic sexual organisms, multicellular plants, ani-

mals, and fungi; and the rest of the history of life on

earth.

In this vast landscape of causation, it is now possi-

ble to locate "Man’s place in nature" to use Huxley’s

famous phrase and, therefore, to understand for the

ﬁrst time what humankind is and why we have the

characteristics that we do. From this vantage point,

humans are self-reproducing chemical systems, mul-

ticellular heterotrophic mobile organisms (animals),

appearing very late in the history of life as some-

what modiﬁed versions of earlier primate designs.

Our developmental programs, as well as the physi-

ological and psychological mechanisms that they re-

liably construct, are the natural product of this evo-

lutionary history. Human minds, human behavior,

human artifacts, and human culture are all biologi-

cal phenomena-aspects of the phenotypes of humans

and their relationships with one another.

The rich complexity of each individual is produced

by a cognitive architecture, embodied in a physio-

logical system, which interacts with the social and

nonsocial world that surrounds it. Thus humans,

like every other natural system, are embedded in

the contingencies of a larger principled history, and

explaining any particular fact about them requires

the joint analysis of all the principles and contin-

gencies involved. To break this seamless matrix of

causation-to attempt to dismember the individual

into "biological" versus "nonbiological" aspects-is to

embrace and perpetuate an ancient dualism endemic

to the Western cultural tradition: material/spiritual,

body/ mind, physical/mental, natural/human, an-

imal/human, biological/social, biological/ cultural.

This dualistic view expresses only a premodern ver-

sion of biology, whose intellectual warrant has van-

ished.

This expansive new landscape of knowledge has

not always been welcome, and many have found it

uncongenial in one respect or another. The intel-

lectual worlds we built and grew attached to over

the last 3,000 years were laid out before much was

known about the nature of the living, the men-

tal, and the human. As a result, these intellectual

worlds are, in many important respects, inconsis-

tent with this new uniﬁed scientiﬁc view and, hence,

are in need of fundamental reformulation. These

established intellectual traditions and long-standing

habits of mind seem, to many, to be more nourish-

ing, more comfortable and, therefore, more valu-

able than the alternative prospect of new and un-

familiar scientiﬁc knowledge. To pick a single exam-

ple, the shift from a universe designed to embody

amoral and spiritual order to a universe that is un-

designed and is structured only by a causal order

engendered an immeasurably greater cultural dislo-

cation than that which occurred when Copernicus

identiﬁed the sun rather than the earth as the cen-

ter of the planetary orbits. Consequently, the de-

mystiﬁcations that have taken place since 1859 have

been painful and have precipitated considerable re-

sistance to accepting these discoveries and their im-

plications. With the appearance of Darwinism, the

full scope of the emerging uniﬁed account was, for

the ﬁrst time, apparent. Therefore, much of the op-

position has speciﬁcally revolved around evolution

and its application to humans. Gladstone, for exam-

ple, in a debate with Huxley, captured in his choice

of language the widely shared, visceral sense of re-

vulsion caused by the claim "that natural selection

and the survival of the ﬁttest, all in the physical or-

der, exhibit to us the great arcanum of creation, the

sun and the center of life, so that mind and spirit are

dethroned from their old supremacy, are no longer

sovereign by right, but may ﬁnd somewhere by char-

ity a place assigned them, as appendages, perhaps

only as excrescences, of the material creation" (Glad-

stone, quoted in Gould, 1988, p. 14 ). The dislo-

cations in worldview stemming from this process of

conceptual uniﬁcation led to a growing demand for,

and production of, conceptual devices and rationales

to divorce the natural sciences from the human so-

cial and inner landscape, to blunt the implications

of monism and Darwinism, and to restore a comfort-

able distance between the human sciences and the

world of natural causation. To many scholarly com-

munities, conceptual uniﬁcation became an enemy,

and the relevance of other ﬁelds a menace to their

freedom to interpret human t reality in any way they

chose. Il Thus, despite some important exceptions,

the social sciences have largely kept I i themselves

isolated from this crystallizing process of scientiﬁc

integration. Although social scientists imitated many

of the outward forms and practices of natural scien-

tists (quantitative measurement, controlled observa-

tion, mathematical models, experimentation, etc.),

they have tended to neglect or even reject the cen-

tral principle that valid scientiﬁc knowledge-whether

from the same or different ﬁelds-should be mutually

consistent (see Cosmides, Tooby, & Barkow, this vol-

ume). It is this principle that makes different ﬁelds

relevant to each other, and part of the same larger

system of knowledge. In consequence, this insularity

is not just an accident. For many scholars, it has been

a conscious, deeply held, and strongly articulated po-

sition, advanced and defended since the inception of

the social sciences, particularly in anthropology and

sociology. Durkheim, for example, in his Rules of

the Sociological Method, argued at length that so-

cial phenomena formed an autonomous system and

could be only explained by other social phenomena (

1895/ 1962). The founders of American anthropol-

ogy, from Kroeber and Boas to Murdock and Lowie,

were equally united on this point. For Lowie, "the

principles of psychology are as incapable of account-

ing for the phenomena of culture as is gravitation

to account for architectural styles," and "culture is

a thing sui generis which can be explained only in

terms of itself. ...Omnis cultura ex cultura" ( 1917/

1966, p. 25-26; p. 66). Murdock, in his inﬂuen-

tial essay "The science of culture," summed up the

conventional view that culture is "independent of the

laws of biology and psychology" ( 1932, p. 200).

Remarkably, while the rest of the sciences have

been weaving themselves together through acceler-

ating discoveries of their mutual relevance, this doc-

trine of intellectual isolationism, which has been the

reigning view in the social sciences, has only be-

come more extreme with time. With passionate ﬁ-

delity, reasoned connections with other branches of

knowledge are dismissed as ignorant attempts at

crude reductionism, and many leading social scien-

tists now openly call for abandoning the scientiﬁc

enterprise instead. For example, Clifford Geertz ad-

vocates abandoning the ground of principled causal

analysis entirely in favor of treating social phenom-

ena as "texts" to be interpreted just as one might in-

terpret literature: We should "turn from trying to ex-

plain social phenomena by weaving them into grand

textures of cause and effect to trying to explain them

by placing them into local frames of awareness" (

1983, p. 6). Similarly, Edmund Leach rejects scien-

tiﬁc explanation as the focus of anthropology: "So-

cial anthropology is not, and should not aim to be,

a ’science’ in the natural science sense. If anything

it is a form of art Social anthropologists should not

see themselves as seekers after objective truth. ..."

(Leach, 1982, p. 52). These positions have a grow-

ing following, but less, one suspects, because they

have provided new illumination than because they

offer new tools to extricate scholars from the unwel-

come encroachments of more scientiﬁc approaches.

They also free scholars from all of the arduous tasks

inherent in the attempt to produce scientiﬁcally valid

knowledge: to make it consistent with other knowl-

edge and to subject it to critical rejection on the ba-

sis of empirical disproof, logical inconsistency, and

incoherence. In any case, even advocates of such

avenues of retreat do not appear to be fully serious

about them because few are actually willing to ac-

cept what is necessarily entailed by such a stance:

Those who jettison the epistemological standards of

science are no longer in a position to use their in-

tellectual product to make any claims about what

is true of the world or to dispute the others’ claims

about what is true.

Not only have the social sciences been unusual

in their self-conscious stance of intellectual autarky

but, signiﬁcantly, they have also been relatively un-

successful as sciences. Although they were founded

in the 18th and 19th centuries amid every expecta-

tion that they would soon produce intellectual dis-

coveries, grand "laws," and validated theories to ri-

val those of the rest of science, such success has

remained elusive. The recent wave of antiscien-

tiﬁc sentiment spreading through the social sciences

draws much of its appeal from this endemic fail-

ure. This disconnection from the rest of science has

left a hole in the fabric of our organized knowledge

of the world where the human sciences should be.

After more than a century, the social sciences are

still adrift, with an enormous mass of half -digested

observations, a not inconsiderable body of empiri-

cal generalizations, and a contradictory stew of un-

grounded, middle-level theories expressed in a ba-

bel of incommensurate technical lexicons. This is ac-

companied by a growing malaise, so that the single

largest trend is toward rejecting the scientiﬁc enter-

prise as it applies to humans.

We suggest that this lack of progress, this "failure

to thrive," has been caused by the failure of the so-

cial sciences to explore or accept their logical con-

nections to the rest of the body of science-that is,

to causally locate their objects of study inside the

larger network of scientiﬁc knowledge. Instead of

the scientiﬁc enterprise, what should be jettisoned is

what we will call the Standard Social Science Model

(SSSM): The consensus view of the nature of social

and cultural phenomena that has served for a cen-

tury as the intellectual framework for the organiza-

tion of psychology and the social sciences and the

intellectual justiﬁcation for their claims of autonomy

from the rest of science. Progress has been severely

limited because the Standard Social Science Model

mischaracterizes important avenues of causation, in-

duces researchers to study complexly chaotic and

unordered phenomena, and misdirects study away

from areas where rich principled phenomena are to

be found. In place of the Standard Social Science

Model, there is emerging a new framework that we

will call the Integrated Causal Model. This alterna-

tive framework makes progress possible by accept-

ing and exploiting the natural connections that ex-

ist among all the branches of science, using them

to construct careful analyses of the causal interplay

among all the factors that bear on a phenomenon. In

this alternative framework, nothing is autonomous

and all the components of the model must mesh.

In this chapter, we argue the following points:

I. There is a set of assumptions and inferences

about humans, their minds, and their collective

interaction-the Standard Social Science Model-that

has pro- vided the conceptual foundations of the so-

cial sciences for nearly a century and has served as

the intellectual warrant for the isolationism of the

social sciences.

2. Although certain assumptions of this model are

true, it suffers from a series of major defects that

make it a profoundly misleading framework. These

defects have been responsible for the chronic difﬁ-

culties encountered by the social sciences.

3. Advances in recent decades in a number of

different disciplines, including evolutionary biology,

cognitive science, behavioral ecology, psychology,

hunter- gatherer studies, social anthropology, bio-

logical anthropology, primatology, and neurobiol-

ogy have made clear for the ﬁrst time the nature

of the phenomena studied by social scientists and

the connections of those phenomena to the princi-

ples and ﬁndings in the rest of science. This allows

anew model to be constructed-the Integrated Causal

Model-to replace the Standard Social Science Model.

4. Brieﬂy, the ICM connects the social sciences to

the rest of science by recognizing that:

a. the human mind consists of a set of evolved

information-processing mechanisms instantiated in

the human nervous system;

b. these mechanisms, and the developmental pro-

grams that produce them, are adaptations, produced

by natural selection over evolutionary time in ances-

tral environments;

c. many of these mechanisms are functionally spe-

cialized to produce behavior that solves particular

adaptive problems, such as mate selection, language

acquisition, family relations, and cooperation;

d. to be functionally specialized, many of these

mechanisms must be richly structured in a content-

speciﬁc way;

e. content-speciﬁc information-processing mecha-

nisms generate some of the particular content of hu-

man culture, including certain behaviors, artifacts,

and linguistically transmitted representations;

f. the cultural content generated by these and

other mechanisms is then present to be adopted or

modiﬁed by psychological mechanisms situated in

other members of the population;

g. this sets up epidemiological and historical

population-level processes; and

h. these processes are located in particular ecolog-

ical, economic, demographic, and intergroup social

contexts or environments.

On this view, culture is the manufactured prod-

uct of evolved psychological mechanisms situated in

individuals living in groups. Culture and human so-

cial behavior is complexly variable, but not because

the human mind is a social product, a blank slate,

or an externally programmed general-purpose com-

puter, lacking a richly deﬁned evolved structure. In-

stead, human culture and social behavior is richly

variable because it is generated by an incredibly in-

tricate, contingent set of functional programs that

use and process information from the world, includ-

ing information that is pro- vided both intentionally

and unintentionally by other human beings.

THE STANDARD SOCIAL SCIENCE MODEL

The Central Logic of the Standard Social Sci-

ence Model

But one would be strangely mistaken about

our thought if, from the foregoing, he drew

the conclusion that sociology, according to

us, must, or even can, make an abstraction

of man and his faculties. It is clear; on the

contrary, that the general characteristics of

human nature participate in the work of

elaboration from which social life results.

But they are not the cause of it, nor do they

give it its special form; they only make it

possible. Collective representations, emo-

tions, and tendencies are caused not by cer-

tain states of the consciousnesses of indi-

viduals but by the conditions in which the

social group, in its totality, is placed. Such

actions can, of course materialize only if

the individual natures are not resistant to

them; but these individual natures are

merely the indeterminate material that

the social factor molds, and transforms.

Their contribution consists exclusively in

very general attitudes. in vague and con-

sequently plastic predispositions which,

by themselves, if other agents did not in-

tervene, could not take on the deﬁnite

and complex forms which characterize

social phenomena.

-DURKHEIM, 1895/1962, pp.105-106, em-

phasis added.

Humans everywhere shows striking patterns of

local within-group similarity in their behavior and

thought, accompanied by profound intergroup dif-

ferences. The standard Social Science Model (SSSM

or Standard Model) draws its enduring persuasive

power by starting with these and a few other facts,

rooted in direct experience and common knowledge.

It then focuses on one salient causal and temporal se-

quence: how individuals change over their develop-

ment from "unformed" infants into complexly com-

petent adult members of their local social group, and

how they do so in response to their local human en-

vironment. The central precepts of the SSSM are

direct and seemingly inescapable conclusions drawn

from these facts (0. E. Brown, 1991), and the same

reasoning appears in author after author, from per-

haps its most famous early expression in Durkheim

(1895/1962), to its fully conventional modern ad-

herents (with updated conceptual ornamentation)

such as Geertz (1973).

The considerations that motivate the Standard So-

cial Science Model are as follows:

Step 1. The existence of rapid historical change

and the multitude of spontaneous, human "cross-

fostering experiments" effectively disposes of the

racialist notion that human intergroup behavioral

differences of any signiﬁcance are attributable to

genetic differences between groups. Infants every-

where are born the same and have the same devel-

opmental potential, evolved psychology, or biolog-

ical endowment-a principle traditionally known as

the psychic unity of humankind. The subsequent

growth of knowledge over this century in genetics

and human development has given strong empiri-

cal support to the conclusion that infants from all

groups have essentially the same basic human design

and potential. Human genetic variation, which is

now directly detectable with modern electrophoretic

techniques, is overwhelmingly sequestered into func-

tionally superﬁcial biochemical differences, leaving

our complex functional design universal and species-

typical (Tooby & Cosmides, 1990a). Also, the bulk of

the variation that does exist is overwhelmingly inter-

individual and within-population, and not between

"races" or populations. By the nature of its known

distribution, then, genetic variation cannot explain

why many behaviors are shared within groups, but

not between groups. That is, genetic variation does

not explain why human groups dramatically differ

from each other in thought and behavior. (Signiﬁ-

cantly, this is the only feature of the SSSM that is

correct as it stands and that is incorporated unmod-

iﬁed into the Integrated Causal Model. Why it turns

out to be true, however, depends on the existence

of complex evolved psychological and physiological

adaptations-something explicitly or implicitly denied

by adherents of the SSSM.)

Step 2. Although infants are everywhere the same,

adults everywhere differ profoundly in their behav-

ioral and mental organization.

These ﬁrst two steps, just by themselves, have

led to the following widely accepted deduction: Be-

cause, it is reasoned, a "constant" (the human bio-

logical endowment observable in infants) cannot ex-

plain a "variable" (intergroup differences in complex

adult mental or social organization) the SSSM con-

cludes that "human nature" (the evolved structure of

the human mind) cannot be the cause of the mental

organization of adult humans, their social systems,

their culture, historical change, and so on.

Step 3. Even more transparently, these complexly

organized adult behaviors are absent from infants.

Infants do not emerge speaking, and they appear to

lack virtually every recognizable adult competency.

Whatever "innate" equipment infants are born with

has traditionally been interpreted as being highly

rudimentary, such as an unorganized set of crude

urges or drives, plus the ability to learn-certainly

nothing resembling adult mental organization. Be-

cause adult mental organization (patterned behav-

ior, knowledge, socially constructed realities, and so

on) is clearly absent from the infant, infants must

"acquire" it from some source outside themselves in

the course of development.

Step 4. That source is obvious: This mental orga-

nization is manifestly present in the social world in

the form of the behavior and the public representa-

tions of other members of the local group. Thus, the

stuff of mental organization is categorizable accord-

ing to its source: (I) the "innate" (or inborn or genet-

ically determined, etc.), which is supplied "biologi-

cally" and is what you see in the infant, and (2) the

social (or cultural or learned or acquired or environ-

mental), which contains everything complexly orga-

nized and which is supplied by the social environ-

ment (with a few exceptions supplied by the physi-

cal environment and nonsocial learning). "Cultural

phenomena ...are in no respect hereditary but are

characteristically and without exception acquired"

(Murdock, 1932, p. 200). This line of reasoning

is usually sup- ported by another traditional argu-

ment, the deprivation thought experiment: "Undi-

rected by culture patterns-organized systems of sig-

niﬁcant symbols-man’s behavior would be virtually

ungovernable, a mere chaos of pointless acts and ex-

ploding emotions, his experience virtually shapeless"

(Geertz, 1973, p. 46). Humans raised without a so-

cial or cultural environment would be "mental basket

cases" with "few useful instincts, fewer recognizable

sentiments, and no intellect" (Geertz, 1973, p. 49).

Because, it is reasoned, an effect disappears when its

cause is withdrawn, this thought experiment is be-

lieved to establish that the social world is the cause

of the mental organization of adults.

Step 5. The causal arrow in this process has a

clear directionality, which is directly observable in

the individual’s development. The cultural and so-

cial elements that mold the individual precede the

individual and are external to the individual. The

mind did not create them; they created the mind.

They are "given," and the individual "ﬁnds them al-

ready current in the community when he is born"

(Geertz, 1973, p. 45). Thus, the individual is the

creation of the social world and, it appears to follow,

the social world cannot be the creation of "the indi-

vidual. " If you are reading this chapter, you learned

English and did not create it. Nor did you choose

to learn English (assuming you are a native speaker)

any more than any effect chooses its cause; this ac-

tion of the social world on the individual is com-

pulsory and automatic-"coercive," to use Durkheim’s

phrase. Adult mental organization is socially deter-

mined. Moreover, by looking at social processes in

the vast modern societies and nation-states, it is ob-

vious that the "power asymmetry" between "the indi-

vidual" and the social world is huge in the determi-

nation of outcomes and that the reciprocal impact of

the individual on the social world is negligible. The

causal ﬂow is overwhelmingly or entirely in one di-

rection. The individual is the acted upon (the effect

or the outcome) and the sociocultural world is the

actor (the cause or the prior state that determines

the subsequent state).

Step 6. Accordingly, what complexly organizes

and richly shapes the substance of human life-what

is interesting and distinctive and, therefore, worth

studying-is the variable pool of stuff that is usually

referred to as "culture." Sometimes called "extra- so-

matic" or "extragenetic" (e.g., Geertz, 1973) to em-

phasize its nonbiological origins and nature, this

stuff is variously described as behavior, traditions,

knowledge, signiﬁcant symbols, social facts, control

programs, semiotic systems, information, social or-

ganization, social relations, economic relations, in-

tentional worlds, or socially constructed realities.

However different these characterizations may ap-

pear to be in some respects, those who espouse them

are united in afﬁrming that this substance-what- ever

its character-is (in Durkheim’s phrase) "external to

the individual." Even so psychological a phenomenon

as thinking becomes external: "Human thought is

basically both social and public-?its natural habitat

is the house yard, the marketplace, and the town

square. Thinking consists not of ’happenings in the

head’ (though happenings there and elsewhere are

necessary for it to occur) but of a trafﬁc in what

have been called, by G.H. Mead and others, sig-

niﬁcant symbols-words for the most part " (Geertz,

1973, p. 45). "The individual" contributes only the

infant’s impoverished drives, unformed tendencies,

and capacity to be socialized. These ﬁrst six steps

constitute the SSSM’s account of the causal process

whereby what is assumed to be an initially form-

less infant is transformed into a fully human (i.e.,

fully cultural) being. The next important element

in the SSSM is its approach to answering the ques-

tion, "If culture creates the individual, what then

creates culture?" Before describing the SSSM’s an-

swer to this question, however, we need to make

an important aspect of the question explicit: Human

life is complexly and richly ordered. Human life is

not (solely) noise, chaos, or random effect (contra

Macbeth). Although the substance of human life,

like human speech, is various and contingent, it is

still, like human speech, intricately patterned. Many

attempt to capture this perception with the phrase

that human cultures (e.g., human symbol systems)

are "meaningful." Human conduct does not resemble

white noise. In away that is analogous to William Pa-

ley’s argument from design in his Natural Theology,

one must ask: If there is complex and meaningful

organization in human sociocultural life, what is the

creator or artiﬁcer of it? Entropy, perturbation, er-

ror, noise, interaction with other systems, and so on,

are always operating to inﬂuence culture (and ev-

erything else), so clearly not everything in culture is

orderly. Equally, if these processes were all that were

operating, complex order would never appear and

would quickly degrade even if it did. Just as ﬁnding

a watch on the heath, already complexly organized,

requires that one posit a watchmaker (Paley, 1828),

ﬁnding out that human life is complexly ordered ne-

cessitates the search for the artiﬁcer or source of this

order (see Dawkins, 1986, for an exceptionally lucid

general analysis of the problem of explaining com-

plex order, its importance as a question, and the ex-

tremely narrow envelope of coherent answers). So,

the question is not so much, what are the forces that

act on and inﬂuence human culture and human af-

fairs? but rather, What is the generator of complex

and signiﬁcant organization in human affairs?

Step 7. The advocates of the Standard Social Sci-

ence Model are united on what the artiﬁcer is not

and where it is not: It is not in "the individual"-in

human nature or evolved psychology-which, they as-

sume, consists of nothing more than what the infant

comes equipped with, bawling and mewling, in its

apparently unimpressive initial performances. Be-

cause the directional ﬂow of the organization is from

the outer world inward into "the individual," the di-

rection toward which one looks for the source of

the organization is likewise clear: outward into the

social world. As Durkheim says, "[w]hen the indi-

vidual has been eliminated, society alone remains"

(1895/1962, p. 102).

Step 8. The SSSM maintains that the generator

of complex and meaningful organization in human

life is some set of emergent processes whose de-

terminants are realized at the group level. The so-

ciocultural level is a distinct, autonomous, and self-

caused realm: "Culture is a thing sui generis which

can be explained only in terms of itself. ...Omnis

cultura ex cultura" (Lowie, 1917/1966, p. 25-26).

For Alfred Kroeber, "the only antecedents of histori-

cal phenomena are historical phenomena" (Kroeber,

1917). Durkheim was equally emphatic: "The de-

termining cause of a social fact should be sought

among the social facts preceding it and not among

the states of individual consciousness"; that is, phe-

nomena at the sociocultural level are mostly or en-

tirely caused by other phenomena at the sociocul-

tural level (Durkheim, 1895/1962, p. 110). It must

be emphasized that this claim is not merely the ob-

vious point that social phenomena (such as tulip

bulb mania, the contagious trajectory of deconstruc-

tionist fashions, or the principles of supply and de-

mand) cannot be understood simply by pointing in-

side the head of a single individual. It is, instead,

a claim about the generator of the rich organization

everywhere apparent in human life. What is gen-

erated even includes individual adult psychological

phenomena, which are themselves simply additional

social constructions. For Durkheim (and for most an-

thropologists today), even emotions such as "sexual

jealousy" and "paternal love" are the products of the

social order and have to be explained "by the con-

ditions in which the social group, in its totality, is

placed." As Geertz argues, "Our ideas, our values, our

acts, even our emotions, are, like our nervous sys-

tem itself, cultural products-products manufactured,

indeed, out of tendencies, capacities, and disposi-

tions with which we were born, but manufactured

nonetheless" ( 1973, p. 50). Similarly, Shweder de-

scribes "cultural psychology" as "the study of the way

cultural traditions and social practices regulate, ex-

press, transform, and permute the human psyche,

resulting less in psychic unity for humankind than

in ethnic divergences in mind, self and emotion "

(Shweder, 1990, p. 1).

Step 9. Correspondingly, the SSSM denies that

"human nature"-the evolved architecture of the hu-

man mind can play any notable role as a generator

of signiﬁcant organization in human life (although

it is acknowledged to be a necessary condition for

it). In so doing, it removes from the concept of hu-

man nature all substantive, content, and relegates

the architecture of the human mind to the delim-

ited role of embodying "the capacity for culture." Hu-

man nature is "merely the indeterminate material

that the social factor molds and transforms. [This]

contribution consists exclusively in very general at-

titudes, in vague and consequently plastic predis-

positions which, by themselves, if other agents did

not intervene, could not take on the deﬁnite and

complex forms which characterize social phenom-

ena" (Durkheim, 1895/1962, p. 106). As Hatch

comments, the "view that the Boasians had struggled

to roster within the social sciences since almost the

turn of the century" was that the human mind is "al-

most inﬁnitely malleable" ( 1973, p. 236). Socializa-

tion is the process of externally supplied "conceptual

structures molding formless talents" (Geertz, 1973,

p. 50).

Social scientists who paid any attention to neu-

roscience, ethology, and cognitive psychology were

increasingly, if uneasily, aware of the evidence that

the nervous sys- tem was complex and not well char-

acterized by the image of the "blank slate." None-

the less, aside from paying some lip service to the

notion that tabula rasa empiricism was untenable,

this changed nothing important in the SSSM. The

blank slate was traded in for blank cognitive proce-

dures.2 The mind could be seen as complex, but its

procedures were still assumed to be content-free. As

long as environmental input could enter and mod-

ify the system, as it clearly could, environmental in-

put was presumed to orchestrate the system, giv-

ing it its functional organization. It doesn’t mat-

ter if the clay of the human mind has some initial

shape (tendencies, dispositions), so long as it is soft

enough to be pounded by the external forces into

any new shape required. Thus, for Geertz, who is

attracted to the language if not the actual substance

of cognitive science, the mind is not a slate, blank

or otherwise (he dismisses this as a straw man po-

sition "which no one of any seriousness holds" or

perhaps ever held [Geertz, 1984, p. 268]), but

it is instead the tabula rasa’s fully modern equiva-

lent, a general-purpose computer. Such a computer

doesn’t come pre-equipped with its own programs,

but instead-and this is the essential point-it obtains

the programs that tell it what to do from the out-

side, from "culture." Thus, the human mind is a com-

puter that is "desperately dependent upon such ex-

tragenetic, outside-the-skin control mechanisms" or

"programs" "for the governing of behavior" (Geertz,

1973, p. 44).

This eliminates the concept of human nature or

its alternative expression, the evolved psychological

architecture, as useful or informative concepts. As

Geertz puts it, "[t]he rise of the scientiﬁc concept of

culture amounted to. ..the overthrow of the view of

human nature dominant in the Enlightenment. ..",

that is, that "[man] was wholly of apiece with nature

and shared in the general uniformity of composition

which natural science. ..had discovered there" with

"a human nature as regularly organized, as thor-

oughly invariant, and as marvelously simple as New-

ton’s universe" (Geertz, 1973, p. 34). Instead, the

view entailed in the modern "scientiﬁc concept of

culture" is that "humanity is as various in its essence

as in its expression" (Geertz, 1973, p. 37). Geertz

does not mean, of course, that infants vary due to

genetic differences, but that all signiﬁcant aspects of

adult mental organization are supplied culturally. As

deeply as one can go into the mind, people here are

different from people there, leading to "the decline

of the uniformitarian view of man" (Geertz, 1973, p.

35).

The conclusion that human nature is an empty

vessel, waiting to be ﬁlled by social processes, re-

moved it as a legitimate and worthwhile object of

study. Why study paper when what is interesting is

the writing on it and, perhaps even more important,

the author (the perennially elusive generative social

processes)? Since there could be no content, per se,

to the concept of human nature, anything claimed

to be present in human nature was merely an eth-

nocentric projection of the scholar making the claim.

Thus, attempts to explore and characterize human

nature became suspect. Such efforts were (and are)

viewed as simply crude attempts to serve ideological

ends, to manufacture propaganda, or to deﬁne one

way of being as better and more natural than others.

Step 10. In the SSSM, the role of psychology is

clear. Psychology is the discipline that studies the

process of socialization and the set of mechanisms

that comprise what anthropologists call "the capac-

ity for culture" (Spuhler, 1959). Thus, the central

concept in psychology is learning. The prerequi-

site that a psychological theory must meet to par-

ticipate in the SSSM is that any evolved compo-

nent, process, or mechanism must be equipotential,

content-free, content-independent, general-purpose,

domain- general, and so on (the technical terms vary

with movement and era). In short, these mecha-

nisms must be constructed in such away that they

can absorb any kind of cultural message or environ-

mental input equally well. Moreover, their struc-

tures must themselves impose no particular substan-

tive content on culture. As Rindos ( 1986, p.315)

puts it, "the speciﬁcs that we learn are in no sense

predetermined by our genes." Learning is thus the

window through which the culturally manufactured

pre-existing complex organization outside of the in-

dividual manages to climb inside the individual. Al-

though this approach deprives psychological mech-

anisms of any possibility of being the generators of

signiﬁcant organization in human affairs, psycholo-

gists get something very appealing in exchange. Psy-

chology is the social science that can hope for gen-

era1laws to rival those of the natural sciences: gen-

eral laws of learning, or (more recently) of cogni-

tive functioning. The relationship of psychology to

biology is also laid out in advance by the SSSM: In

human evolution, natural selection removed "genet-

ically determined" systems of behavior and replaced

them with general-purpose learning mechanisms or

content-independent cognitive processes. Suppos-

edly, these more general systems were favored by

evolution because they did not constrain human be-

havior to be maladaptively inﬂexible (e.g., Geertz,

1973; Harris, 1979; Montagu, 1964 ). Neurobiology

is the account of how these general mechanisms are

instantiated in our nervous system.

Consequently, the concepts of learning, social-

ization, genera1-purpose (or content- independent)

cognitive mechanisms, and environmentalism have

(under various names and permutations) dominated

scientiﬁc psychology for at least the last 60 years.

Skinnerian behaviorism, of course, was one of the

most institutionally successful manifestations of the

SSSM’s program for psychology, but its antimen-

talism and doctrinaire scientism made it unconge-

nial to those who wanted an account of their in-

ternal experience. More importantly, its emphasis

on individual histories of reinforcement limited the

avenues through which culture could have its ef-

fect. It proved an easy target when cognitive science

provided precise ways of characterizing and inves-

tigating the mental as a system that processes in-

formation, a characterization that seemed to offer

easier avenues for cultural transmission than labo-

riously organized schedules of reinforcement. Al-

though cognitive psychologists threw out behavior-

ism’s cumbersome antimentalism, they uncritically

adopted behaviorism’s equipotentiality assumption.

In mainstream cognitive psychology, it is assumed

that the machine is free of content- specialized

processes and that it consists primarily of general-

purpose mechanisms. Psychologists justify this as-

sumption by an appeal to parsimony: It is "unsci-

entiﬁc" to multiply hypothesized mechanisms in the

head. The goal, as in physics, is for as few principles

as possible to account for as much as possible. Con-

sequently, viewing the mind as a collection of spe-

cialized mechanisms that perform, speciﬁc tasks ap-

pears to be a messy approach, one not worth pursu-

ing. Anthropologists and sociologists easily accom-

modated themselves to these theoretical changes in

psychology: Humans went from being viewed as

relatively simple equipotential learning systems to

very much more complex equipotential information-

processing systems, general-purpose computers, or

symbol manipulators (see, e.g., Sahlins, 1976a,

1976b).

Within psychology there are, of course, impor-

tant research communities that fall outside of the

SSSM and that have remained more strongly con-

nected to the rest of science, such as physiological

psychology, perception, psychophysics, (physiologi-

cal) motivation, psycholinguistics, much of compar-

ative psychology, and a few other areas. Moreover,

to explain how organisms remain alive and repro-

duce (and to make some minimal attempt to account

for the focused substance of human life), psycholo-

gists have found it necessary to posit a few content-

oriented mechanisms: hunger, thirst, sexual moti-

vation, and so on. Nevertheless, the tendency has

been to keep these elements restricted to as small a

class as possible and to view them as external to the

important central learning or cognitive processes.

They are incorporated as, for exam- pie, reinforces

operating by drive reduction. Cognitive psycholo-

gists have, for the most part, labored to keep any

such content-inﬂuenced elements extrinsic to the pri-

mary cognitive machinery. Indeed, they have usually

avoided addressing how functional action-such as

mate choice, food choice, or effort calculation-takes

place at all. The principles of concept formation, of

reasoning, of remembering, and so forth, have tra-

ditionally been viewed as uninfected prior to experi-

ence with any content, their procedures lacking fea-

tures designed for dealing with particular types of

content. Modular or domain-speciﬁc cognitive psy-

chologists, in dissenting from this view, are abandon-

ing the assumptions of the Standard Social Science

Model. Of course, readers should recognize that by

so brieﬂy sketching large expanses of intellectual his-

tory and by so minimally characterizing entire re-

search communities, we are doing violence to the

speciﬁc reality of, and genuine differences among

hundreds of carefully developed intellectual systems.

We have had to leave out the qualiﬁcations and com-

plexities by which positions are softened, pluralisms

espoused, critical distinctions lost, and, for that mat-

ter, lip service paid. This is inevitable in attempting

so synoptic a view. In what is surely a graver defect,

we have had to omit discussion of the many impor-

tant dissident subcommunities in sociology, anthro-

pology, economics, and other disciplines, which have

sloughed off or never adopted the Standard Social

Science Model. In any case, we simply hope that this

sketch captures a few things that are true and impor-

tant, to compensate for the unavoidable simplifying

distortions and omissions. Most obviously, there are

no pure types in the world, and scholars are quoted

not to characterize the full richness of their individ-

ual views, which usually undergo considerable evo-

lution over their intellectual development anyway,

but rather to illustrate instances of a larger intellec-

tual system. It is the larger intellectual system we

are criticizing, and not the multitude of worthwhile

research efforts that have gone on inside its struc-

ture. We think the roof of the Standard Social Sci-

ence Model has collapsed, so to speak, because the

overall architectural plan is unsound, not because

the bricks and other building materials are defec-

tive. The detailed research efforts of hundreds of

scientists have produced critically important knowl-

edge that has transformed our understanding of the

world. In this criticism, we are looking for an archi-

tectural design for the social sciences that is worthy

of the intelligence and labor of those whose research

goes on within their compass.

The Standard Social Science Model’s Treatment

of Culture

This logic has critically shaped how nearly every

issue has been approached and debated in the social

sciences. What we are concerned with here, how-

ever, is the impact of the Standard Social Science

Model on the development of modern conceptions

of culture, its causal role in human life, and its rela-

tionship to psychology. Brieﬂy, standard views of cul-

ture are organized according to the following propo-

sitions (see also D. E. Brown, 1991, p. 146; Tooby &

Cosmides, 1989a):

I. Particular human groups are properly character-

ized typologically as having "a" culture, which con-

sists of widely distributed, or nearly group-universal

behavioral practices, beliefs, ideational systems, sys-

tems of signiﬁcant symbols, or informational sub-

stance of some kind. Cultures are more or less

bounded entities, although cultural elements may

diffuse across boundaries. 2. These common el-

ements are maintained and transmitted "by the

group," an entity that has cross-generational conti-

nuity.

3. The existence of separate streams of this infor-

mational substance, culture, transmitted from gen-

eration to generation, is the explanation for human

within-group similarities and between-group differ-

ences. In fact, all between- group differences in

thought and behavior are referred to as cultural

differences and all within-group similarities are re-

garded as the expressions of a particular culture.

Since these similarities are considered to be "cul-

tural," they are, either implicitly or explicitly, con-

sidered to be the consequence of informational sub-

stance inherited jointly from the preceding genera-

tion by all who display the similarity.

4. Unless other factors intervene, the culture (like

the gene pool) is accurately replicated from genera-

tion to generation.

5. This process is maintained through learning,

a well-understood and unitary process that acts to

make the child like the adult of her culture.

6. This process of learning can be seen, from the

point of view of the group, as a group-organized pro-

cess called socialization, imposed by the group on

the child.

7. The individual is the more or less passive recip-

ient of her culture and is the product of that culture.

8. What is organized and contentful in the minds

of individuals comes from culture and is socially

constructed. The evolved mechanisms of the hu-

man mind are themselves content-independent and

content-free and, therefore, what- ever content ex-

ists in human minds originally derives from the so-

cial or (sometimes) nonsocial environment.

9. The features of a particular culture are the re-

sult of emergent group-level processes, whose deter-

minants arise at the group level and whose outcome

is not given speciﬁc shape or content by human bi-

ology, human nature, or any inherited psychological

design. These emergent processes, operating at the

sociocultural level, are the ultimate generator of the

signiﬁcant organization, both mental and social, that

is found in human affairs.

10. In discussing culture, one can safely neglect a

consideration of psychology as anything other than

the nondescript "black box" of learning, which pro-

vides the capacity for culture. Learning is a sufﬁ-

ciently speciﬁed and powerful explanation for how

any behavior acquires its distinct structure and must

be the explanation for any aspect of organized hu-

man life that varies from individual to individual and

from group to group.

11. Evolved, "biological," or "innate" aspects of hu-

man behavior or psychological organization are neg-

ligible, having been superseded by the capacity for

culture. The evolution of the capacity for culture

has led to a ﬂexibility in human behavior that be-

lies any signiﬁcant "instinctual" or innate component

(e.g., Geertz, 1973; Montagu, 1968, p. II; Sahlins,

1976a &b), which, if it existed, would have to reveal

itself as robot like rigid behavioral universals. To the

extent that there may be any complex biological tex-

tures to individual psychology, these are nevertheless

organized and given form and direction by culture

and, hence, do not impart any substantial character

or content to culture.

On the Reasonableness of the Standard Social

Science Model

There are, of course, many important elements of

truth in the tenets of the SSSM, both in its core logic

and in its treatment of culture. The SSSM would

not have become as decisively inﬂuential if it did not

have a strong surface validity, anchored in important

realities. For example: It is true that infants are ev-

erywhere the same. Genetic differences are superﬁ-

cial. There is within-group similarity of behavior and

there are between-group differences, and these per-

sist across generations, but also change over histor-

ical time. Highly organized socially communicated

information exists outside of any particular individ-

ual at anyone time (in the cognitive mechanisms of

other individuals), and over time this information

can be internalized by the speciﬁc individual in ques-

tion. And so on.

Nevertheless, the Standard Social Science Model

contains a series of major defects that act to make it,

as a framework for the social sciences, deeply mis-

leading. As a result, it has had the effect of stunt-

ing the social sciences, making them seem falsely

autonomous from the rest of science (i.e., from the

"natural sciences") and precluding work on answer-

ing questions that need to be answered if the so-

cial sciences are to make meaningful progress as

sciences. After a century, it is time to reconsider

this model in the light of the new knowledge and

new understanding that has been achieved in evolu-

tionary biology, development, and cognitive science

since it was ﬁrst formulated.. These defects cluster

into several major categories, but we will limit our

discussion to the following three:

I. The central logic of the SSSM rests on naive

and erroneous concepts drawn from outmoded the-

ories of development. For example, the fact that

some aspect of adult mental organization is absent

at birth has no bearing on whether it is part of our

evolved architecture. Just as teeth’ or breasts are

absent at birth, and yet appear through maturation,

evolved psychological mechanisms or modules (com-

plex structures that are functionally organized for

processing information) could develop at any point

in the life cycle. For this reason, the many features

of adult mental organization absent at birth need not

be attributed to exposure to transmitted culture, but

may come about through a large number of causal

avenues not considered in traditional analyses.

2. More generally, the SSSM rests on a faulty anal-

ysis of nature-nurture issues, stemming from a fail-

ure to appreciate the role that the evolutionary pro-

cess plays in organizing the relationship between our

species-universal genetic endowment, our evolved

developmental processes, and the recurring features

of developmental environments. To pick one misun-

derstanding out of a multitude, the idea that the phe-

notype can be partitioned dichotomously into genet-

ically determined and environmentally determined

traits is deeply ill-formed, as is the notion that traits

can be arrayed along a spectrum according to the

degree that they are genetically versus environmen-

tally caused. The critique of the SSSM that has been

emerging from the cognitive and evolutionary com-

munities is not that traditional accounts have under-

estimated the importance of biological factors rela-

tive to environmental factors in human life. Instead,

the target is the whole framework that assumes that

"biological factors" and "environmental factors" re-

fer to mutually exclusive sets of causes that exist in

some kind of explanatory zero-sum relationship, so

that the more one explains "bio- logically" the less

there is to explain "socially" or "environmentally. "

On the contrary, as we will discuss, environmental-

ist claims necessarily require the existence of a rich,

evolved cognitive architecture.

3. The Standard Social Science Model requires an

impossible psychology. Results out of cognitive psy-

chology, evolutionary biology, artiﬁcial intelligence,

developmental psychology, linguistics, and philoso-

phy converge on the same conclusion: A psychologi-

cal architecture that consisted of nothing but equipo-

tential, general-purpose, content-independent, or

content-free mechanisms could not successfully per-

form the tasks the human mind is known to perform

or solve the adaptive problems humans evolved to

solve-from seeing, to learning a language, to rec-

ognizing an emotional expression, to selecting a

mate, to the many disparate activities aggregated un-

der the term "learning culture" (Cosmides & Tooby,

1987; Tooby & Cosmides, 1989a). It cannot account

for the behavior observed, and it is not a type of de-

sign that could have evolved.

The alternative view is that the human psycholog-

ical architecture contains many evolved mechanisms

that are specialized for solving evolutionarily Iong-

enduring adaptive problems and that these mecha-

nisms content-specialized representational formats,

procedures, cues, and so on. These richly content-

sensitive evolved mechanisms tend to impose cer-

tain types of content and conceptual organization

on human mental life and, hence, strongly shape

the nature of human social life and what is cul-

turally transmitted across generations. Indeed, a

post-Standard Model psychology is rapidly coalesc-

ing, giving a rapidly expanding empirical foundation

to this new framework. In fact, historically, most of

the data already gathered by psychologists supports

such a view. It required a strongly canalized inter-

pretative apparatus to reconcile the raw data of psy-

chology with the central theoretical tenets of SSSM

psychology.

Before examining in detail what is wrong with the

SSSM and why the recognition of these defects leads

to the formulation of anew model with greater ex-

planatory power, it is necessary ﬁrst to alleviate the

fears of what would happen if one "falls off the edge"

of the intellectual world created by the SSSM. These

fears have dominated how alternative approaches to

the SSSM have been treated in the past and, un-

less addressed, will prevent alternatives from being

fairly evaluated now. Moreover, the Standard Model

has become so well-internalized and has so strongly

shaped how we now experience and interpret social

science phenomena that it will be difﬁcult to free

ourselves of the preconceptions that the Standard

Model imposes until its Procrustean operations on

psychology and anthropology are examined.

THE WORLD BUILT BY THE STANDARD SOCIAL

SCIENCE MODEL

The Moral Authority of the Standard Social Sci-

ence Model

The Case Against Nativism

The overwhelming success of the Standard So-

cial Science Model is attributable to many factors

of which, arguably, the most signiﬁcant has been its

widespread moral appeal. Over the course of the

century, its strong stand against explaining differ-

ences between races, classes, sexes, or individuals by

hypothesizing underlying biological differences has

been an important element in combating a multitude

of searing horrors and oppressions, from the exter-

mination of ethnic groups and the forced steriliza-

tion of the poor to restrictive immigration laws and

legally institutionalized sex and race discrimination.

The depth of these tragedies and the importance of

the issues involved have imbued the SSSM and its

central precept, "environmentalism," with an impos-

ing moral stature. Consequently, the positions of in-

dividual scholars with respect to the SSSM have been

taken to imply allegiances with respect to the larger

social and moral conﬂicts around the world. Thus, to

support the SSSM was to oppose racism and sexism

and to challenge the SSSM was, intentionally or not,

to lend support to racism, sexism, and, more gener-

ally (an SSSM way of deﬁning the problem), "biolog-

ical determinism. " If biological ideas could be used

to further such ends, then ideas that minimized the

relevance of biology to human affairs, such as the

tenets of the SSSM, could only be to the good.

In this process, all approaches explicitly involv-

ing nativist elements of whatever sort became sus-

pect. In consequence, fundamentally divergent-even

opposing- programs and claims have become endur-

ingly conﬂated in the minds of 20th-century social

scientists. Most signiﬁcant was the failure to dis-

tinguish adaptationist evolutionary biology from be-

havior genetics. Although the adaptationist inquiry

into our universal, inherited, species-typical design

is quite distinct from the behavior genetics ques-

tion about which differences between individuals or

sets of individuals are caused by differences in their

genes, the panspeciﬁc nativism typical of adaptation-

ist evolutionary biology and the idiotypic nativism of

behavior genetics became confused with each other

(Tooby and Cosmides, 1990a). Obviously, claims

about a complexly organized, universal human na-

ture, by their very character, cannot participate in

racist explanations. Indeed, they contradict the cen-

tral premises of racialist approaches. Yet, despite this

fact, adaptationist approaches and behavior genetics

remain inextricably intertwined in the minds of the

majority of social scientists.

The second strong moral appeal the Standard So-

cial Science Model derives from its emphasis on hu-

man malleability and the hope it, therefore, gave for

social melioration or social revolution. The claim of

John B. Watson, the founder of behaviorism, exem-

pliﬁes this optimism about the power of scientiﬁcally

directed socialization (as well as the usual implicit

conﬂation of idiotypic and panspeciﬁc nativism):

Give me a dozen healthy infants, well-formed, and

my own speciﬁed world to bring them up in and I’ll

guarantee to take anyone at random and train him to

become any type of specialist I might select-doctor,

lawyer, artist, merchant-chief, and yes, even beggar-

man and thief, regardless of his talents, penchants,

tendencies, abilities, vocations, and race of his an-

cestors (Watson, 1925, p. 82).

As D. E. Brown (1991, p. 61) comments, "In hind-

sight it is clear that this famous statement about the

inﬂuence of the environment on individual differ-

ences is entirely compatible with the most extreme

of the faculty or modular views of the human mind-

in which it comprises numerous innate and highly

speciﬁc mechanisms." But this thought experiment

was interpreted by the social science mainstream,

Watson included, as demonstrating that "people are

the products of their societies or cultures. " There-

fore, "change society or culture and you change peo-

ple. ...Intelligent, scientiﬁc socialization can make

us whatever we want to be" (0. E. Brown, 1991,

p. 61 ). More- over, "[t]he equation of an arch en-

vironmentalism (including cultural relativism) with

optimism about the practical application of social sci-

ence to the problems of society remains a force to

the present" (0. E. Brown, 1991, p. 62). More crit-

ically, the belief that the mind is "almost inﬁnitely

malleable" (or, in more modem terms, is a general-

purpose computer) means that humans are not con-

demned to the status quo, and need not inevitably

ﬁght wars, or have social classes, or manifest sex-

differentiated roles, or live in families, and so on.

If the "happy" ability of the mind to "quite read-

ily take any shape that is presented" (Benedict,

1934/1959, p. 278) is the ameliorator’s ideal be-

cause it is believed to be logically necessary to allow

social change, then dissent from the SSSM tends to

be framed as claims about "constraints" or limits on

this malleability. This, in turn, is taken to imply a

possible intractability to social problems-the stronger

the biological forces are, the more we may be con-

strained to suffer from certain inevitable expressions

of human darkness. Thus, the debate on the role of

biology in human life has been consistently framed

as being between optimistic environmentalists who

plan for human betterment and sorrowful, but real-

istic nativists who lament the unwelcome inevitabil-

ity of such things as aggression ( e.g., Ardrey, 1966;

Lorenz, 1966), or who (possibly even gleefully) de-

fend the status quo as inevitable and natural (e.g.,

Gol 1973, on patriarchy). These nativists are, in

turn, "debunked" by the tireless opponents of "bio-

logical determinism" (e.g., Chorover, 1979; Lewon-

tin, Rose, & Kamin, 1984; Montagu, 1968, 1978),

who place each new biological intrusion onto social

science territory in the context of the bitter lessons

of the century .(Environmentalist holocausts are, of

course, edited out of this chronology. )

This morality play, seemingly bound forever to the

wheel of intellectual life, has been through innu-

merable incarnations, playing itself out in different

arenas in different times (rationalism versus empiri-

cism, heredity versus environment, instinct versus

learning, nature versus nurture, human universals

versus cultural relativism, human nature versus hu-

man culture, innate behavior versus acquired behav-

ior, Chomsky versus Piaget, biological determinism

versus social determinism, essentialism versus social

construction, modularity versus domain-generality,

and so on). It is perennial because it is inherent in

how the issues have been deﬁned in the SSSM itself,

which even governs how the dissidents frame the na-

ture of their dissent. Accordingly, the language of

constraint and limitation is usually adopted by bio-

logically oriented behavioral scientists themselves in

describing the signiﬁcance of their own work. Thus

we even have titles such as The Tangled Wing: Bi-

ological Constraints on the Human Spirit (Konner,

1982), Biological Boundaries of Learning (Seligman

and Hager, 1972), "Constraints on Learning" (Shet-

tleworth, 1972), and "Structural Constraints on Cog-

nitive Development" (Gelman, 1990a). Biologically

oriented social and behavioral scientists often see

themselves as deﬁning limits on the possible. Envi-

ronmentalists see themselves as expanding the bor-

ders of the possible. As we will see, this framing is

profoundly misleading.

Wrong Diagnosis, Wrong Cure

Driven by these fears to an attitude that Daly and

Wilson ( 1988) have termed "bio- phobia," the so-

cial science community lays out implicit and some-

times explicit ground rules in its epistemological hi-

erarchy: The tough-minded and moral stance is to

be skeptical of panspeciﬁc "nativist" claims; that is,

of accounts that refer in any way to the participation

of evolved psychological mechanisms together with

environ- mental variables in producing outcomes, no

matter how logically inescapable or empirically well-

supported they may be. They are thought to be

explanations of last resort and, because the tough-

minded and skeptical can generate particularistic al-

ternative accounts for any result at will, this last re-

sort is rarely ever actually arrived at. For the same

reason, it is deemed to be the moral stance to be

correspondingly credulous of "environmentalist" ac-

counts, no matter how vague, absurd, incoherent, or

empirically contradicted they may be. These proto-

cols have become second nature (so

Ito speak) to nearly everyone in the social science

community. This hierarchy is driven by the fear of

falling off the edge of the Standard Social Science

Model, into unknown

Iregions where monsters such as "biological" or

"genetic determinism" live.

What, in fact, is an environmentalist account?

There are two brands of environmentalism: coher-

ent environmentalism and incoherent environmen-

talism, which correspond approximately to environ-

mentalism as defended and environmentalism as

practiced. As Daly and Wilson ( 1988, p. 8) com-

ment, "[a]ll social theorists, including the staunchest

antinativists, seek to describe human nature at some

cross culturally general level of abstraction " and

would be "distressed should their theories. .. prove

applicable to Americans but not to Papuans." Both

Skinner (1957) and Chomsky ( 1975)-opponents in

a paradigmatic case of an environmentalist-nativist

debate-posit the existence of universal evolved psy-

chological mechanisms, or what cognitive psychol-

ogists have called "innate mechanisms." As Symons

(1987) points out, most of what passes for the

nature-nurture debate is not about the need to posit

evolved mechanisms in theories. Everyone capable

of reasoning logically about the problem accepts the

necessity of this. As Symons makes clear, what the

debate often seems to be about is how general or

content-speciﬁc the mechanisms are: Skinner pro-

poses conditioning mechanisms that apply to all sit-

uations, while Chomsky proposes specialized mech-

anisms particularly designed for language. Conse-

quently, coherent

I environmentalists acknowledge that they are

positing the existence of evolved developmental or

psychological mechanisms and are willing to de-

scribe ( 1) the explicit structure of these mecha-

nisms, and (2) what environmental variables they in-

teract with to produce given outcomes. By this stan-

dard, of course, Chomsky is an environ- mentalist, as

was Skinner, as are we, along with most other evo-

lutionary psychologists and evolutionarily informed

behavioral scientists. Equally, all coherent behav-

ioral scientists of whatever orientation must be na-

tivists in this same sense, and no coherent and fully

speciﬁed hypothesis about behavior can avoid mak-

ing nativist claims about the involvement of evolved

structure.

Incoherent environmentalists, on the other hand,

are those who propose theories of how environments

regulate behavior or even psychological phenomena

without describing or even mentioning the evolved

mechanisms their theories would require to be com-

plete or coherent. In practice, communities whose

rules of discourse are governed by incoherent envi-

ronmentalism consider any such trend toward ex-

plicitness to be introducing vague and speculative

variables and-more to the point-to be in bad taste as

well. The simple act of providing a complete model

is to invoke evolved design and, hence, to court be-

ing called a genetic or biological determinist. Given

that all coherent (fully speciﬁed) models of psycho-

logical processes necessarily entail an explication of

how environmental variables relate to the inherited

architecture or developmental machinery, this atti-

tude has the effect of portraying psychologists who

are clear about all causal steps as more soft-minded

and speculative than those who remain vague about

the crucial elements necessary to make their theo-

ries coherent (e.g., Cheng & Holyoak, 1989). (This

criticism is, of course, fully symmetrical: Incoherent

nativists are those scholars who talk about how evo-

lution structures behavior without attempting to de-

scribe the structure of the evolved adaptations that

link evolution, environment, and behavior in adap-

tively patterned ways; for discussion, see Symons,

this volume; Tooby & Cosmides, 1990b.)

The problem with an epistemological hierarchy

that encourages incoherence and discourages coher-

ence (aside from the fully sufﬁcient objection that

it has introduced major distortions into the body of

scientiﬁc knowledge) is that it is completely unnec-

essary, even on its own terms. Not only is the cure

killing the patient-social science- but also the diag-

nosis is wrong and the patient is not menaced by the

suspected mal- ady. In the ﬁrst place, as discussed,

models of a robust, universal human nature by their

very character cannot participate in racist explana-

tions of intergroup differences. This is not just a

deﬁnitional trick of deﬁning human nature as what-

ever is universal. There are strong reasons to believe

that selection usually tends to make complex adap-

tations universal or nearly universal, and so humans

must share a complex, species-typical and species-

speciﬁc architecture of adaptations, however much

variation there might be in minor, superﬁcial, non-

functional traits. As long lived sexual reproducers,

complex adaptations would be destroyed by the ran-

dom processes of sexual recombination every gener-

ation if the genes that underlie our complex adap-

tations varied from individual to individual. Selec-

tion in combination with sexual recombination tends

to enforce uniformity in adaptations, whether phys-

iological or psychological, especially in long-lived

species with an open population structure, such as

humans (Tooby & Cosmides, 1990b). Empirically, of

course, the fact that any given page out of Gray’s

Anatomy describes in precise anatomical detail indi-

vidual humans from around the world demonstrates

the pronounced monomorphism present in complex

human physiological adaptations. Although we can-

not yet directly "see" psychological adaptations (ex-

cept as described neuroanatomically), no less could

be true of them. Human nature is everywhere the

same.

The Malleability of Psychological Architecture

versus the Volatility of Behavioral Outcomes

If the fear that leaving the Standard Social Science

Model will lead to racist doctrines is unfounded,

what of the issue of human malleability? Does a bio-

logically informed approach necessarily imply an in-

tractability of undesired social and behavioral out-

comes and an inevitability of the status quo? Af-

ter all, isn’t the basic thrust of biologically informed

accounts against malleability and in favor of con-

straints and limits on human aspirations?

No. The central premise of an opposition be-

tween the mind as an inﬂexible bio- logical prod-

uct and the mind as a malleable social product is

ill-formed: The notion that inherited psychological

structure constrains is the notion that without it we

would be even more ﬂexible or malleable or envi-

ronmentally responsive than we are. This is not

only false but absurd. Without this evolved struc-

ture, we would have no competences or contingent

environmental responsiveness whatsoever. Evolved

mechanisms do not prevent, constrain, or limit the

system from doing things it otherwise would do in

their absence. The system could not respond to "the

environment" (that is, to selected parts of the en-

vironment in an organized way) without the pres-

ence of mechanisms designed to create that con-

nection. Our evolved cognitive adaptations-our in-

herited psychological mechanisms-are the means by

which things are afﬁrmatively accomplished. It is an

absurd model that proposes that the potentially un-

fettered human mind operates by ﬂailing around and

is only given structure and direction by the "limits"

and "constraints" built in by "biology." Instead, any

time the mind generates any behavior at all, it does

so by virtue of speciﬁc generative programs in the

head, in conjunction with the environmental inputs

with which they are presented. Evolved structure

does not constrain; it creates or enables (Cosmides

& Tooby, 1987).

Given that we are all discussing universal human

design and if, as Symons argues, all coherent behav-

ioral scientists accept the reality of evolved mech-

anisms, then the modern nature-nurture debate is

really about something else: the character of those

evolved mechanisms (Symons, 1987). Does the

mind consist of a few, general-purpose mechanisms,

like operant conditioning, social learning, and trial-

and-error induction, or does it also include a large

number of specialized mechanisms, such as a lan-

guage acquisition device (Chomsky, 1975; Pinker,

1984; Pinker & Bloom, this volume), mate prefer-

ence mechanisms (Buss, 1989, this volume; Ellis,

this volume; Symons, 1979), sexual jealousy mech-

anisms (Daly, Wilson, & Weghorst, 1982; Wilson &

Daly, this volume), mother-infant emotion communi-

cation signals (Fernald, this volume), social contract

algorithms (Cosmides, 1989; Cosmides & Tooby,

1989, this volume; Gigerenzer & Hug, in press), and

so on? This is the point of separation between the

Standard Social Science Model and the Integrated

Causal Model, and it is the main focus of this vol-

ume.

How, then, does the issue of the number and speci-

ﬁcity of evolved mental mechanisms bear on the is-

sue of the inevitability of undesired behavioral out-

comes? As we will discuss and review later, the same

answer applies: General mechanisms turn out to be

very weak and cannot unassisted perform at least

most and perhaps all of the tasks humans routinely

perform and need to perform. Our ability to perform

most of the environmentally engaged, richly contin-

gent activities that we do depends on the guiding

presence of a large number of highly specialized psy-

chological mechanisms (Cosmides & Tooby, 1987;

Rozin, 1976; Symons, 1987; Tooby & Cosmides,

1990b). Far from constraining, specialized mecha-

nisms enable competences and actions that would

not be possible were they absent from the architec-

ture. This rich array of cognitive specializations can

be likened to a computer program with millions of

lines of code and hundreds or thousands of func-

tionally specialized subroutines. It is because of, and

not despite, this speciﬁcity of inherent structure that

the output of computational systems is so sensitively

contingent on environmental inputs. It is just this

sensitive contingency to subtleties of environmental

variation that make a narrow intractability of out-

comes unlikely.

The image of clay, and terms such as "malleability,"

"ﬂexibility," and "plasticity" confuse two separate is-

sues: ( 1) the detailed articulation of human evolved

psychological design (i.e., what is the evolved design

of our developmental programs and of the mech-

anisms they reliably contstruct), and (2) the ﬁxity

or intractability of expressed outcomes (what must

people do, regardless of circumstance). The ﬁrst

question asks what evolved organization exists in

the mind, while the second asks what events will in-

evitably occur in the world. Neither "biology," "evo-

lution," "society," or "the environment" directly im-

pose behavioral outcomes, without an immensely

long and intricate intervening chain of causation in-

volving interactions with an entire conﬁguration of

other causal elements. Each link of such a chain of-

fers a possible point of intervention to change the

ﬁnal outcome. For this reason, computer programs

present afar better model of the situation: The com-

puter does nothing without them, they frequently

involve superbly complex contingent branching and

looping alternatives, they can (and the procedures

in the human mind certainly do) take as input en-

vironmental variables that create cascading changes

in subsequent computational events and ﬁnal out-

comes, and the entire system may respond dramati-

cally and dynamically to direct intervention (for ex-

ample, the alteration of even a single instruction) at

any of a great number of locations in the program.

Moreover, we know in advance that the human

psychological system is immensely ﬂexible as to out-

come: Everything that every individual has ever

done in all of human history and prehistory estab-

lishes the minimum boundary of the possible. The

maximum, if any, is completely unknown. Given the

fact that we are almost entirely ignorant of the com-

putational speciﬁcs of the hundreds or thousands of

mechanisms that comprise the human mind, it is far

beyond the present competence of anyone living to

say what are and are not achievable outcomes for

human beings.

It is nevertheless very likely to be the case that we

will ﬁnd adaptive specializations in the human mind

that evolved to make, under certain circumstances,

choices or decisions that are (by most standards)

ethically unacceptable and often lead to con- sensu-

ally undesirable outcomes ( e.g., male sexual propri-

etariness, Wilson & Daly, this volume; discriminative

parental solicitude, Daly & Wilson, 1981; Mann, this

volume). If one is concerned about something like

family violence, however, knowing the details of the

mechanisms involved will prove crucial in taking any

kind of constructive or ameliatory action. "Solutions"

that ignore causation can solve nothing.

In any case, the analysis of the morality or practi-

cality of intervention to prevent undesired outcomes-

"ontogenetic engineering" (Daly, Wilson & Weghorst,

1982)- is beyond the scope of this discussion. Our

point here is simply that leaving behind the SSSM

does not entail accepting the inevitability of any

speciﬁc outcome, nor does it entail the defense of

any particular aspect of the status quo. Instead, for

those genuinely concerned with such questions, it of-

fers the only realistic hope of understanding enough

about human nature to eventually make possible

successful intervention to bring about humane out-

comes. Moreover, a program of social melioration

carried out in ignorance of human complex design is

something like letting a blindfolded individual loose

in an operating room with a scalpel-there is likely

to be more blood than healing. To cure, one needs

to understand; lamenting disease or denouncing the

researchers who study its properties has never yet

saved a life. At present, we are decades away from

having a good model of the human mind, and this

is attributable in no small measure to a misguided

antinativism that has, for many, turned from being

a moral stance into a tired way of defending a stag-

nated and sterile intellectual status quo. There are,

of course, no guarantees, and it is at least logically

possible that under- standing our complex array of

evolved mechanisms will offer no way to improve

.the human condition. But, if that is the case, it will

be the ﬁrst time in history that major sets of new dis-

coveries turned out to be useless.

The Empirical Authority of the Standard Social

Science Model

The Division of Labor: Content-independent Psy-

chology

One major consequence of the adoption of the

Standard Social Science Model has been the assign-

ment of a division of labor among the social sci-

ences. It gave each ﬁeld its particular mission,

stamped each of them with its distinctive charac-

ter, and thereby prevented them from making much

progress beyond the accumulation of particularistic

knowledge. Anthropology, as well as sociology and

history, study both the important and variable con-

tent of human life (the signal) and the more vaguely

deﬁned processes and contingent events that gener-

ated it (the artiﬁcer or author of the signal). Psy-

chology studies the medium on which this socially

generated content is inscribed, the process of inscrip-

tion, and the mechanisms that enable the inscription

to take place. (The SSSM also assigns to psychol-

ogy and to psychological anthropology the task of

cataloging, at the individual level, the particularis-

tic psychological phenomena that are created by the

action of each culture on individuals; e.g., what do

American college sophomores get anxious about?).

In advance of any data, the Standard Model de-

ﬁned for psychology the general character of the

mechanisms that it was supposed to ﬁnd (general-

purpose, content- independent ones), its most im-

portant focus (learning), and how it would interpret

the data it found (no matter what the outcome, the

origin of content was to be located externally-for ex-

ample, in the unknowably complex unobserved prior

history of the individual-and not "internally" in the

mind of the organism). Psychologists certainly were

not forced by the character of their data into these

types of conclusions (e.g., Breland & Breland, 1961

). Instead, they had to carefully design their exper-

iments so as to exclude evolutionarily organized re-

sponses to biologically signiﬁcant stimuli by elimi-

nating such stimuli from their protocols (e.g., by us-

ing stimulus-impoverished Skinner boxes or the cur-

rently widespread practice of eliminating "emotion-

ally charged" stimuli from cognitive experiments).

This was done in the name of good experimental de-

sign and with the intention of eliminating contam-

inating "noise" from the exploration of the content-

independent mechanisms that were thought to exist.

The Division of Labor: Particularistic, Content-

Speciﬁc Anthropology

Even more than psychology, anthropology was

shaped by the assumptions inherent in the SSSM’s di-

vision of labor: A content-independent (or content-

free) psychology symbiotically requires a content-

supplying anthropology to provide the agent-culture-

that transforms malleable generalized potential into

speciﬁcally realized human beings. So anthropol-

ogy’s mission was to study the particular (Geertz,

1973, p. 52). Consequently, anthropology became

the custodian of the key explanatory concept in the

paradigm, "culture." Belief in culture, as a substance

passed across generations causing the richly deﬁned

particularity of adult mental and social organization

deﬁnes one’s membership in the modern social sci-

ence community. The invocation of culture became

the universal glue and explanatory variable that held

social science explanations together: Why do par-

ents take care of their children? It is part of the

social role their culture assigns to them. Why are

Syrian husbands jealous? Their culture tied their sta-

tus to their wife’s honor. Why are people sometimes

aggressive? They learn to be because their culture

socializes them to be violent. Why are there more

murders in America than in Switzerland? Americans

have a more individualistic culture. Why do women

want to look younger? Youthful appearance is val-

ued in our culture. And so on.

Although using culture as an ail-purpose explana-

tion is a stance that is difﬁcult or impossible to fal-

sify, it is correspondingly easy to "conﬁrm. " If one

doubts that the causal agent for a particular act is

transmitted culture, one can nearly always ﬁnd sim-

ilar prior acts (or attitudes, or values, or representa-

tions) by others, so a source of the contagion can al-

ways be identiﬁed. Culture is the protean agent that

causes everything that needs explaining in the social

sciences, apart from those few things that can be ex-

plained by content-general psychological laws, a few

drives, and whatever super organic processes (e.g.,

history, social conﬂict, economics) that are used to

explain the particularities of a speciﬁc culture. Psy-

chologists, then, need not explain the origin of com-

plexly speciﬁc local patterns of behavior. They can

be conﬁdent that anthropologists have done this job

and have tracked, captured, deﬁned, and analyzed

the causal processes responsible for explaining why

men are often sexually jealous or why women often

prefer to look youthful.

In deﬁning culture as the central concept of an-

thropology, the SSSM precluded the development of

the range of alternative anthropologies that would

have resulted if, say, human nature, economic and

subsistence activity, ecological adaptation, human

universals, the organization of incentives inside

groups, institutional propagation, species-typical

psychology, or a host of other reasonable possibili-

ties had been selected instead. More critically, be-

cause of the way in which the SSSM frames the rela-

tionship between culture and the human mind, an-

thropology’s emphasis on relativity and explanatory

particularism becomes inescapable, by the following

logic: If the psyche is general-purpose, then all orga-

nized content comes from the outside, from culture.

Therefore, if something is contentful, then it must be

cultural; if it is cultural, then- by the nature of what it

is to be cultural-it is plastically variable; if it is plasti-

cally variable, then there can be no ﬁrm general laws

about it. Ergo, there can be no general principles

about the content of human life (only the content-

less laws of learning). The conclusion is present in

the premises. The relativity of human behavior, far

from being the critical empirical discovery of anthro-

pology (Geertz, 1973, 1984), is something imposed

a priori on the ﬁeld by the assumptions of the SSSM,

because its premises deﬁne a program that is inca-

pable of ﬁnding anything else. Relativity is no more

"there" to be found in the data of anthropologists

than a content-independent architecture is "there" to

be found in the data of psychologists. These con-

clusions are present in the principles by which these

ﬁelds approach their tasks and organize their data,

and so are not "ﬁndings" or "discoveries" at all.

The consequences of this reasoned arrival at par-

ticularism reverberate throughout the social sci-

ences, imparting to them their characteristic ﬂavor,

as compared with the natural sciences. This ﬂa-

vor is not complexity, contingency, or historicity:

Sciences from geology to astronomy to meteorol-

ogy to evolutionary biology have these in full mea-

sure. It is, instead, that social science theories are

usually provisional, indeterminate, tentative, indeﬁ-

nite, enmeshed in an endlessly qualiﬁed explanatory

particular- ism, for which the usual explanation is

that human life is much more complex than mere

Schrödinger equations or planetary ecosystems. Be-

cause culture was held to be the proximate (and

probably the ultimate) cause of the substance and

rich organization of human life, the consensus was,

naturally, that documenting its variability and partic-

ularity deserved to be the primary focus of anthropo-

logical study (e.g., Geertz, 1973). This single propo-

sition alone has proven to be a major contributor to

the failure of the social sciences (Tooby & Cosmides,

1989a). Mainstream sociocultural anthropology has

arrived at a situation resembling some nightmarish

short story Borges might have written, where scien-

tists are condemned by their unexamined assump-

tions to study the nature of mirrors only by cata-

loging and investigating everything that mirrors can

reﬂect. It is an endless process that never makes

progress, that never reaches closure, that generates

endless debate between those who have seen differ-

ent reﬂected images, and whose enduring product is

voluminous descriptions of particular phenomena.

The Empirical Disproof of a Universal Human

Nature.

The view that the essence of human nature lies

in its variousness and the corresponding rejection of

a complex, universal human nature is not advanced

by anthropologists simply as an assertion. Instead,

it is presented as a dramatic and empirically well-

sup- ported scientiﬁc discovery (Geertz, 1984) and

is derived from a particular method through which

the limits of human nature are explored and de-

ﬁned. This method, a logical process of elimina-

tion, "conﬁrmed" that the notion of human nature

was empirically almost vacuous. Since infants are

everywhere the same, then anything that varies in

adults can only (it was reasoned) be cultural and,

hence, socially inherited and, hence, socially man-

ufactured. The method depends on accepting the

premise that behavior can only be accounted for in

these two ways: (1) as something "biological," or in-

born, which is, therefore, inﬂexibly rigid regardless

of environment and (because of the psychic unity of

humankind) everywhere the same, or (2) as socio-

cultural, which includes everything that varies, at a

minimum, and perhaps many things that happen by

accident to be universal as well.

Whenever it is suggested that something is "in-

nate" or "biological," the SSSM- oriented anthropol-

ogist or sociologist rifﬂes through the ethnographic

literature to ﬁnd a report of a culture where the

behavior ( or whatever) varies (for a classic ex-

ample, see Mead’s 1949 Male and Female). Upon

ﬁnding an instance of reported variation (or invent-

ing one through strained interpretation; see again,

Mead, 1949), the item is moved from the category

of "innate," "biological," "genetically determined," or

"hardwired" to the category of "learned," "cultural,"

or "socially constructed." Durkheim succinctly runs

through the process, discussing why sexual jealousy,

ﬁlial piety, and paternal love must be social construc-

tions, despite claims to the contrary: "History, how-

ever, shows that these inclinations, far from being in-

herent in human nature, are often totally lacking. Or

they may present such variations in different human

societies that the residue obtained after eliminat-

ing all these differences-which alone can be consid-

ered of psychological origin-is reduced to something

vague and rudimentary and far removed from the

facts that need explanation" (Durkheim, 1895/1962,

p. 106). Because almost everything human is vari-

able in one respect or another, nearly everything has

been subtracted from the "biologically determined"

column and moved to the "socially determined" col-

umn. The leftover residue of "human nature," af-

ter this process of subtraction has been completed

is weak tea indeed, compared to the rich and en-

gaging list of those dimensions of life where humans

vary. No wonder Geertz (1973) ﬁnds such watered-

down universals no more fundamental or essential

to humans than the behaviors in which humans vary.

Psychologists have, by and large, accepted the pro-

fessional testimony of anthropologists and have, as

part of their standard intellectual furniture, the con-

ﬁdence that other cultures violate virtually every uni-

versal claim about the content of human life. (D. E.

Brown [1991] offers a pivotal examination of the his-

tory and logic of anthropological approaches to hu-

man universals, cultural variation, and biology, and

this entire discussion is informed by his work.)

Discovering Regularities Depends on Selecting Ap-

propriate Frames of Reference

Because of the moral appeal of antinativism, the

process of discrediting claims about a universal hu-

man nature has been strongly motivated. Anthropol-

ogists, by each new claim of discovered variability,

felt they were expanding the boundaries of their dis-

cipline (and, as they thought, of human possibility

itself) and liberating the social sciences from biolog-

ically deterministic accounts of how we are inﬂexibly

constrained to live as we do. This has elevated par-

ticularism and the celebration of variability to cen-

tral values inside of anthropology, strongly asserted

and ﬁercely defended.

The most scientiﬁcally damaging aspect of this dy-

namic has not been the consequent rhetorical em-

phasis most anthropologists have placed on the un-

usual (Bloch, 1977; Goldschmidt, 1960; Symons,

1979; see, especially, D. E. Brown, 1991). As Bloch (

1977, p. 285) says, it is the "professional malpractice

of anthropologists to exaggerate the exotic character

of other cultures." Nor is the most damaging aspect

of this dynamic the professionally cultivated credu-

lousness about claims of wonders in remote parts of

the world, which has led anthropologists routinely

to embrace, perpetuate, and defend not only gross

errors (see Freeman, 1983, on Mead and Samoa;

Suggs, 1971, on Linton and the Marquesas) but also

obvious hoaxes (e.g., Casteneda’s UCLA dissertation

on Don Juan; or the gentle "Tasaday," which were

manufactured by ofﬁcials of the Marcos regime).

The most scientiﬁcally damaging aspect of this

value system has been that it leads anthropologists to

actively reject conceptual frameworks that identify

meaningful dimensions of cross-cultural uniformity

in favor of alternative vantage points from which cul-

tures appear maximally differentiated. Distinctions

can easily be found and endlessly multiplied, and it

is an easy task to work backward from some partic-

ular difference to ﬁnd a framework from which the

difference matters (e.g., while "mothers" may exist

both there and here, motherhood here is completely

different from motherhood there because mothers

there are not even conceptualized as being blood kin,

but rather as the wife of one’s father, etc., etc.). The

failure to view such variation as always profoundly

differentiating is taken to imply the lack of a sophis-

ticated and professional appreciation of the rich de-

tails of ethnographic reality.

But whether something is variable or constant is

not just "out in the world"; it is also a function of the

system of categorization and description that is cho-

sen and applied. The distance from Paris to Mars is

complexly variable, so is the location of Paris "con-

stant" and "inﬂexible" or is it "variable?" In geogra-

phy, as in the social sciences, one can get whichever

answer one wants simply by choosing one frame of

reference over another. The order that has been

uncovered in physics, for example, depends on the

careful selection of those particular systems of de-

scription and measure that allow in variances to ap-

pear. These regularities would all disappear if physi-

cists used contingently relative deﬁnitions and mea-

sures, such as their own heartbeat to count units of

time (the speed of light would slow down every time

the measurer got excited-"relativity" indeed).

Other sciences select frameworks by how much

regularity these frameworks allow

them to uncover. In contrast, m()st anthropolo-

gists are disposed to select their frame- works so as

to bring out the maximum in particularity, contin-

gency, and variability (e.g., how are the people they

study unique?). Certainly one of the most rewarded

of talents inside anthropology is the literary ability

to express the humanly familiar and intelligible as

the exotic (see, e.g., Geertz’s description of a raid by

the authorities on a cock ﬁght in Bali; Geertz, 1973;

see D. E. Brown, 1991, for a lucid dissection of the

role of universals in this example, and Barkow, 1989,

on how Balinese cock ﬁghting illustrates the conven-

tional psychology of prestige). Anthropologists’ at-

traction to frameworks that highlight particularistic

distinctions and relationships has nearly precluded

the accumulation of genuine knowledge about our

universal design and renders anthropologists’ "em-

pirically" grounded dismissal of the role of biology

a matter of convention and conjuring rather than a

matter of fact.

Beneath Variable Behavior Lie Universal Mech-

anisms

The social science tradition of categorizing every-

thing that varies as "nonbiological" fails to identify

much that is "biological." This is because anthropolo-

gists have chosen ill-suited frames of reference (such

as those based on surface "behavior" or reﬂective

"meaning") that accentuate variability and obscure

the underlying level of universal evolved architec-

ture. There may be good reasons to doubt that the

"behavior" of marriage is a cross-cultural universal or

that the articulated "meaning" , of gender is the same

across all cultures, but there is every reason to think

that every human ( of a given sex) comes equipped

with the same basic evolved design (Tooby & Cos-

mides, 1990a). The critical question is not, for ex-

ample, whether every human male in every culture

engages in jealous behaviors or whether mental rep-

resentations attached to situations of extra-pair mat-

ing are the same in every culture; instead, the most

illuminating question is whether every human male

comes endowed with developmental programs that

are designed to assemble (either conditionally or re-

gardless of normal environmental variation) evolu-

tionarily designed sexual jealousy mechanisms that

are then present to be activated by appropriate cues.

To discern and rescue this underlying universal de-

sign out of the booming, buzzing confusion of ob-

servable human phenomena requires selecting ap-

propriate analytical tools and frames of reference.

Genetics, for example, had enormous difﬁculty

making progress as a science until geneticists

developed the distinction between genotype and

phenotype-between the inherited basis of a trait and

its observable expression. We believe a similar dis-

tinction will prove necessary to the development of

an integrated social science. We will refer to this as

the distinction between the evolved (as in evolved

mechanisms, evolved psychology, evolved architec-

ture, etc.) and the manifest (as in manifest psychol-

ogy, manifest behavior, etc. ). One observes vari-

able manifest psychologies or behaviors between in-

dividuals and across cultures and views them as the

product of a common, under- lying evolved psychol-

ogy, operating under different circumstances. The

mapping between the evolved architecture and man-

ifest behavior operates according to principles of ex-

pression that are speciﬁed in the evolved develop-

mental mechanisms and the psychological mecha-

nisms they reliably construct; manifest expressions

may differ between individuals when different envi-

ronmental inputs are operated on by the same pro-

cedures to produce different manifest outputs (Cos-

mides & Tooby, 1987; Tooby & Cosmides, 1989b).

For example, some individuals speak English while

others do not, yet everyone passes through a life

stage when the same species-typical language acqui-

sition device is activated (Pinker & Bloom, this vol-

ume). In fact, if an individual survives a child- hood

of aberrant social isolation she may never acquire

a language and may be incapable of speaking; yet,

she will have had the same species-typical language

acquisition device as everyone else. So what at the

behavioral level appears variable ("speaks English,"

"speaks Kikuyu"; or, even, "speaks a language," "does

not speak any language") fractionates into variable

environmental inputs and a uniform underlying de-

sign, interacting to produce the observed patterns of

manifest variation. The fog

enveloping most social science debates would

blow away if all hypotheses were completely spelled

out, through analyzing each situation into environ-

mental conditions, evolved architecture, and how

their interaction produces the manifest outcome.

Standard Model partisans have conﬁdently rested

their empirical case on what now appears to be un-

certain ground: that manifest universality across cul-

tures is the observation that evolutionary hypothe-

ses about human nature require and that, on the

other hand, cross-cultural variability establishes that

the behavior in question is uncontaminated by "biol-

ogy" and is, instead, solely the product of "culture"

or "social processes." The recognition that a univer-

sal evolved psychology will produce variable man-

ifest behavior given different environmental condi-

tions exposes this argument as a complete non se-

quitur. From a perspective that describes the whole

integrated system of causation, the distinction be-

tween the biologically determined and the nonbio-

logically determined can be seen to be a nondistinc-

tion.

In its place, the relevant distinction can be drawn

between what Mayr (among others) called open and

closed behavior programs (Mayr, 1976). This ter-

minology distinguishes mechanisms that are open to

factors that commonly vary in the organism’s nat-

ural environment and, hence, commonly vary in

their manifest expression from those that are closed

to the inﬂuence of such factors and are, conse-

quently, uniform in their manifest expression. The

human language acquisition device is an open be-

havior program whose constructed product, adult

competency in the local language, varies depend-

ing on the language community in which the indi-

vidual is raised. Certain facial emotional displays

that manifest themselves uniformly cross-culturally

(Ekman, 1973) may be examples of closed behav-

ior programs. The Standard Social Science Model’s

method of sorting behavior by its cross-cultural uni-

formity or variability of expression into "biologically

determined" and "socially determined" categories in

reality sorts behaviors into those generated by closed

behavior programs, and those generated by open be-

havior programs. In neither case can the analysis

of the "determination" of behavior be made inde-

pendent of "biology," that is, independent of under-

standing the participation of the evolved architec-

ture. For this reason, the whole incoherent oppo-

sition between socially determined (or culturally de-

termined) phenomena and biologically determined

(or genetically determined) phenomena should be

consigned to the dustbin of history, along with the

search for a biology-free social science.

The Search for the Artiﬁcer

If psychology studies the content-independent

laws of mind and anthropology studies the content-

supplying inheritances of particular cultures, one

still needs to ﬁnd the content-determining processes

that manufacture individual cultures and social sys-

tems. The Standard Social Science Model breaks the

social sciences into schools (materialist, structural-

functionalist, symbolic, Marxist, postmodernist, etc.

) that are largely distinguished by how each at-

tempts to afﬁrmatively characterize the artiﬁcer,

which they generally agree is an emergent group-

level process of some kind. It is far beyond the scope

of this chapter to review and critique these attempts

to discover somewhere in the social system what is in

effect a generative computational system. Neverthe-

less, it is important to recognize that the net effect

of the central logic of the Standard Social Science

Model has been to direct the quest for the ultimate

cause or generator of signiﬁcant mental and social

organization outward away from the rich computa-

tional architecture of the human mind. It is there

where sufﬁciently powerful ordering processes ones

capable of explaining the phenomena-are primarily

to be found. As will be discussed later, it is there

where the actual generators of organization are prin-

cipally (though not exclusively) located and could be

productively investigated. And understanding this

architecture is an indispensable ingredient in mod-

eling or under- standing whatever super-individual

processes exist.

This is not to say that there aren’t many important

phenomena and processes operating at the popula-

tion level, which, for example, modify the nature and

distribution of representations (for non-SSSM anal-

yses, see, e.g., Sperber, 1985, 1990, on epidemio-

logical approaches to cultural transmission; see also

Boyd & Richerson, 1985; Campbell, 1965, 1975;

Durham, 1991; and others, who examine analogs

to natural selection operating at the cultural level).

But because the traditional SSSM efforts to charac-

terize these generative processes make them entirely

external to "the individual" as well as independent

of species-typical psychology, these accounts tend to

share a certain ineradicable indeﬁniteness of loca-

tion and substance. The SSSM attempt to abstract

social processes away from the matrix of interact-

ing psychological architectures necessarily fails be-

cause the detailed structure of psychological mech-

anisms is inextricably bound up in how these so-

cial processes operate. One might say that what

mostly remains, once you have removed from the hu-

man world everything internal to individuals, is the

air between them. This vagueness of ontology and

causal mechanism makes it difﬁcult to situate these

hypothetical generative processes with respect to our

knowledge of the rest of the natural world (Sper-

ber, 1986). Moreover, attempt- ing to locate in these

population-level processes the primary generator of

signiﬁcant organization has caused these processes

to be fundamentally misunderstood, mystiﬁed, and

imbued with such unwarranted properties as au-

tonomy, teleology, functionality, organism-like inte-

gration, intelligence, intentionality, emotions, need-

responsiveness, and even consciousness (see, e.g.,

Durkheim, 1895/1962; Harris, 1979; Kroeber, 1915;

Marx, 1867/1909; Merton, 1949; Parsons, 1949;

Radcliffe-Brown, 1952; see Harris, 1968, for review

and discussion).

Of course, the social system is not like a person or

an organism or a mind, self- ordering due to its own

functionally integrated mechanisms. It is more like

an ecosystem or an economy whose relationships are

structured by feedback processes driven by the dy-

namic properties of its component parts. In this case,

the component parts of the population are individual

humans, so any social dynamics must be anchored in

models of the human psychological architecture. In

contrast, the customary insistence on the autonomy

(or analytic separability) of the superorganic level is

why there have been so few successful or convincing

causal models of population-level social processes,

including models of culture and social organization

(apart from those originating in microeconomics or

in analogies drawn from population biology, which

do not usually take SSSM-style approaches; see, e.g.,

Boyd & Richerson, 1985; Schelling, 1978).

Rejecting the design of individuals as central to

the dynamics is fatal to these models, because su-

perorganic (that is, population-level) processes are

not just "out there," external to the individual. In-

stead, human superindividual interactions depend

intimately on the representations and other regula-

tory elements present in the head of every individual

involved and, therefore, on the systems of computa-

tion inside each head. These govern what is selected

from "out there," how this is represented, what pro-

cedures act on the representations, and what behav-

iors result that others can then observe and inter-

act within a population dynamic fashion (Sperber,

1985, 1986, 1990;Tooby & Cosmides, 1989a). These

psychological mechanisms are primarily where there

is sufﬁcient antientropic computational power to ex-

plain the rich patterning of human life. The design of

the human psychological architecture structures the

nature of the social interactions humans can enter

into, as well as the selectively contagious transmis-

sion of representations between individuals. Only

after the description of the evolved human psycho-

logical architecture has been restored as the center-

piece of social theory can the secondary antientropic

effects of population-level social dynamics be fully

assessed and conﬁdently analyzed. Hence, the study

of population-level social and cultural dynamics re-

quires a sophisticated model of human psychology

to undergird it (see Barkow, 1989, this volume; Cos-

mides & Tooby, 1989, this volume; Daly, 1982; Sper-

ber, 1985, 1990; Tooby&Cosmides, 1989a).

The Division of Labor: The Social Sciences ver-

sus the Natural Sciences.

The single most far-reaching consequence of the

Standard Social Science Model has been to intellec-

tually divorce the social sciences from the natural

sciences, with the result that they cannot speak to

each other about much of substance. Where this

divorce has been achieved can be precisely located

within the model. Because biology and evolved psy-

chology are internal to the individual and because

culture-the author of social and mental organization-

is seen as external to the individual, the causal ar-

row from outside to inside logically insulates the so-

cial sciences from the rest of the natural sciences,

making them autonomous and the natural sciences

substantively irrelevant. This set of propositions is

the locus of the primary break between the social

and the natural sciences. Although there has been a

causal ﬂow across four billion years of evolutionary

time, its ability to causally shape the human present

is impermeably dammed at the boundaries of the

individual-in fact, well within the individual, for evo-

lution is thought to provide nothing beyond an ac-

count of the origins of the drives, if any, and of the

general-purpose, content-free learning or computa-

tional equipment that together comprise the SSSM’s

minimalist model of human nature.

Thus, whatever their empirical success may be, the

claims made by (to pick some obvious examples)

ethnologists, sociobiologists, behavioral ecologists,

and evolutionary psychologists about the evolution-

ary patterning of human behavior can be simply dis-

missed out of hand as wrong, without requiring spe-

ciﬁc examination, because causality does not ﬂow

outward from the individual or from psychology, but

inward from the social world (Sahlins, 1976a). Or,

as Durkheim put it nearly a century before, "every

time that a social phenomenon is directly explained

by a psychological phenomenon, we may be sure

that the explanation is false" (Durkheim, 1895/1962,

p. 103). Organic evolution manufactured the bio-

logical substratum, the human capacity for culture-

"the breadth and indeterminateness of [man’s] in-

herent capacities" (Geertz, 1973, p. 45 )-but oth-

erwise reaches a dead end in its causal ﬂow and its

power to explain.

Finally, it would be a mistake t() think that the

Standard Social Science Model reﬂects the views

solely of social scientists and is usually resisted by

biologists and other natural scientists. It is instead

considered the common sense and common decency

of our age. More particularly, it is a very useful doc-

trine for biologists themselves to hold. Many of them

vigorously defend its orthodoxies, adding their pro-

fessional imprimatur to the social scientists’ brief for

the primacy of culture or social forces over "biology"

(see, e.g., Gould, 1977a, 1977b; Lewontin, Rose &

Kamin, 1984). This does not hap- pen simply be-

cause some are drawn to the formidable moral au-

thority of the Standard Model and the mantle it con-

fers. Even for those of a genuinely scientiﬁc tem-

perament, fascinated with biological phenomena for

their own sake, such a doctrine is a godsend. The

Standard Model guarantees them, a priori, that their

work cannot have implications that violate socially

sanctiﬁed beliefs about human affairs because the

Standard Model assures them that biology is intrinsi-

cally disconnected from the human social order. The

Standard Model therefore frees those in the biologi-

cal sciences to pursue their research in peace, with-

out having to fear that they might accidentally stum-

ble into or run afoul of highly charged social or po-

litical issues. It offers them safe conduct across the

politicized mineﬁeld of modem academic life. This

division of labor is, therefore, popular: Natural sci-

entists deal with the nonhuman world and the "phys-

ical" side of human life, while social scientists are the

custodians of human minds, human behavior, and,

indeed, the entire human mental, moral, political,

social, and cultural world. Thus, both social scien-

tists and natural scientists have been enlisted in what

has become a common enterprise: the resurrection

of a barely disguised and archaic physical/mental,

matter/spirit, nature/human dualism, in place of an

integrated scientiﬁc monism.

THE EVOLUTIONARY CONTRIBUTION TO IN-

TEGRA TED EXPLANATION

Rediscovering the Relevance of Evolutionary Bi-

ology

If the adoption of the Standard Social Science

Model has not led to a great deal of natural science-

Iike progress, that is surely not a good argument

against it. Its convenience has no bearing on

whether it is true. What, then, are the reasons for

believing it is false? There are a number of major

problems that independently lead to the rejection of

the SSSM-some emerging from evolutionary biology,

some from cognitive science, and some from their

integration. We will discuss three of these problems,

arguing that (1) the Standard Social Science Model’s

analysis of developmental or "nature-nurture" issues

is erroneous; (2) the general-purpose, content-free

psychology central to the SSSM could not have been

produced by the evolutionary process and, therefore,

is not a viable candidate model of human psychol-

ogy; and (3) a psychology of this kind cannot ex-

plain how people solve a whole array of tasks they

are known to routinely perform.

For advocates of the Standard Social Science

Model, evolution is ignored because it is irrelevant:

The explanatory power of evolution ends with the

emergence of the content-free computational equip-

ment that purportedly constitutes human nature.

This equipment does not impose any form on the so-

cial world, but instead acquires all of its content from

the social world. The supposed erasure of content-

sensitive and content-imparting structure from ho-

minid psychological architecture during our evolu-

tion is what justiﬁes the wall of separation between

the natural and the social sciences. If this view were

correct, then evolution would indeed be effectively

irrelevant to the social sciences and the phenomena

they study.

In contrast, proponents of the Integrated Causal

Model accept that, in addition to whatever content-

independent mechanisms our psychological architec-

ture may contain, it also contains content-speciﬁc

devices, including those computationally responsi-

ble for the generation and regulation of human

cultural and social phenomena. These content-

speciﬁc mechanisms are adaptations (as are content-

independent mechanisms), and evolved to solve

long-enduring adaptive problems characteristic of

our hunter-gatherer past. Because of their design,

their operation continually imparts evolutionarily

patterned content to modern human life. If this

view is correct, then the speciﬁcs of evolutionary bi-

ology have a central signiﬁcance for understanding

human thought and action. Evolutionary processes

are the "architect" that assembled, detail by detail,

our evolved psychological and physiological archi-

tecture. The distinctive characteristics of these pro-

cesses are inscribed in the organizational speciﬁcs

of these designs. Consequently, an understanding

of the principles that govern evolution is an indis-

pensable ally in the enterprise of understanding hu-

man nature and an invaluable tool in the discov-

ery and mapping of the species-typical collection

of information-processing mechanisms that together

comprise the human mind. The complex designs

of these mechanisms are the main causal channels

through which the natural sciences connect to and

shape the substance of the "social" sciences.

Thus, the relevance of evolutionary biology does

not in the least depend on our being "just like" other

species, which we obviously are not (Tooby & De

Vore, 1987). Each species has its own distinctive

properties stemming from its own unique evolution-

ary history .Evolutionary biology is fundamentally

relevant to the study of human behavior and thought

because our species is the product of naturalistic ter-

restrial processes-evolutionary processes-and not of

divine creation or extraterrestrial intervention. How-

ever unusual our properties may be from a zoologi-

cal point of view- and we have every reason to be-

lieve humans followed a unique evolutionary trajec-

tory (Tooby & De V ore, 1987)-we need an account

of how they were produced in the natural world of

causation over evolutionary history (Boyd & Richer-

son, 1985). Such accounts are constructed from ( I)

the principles that govern the evolutionary process

(such as natural selection and drift), (2) the designs

of ancestral hominid species, and (3) the particu-

lar ancestral environments and contingent historical

events hominids encountered during their evolution-

ary history.

Reproduction, Feedback, and the Construction

of Organic Design

These elaborately constructed forms, so dif-

ferent from each other, and dependent on

each other in so complex a manner, have

all been produced by laws acting around us.

These laws, taken in the largest sense, be-

ing. ..Reproduction; Inheritance which is

almost implied by reproduction; Variability,

and as a consequence. Natural Selection.

...

-CHARLES DARWIN, The Origin of Species

While physicists tend to start their causal history

with the Big Bang, biologists usually select a differ-

ent, later event: the emergence of the ﬁrst living or-

ganism. Life ( or the instances we have so far ob-

served) is a phenomenon that originated on earth

three to four billion years ago through the forma-

tion of the ﬁrst living organism by contingent phys-

ical and chemical processes. What is life? What

deﬁning properties qualiﬁed some ancient physical

system as the ﬁrst living organism? From a Dar-

winian perspective, it is the reproduction by sys-

tems of new and similarly reproducing systems that

is the deﬁning property of life. An organism is a

self-reproducing machine. All of the other properties

of living organisms-cellular structure, A TP, polypep-

tides, the use of car- bon’s ability to form indeﬁnitely

large chains, DNA as a regulatory element-are inci-

dental rather than essential, and the logic of Darwin-

ism would apply equally to self- reproducing robots,

to self-reproducing plasma vortices in the sun, or

to anything else that reproduces with the potential

for inheritable change (mutation). From reproduc-

tion, the deﬁning property of life, the entire elegant

deductive structure of Darwinism follows (Dawkins,

1976; Tooby & Cosmides, 1990b; Williams, 1985).

Very simple proto-bacteria emerged early in terres-

trial history, as chemical machines that constructed

additional chemical machines like themselves. Be-

cause reproduction means the construction of off-

spring designs like the parent machines, one could

imagine this leading to an endless chain of prolifer-

ating systems identical to the original parent. This

is not what happened, of course, because muta-

tions or random modiﬁcations are sometimes intro-

duced into offspring designs by accident, with far-

reaching consequences. Most random modiﬁcations

introduce changes into the organism’s organization

that interfere with the complex sequence of actions

necessary for self-reproduction. But a small propor-

tion of random modiﬁcations happen to cause an

enhancement in the average ability of the design

to cause its own reproduction. In the short run,

the frequency of those variants whose design pro-

motes their own reproduction increases, and the fre-

quency of those variants whose design causes them

to produce fewer (or no) offspring decreases. Con-

sequently, one of two outcomes usually ensues: ( 1)

the frequency of a design will drop to zero-i.e., go

extinct (a case of negative feedback); or (2) a de-

sign will outreproduce and thereby replace all alter-

native designs in the population (a case of positive

feedback). After such an event, the population of

reproducing machines is different from its ancestors

because it is equipped with anew and more function-

ally organized design or architecture. Thus, the fact

that alternative design features give rise to repro-

ductive performance differences creates the system

of positive and negative feedback called natural se-

lection. Natural selection guides the incorporation

of design modiﬁcations over generations according

to their consequences on their own reproduction.

Over the long run, down chains of descent, this

cycle of chance modiﬁcation and reproductive feed-

back leads to the systematic accretion within archi-

tectures of design features that promote or formerly

promoted their own propagation. Even more impor-

tantly, the reproductive fates of the inherited proper-

ties that coexist in the same organism are linked to-

gether: What propagates one design feature tends to

propagate others (not perfectly, but sufﬁciently for a

coherent design to emerge; see Cosmides & Tooby,

1981). This means that traits will be selected to

work together to produce the same outcomes and

to enhance each other’s functionality. Frequently,

then, these accumulating properties will sequen-

tially ﬁt themselves together into increasingly func-

tionally elaborated machines for reproduction, com-

posed of constituent mechanisms-called adaptations-

that solve problems that are either necessary for re-

production or increase its likelihood (Darwin, 1859;

Dawkins, 1986; R. Thornhill, 1991 ; Tooby & Cos-

mides, 1990b; Williams, 1966, 1985). As if by the

handiwork of an invisible and nonforesightful engi-

neer, element after element is added to a design over

generations, making it a more functional system for

propagation under the conditions prevailing at the

time each new element was added. At present, there

is no extant alter- native theory for how organisms

acquired complex functional organization over the

course of their evolution (Dawkins, 1986).

What is most compelling about Darwin’s approach

is that the process of natural selection is an inevitable

by-product of reproduction, inheritance, and mu-

tation. Like water running downhill, over genera-

tions organisms tend to ﬂow into new functional de-

signs better organized for effective propagation in

the environmental context in which they evolved.

There is, however, another method, besides selec-

tion, by which mutations can become incorporated

into species-typical design: chance. The sheer im-

pact of many random accidents may cumulatively

propel a useless mutation upward in frequency un-

til it crowds out all alternative design features from

the population. Its presence in the architecture is

not explained by the (nonexistent) functional conse-

quences it has on reproduction, and so it will not be

coordinated with the rest of the organism’s architec-

ture in a functional way.

But despite the fact that chance plays some role

in evolution, organisms are not primarily chance

agglomerations of stray properties. To the extent

that a feature has a signiﬁcant effect on reproduc-

tion, selection will act on it. For this reason, impor-

tant and consequential aspects of organism architec-

tures are shaped by selection. By the same token,

those modiﬁcations that are so minor that their con-

sequences are negligible on reproduction are invisi-

ble to selection and, therefore, are not organized by

it. Thus, chance properties drift through the stan-

dard designs of species in a random way, unable to

account for complex organized design and, corre-

spondingly, are usually peripheralized into those as-

pects that do not make a signiﬁcant impact on the

functional operation of the system (Tooby & Cos-

mides, 1990a, 1990b).

In short, evolution (or descent with modiﬁcation,

to use Darwin’s phrase) takes place due to the action

of both chance and natural selection, causing de-

scendants to diverge in characteristics from their an-

cestors, down chains of descent. Over evolutionary

time, this appears as a succession of designs, each

a modiﬁcation of the one preceding it. Generation

by generation, step by step, the designs of all of the

diverse organisms alive today-from redwoods and

manta rays to humans and yeast-were permuted out

of the original, very simple, single-celled ancestor

through an immensely long sequence of successive

modiﬁcations. Each modiﬁcation spread through the

species either because it caused its own propagation,

or by accident. Through this analytic framework, liv-

ing things in general and each species in particular

can be located in the principled causal history of the

universe. Moreover, the speciﬁc design features of a

species’ architecture can also be causally located in

this history: they exist either because of chance in-

corporation or because they contributed to the func-

tional operation of the architecture. The theory of

evolution by natural selection vastly expanded the

range of things that could be accounted for, so that

not only physical phenomena such as stars, moun-

tain ranges, impact crater’s, and alluvial fans could

be causally located and explained but also things like

whales, eyes, leaves, nervous systems, emotional ex-

pressions, and the language faculty.

The modern Darwinian theory of evolution con-

sists of the logically derivable set of causal princi-

ples that necessarily govern the dynamics of repro-

ducing systems, accounting for the kinds of prop-

erties that they cumulatively acquire over succes-

sive generations. The explicit identiﬁcation of this

core logic has allowed the biological community to

develop an increasingly comprehensive set of prin-

ciples about what kinds of modiﬁcations can and

do become incorporated into the designs of repro-

ducing systems down their chains of descent, and

what kinds of modiﬁcations do not (Dawkins, 1976,

1982, 1986; Hamilton, 1964, 1972; Maynard Smith,

1964, 1982; Williams, 1966). This set of princi-

ples has been tested, validated, and enriched by a

comprehensive engagement with the empirical re-

ality of the biological world, from functional and

comparative anatomy, to biogeography, to genetics,

to immunology, to embryology, to behavioral ecol-

ogy, and so on. Just as the ﬁelds of electrical and

mechanical engineering summarize our knowledge

of principles that govern the design of human-bui1t

machines, the ﬁeld of evolutionary biology summa-

rizes our knowledge of the engineering principles

that govern the design of organisms, which can be

thought of as machines built by the evolutionary pro-

cess (for good summaries, see Daly & Wilson 1984a;

Dawkins, 1976,1982, 1986; Krebs & Davies, 1987).

Modern evolutionary biology constitutes, in effect,

an "organism design theory." Its principles can be

used both to evaluate the plausibility of the psychol-

ogy posited by the Standard Social Science Model

and to guide the construction of a better successor

psychology.

The Peculiar Nature of Biological Functionality

In certain narrowly delimited ways, then, the

spontaneous process of evolution parallels the in-

tentional construction of functional machines by hu-

man action. But, whereas machines built by human

engineers are designed to serve a diverse array of

ends, the causal process of natural selection builds

organic machines that are "designed" to serve only

one very specialized end: the propagation into sub-

sequent generations of the inherited design features

that comprise the organic machine itself.

Because design features are embodied in organ-

isms, they can, generally speaking, propagate them-

selves in only two ways: (I) by solving problems that

will increase the probability that the organism they

are situated in will produce offspring, or (2) by solv-

ing problems that will increase the probability that

the organism’s kin will produce offspring (Hamilton,

1964; Williams & Williams, 1957). An individual’s

relatives, by virtue of having descended from a com-

mon ancestor, have an increased likelihood of having

the same design feature as compared to other con-

speciﬁcs, so their increased reproduction will tend to

increase the frequency of the design feature. Accord-

ingly, design features that promote both direct re-

production and kin reproduction, and that make ef-

ﬁcient trade-offs between the two will replace those

that do not. To put this in standard biological termi-

nology, design features are selected for to the extent

that they promote their inclusive ﬁtness (Hamilton,

1964). For clarity, we will tend to call this propaga-

tion or design-propagation.

Selection, then, is the only known account for the

natural occurrence of complexly organized function-

ality in the inherited design of undomesticated or-

ganisms. More- over, selection can only account

for functionality of a very narrow kind: approxi-

mately, design features organized to promote the re-

production of an individual and his or her relatives

(Dawkins, 1986; Williams, 1966). Fortunately for

the modern theory of evolution, the only naturally

occurring complex functionality that has ever been

documented in plants, animals, or other organisms is

functionality of just this kind, along with its deriva-

tives and by-products. Mammals evolved adapta-

tions to produce milk to feed their own young, infec-

tious micro-organisms mimic human biochemistry to

escape immune surveillance so they can survive and

reproduce, and plants produce oxygen as a waste

product of feeding themselves through photosynthe-

sis, and not for the pleasure of watching humans

breathe or forests burn. Of course, human breeders

artiﬁcially intervene, and one could easily imagine,

as an alternative to a Darwinian world, a benevolent

deity or extraterrestrial being creating the properties

of living things in order to serve human convenience

rather than the organisms’ own reproduction. Wild

horses could be born with saddle-shaped humps, lug-

gage racks, and a spontaneous willingness to be rid-

den; chronic bacterial infections could jolt humans

with caffeine every morning 45 seconds before they

need to get up. Similarly, the non- living world could

be full of intricate functional arrangements not cre-

ated by humans, such as mountains that naturally

mimic hotels down to the details of closet hangers,

electric wiring, and television sets. But this is not

the world we live in. We live in a world of biological

functionality and its derivatives, traceable originally

to the operation of natural selection on reproducing

systems (Darwin, 1859; Dawkins, 1976, 1986). To

be able to understand the world of biological phe-

nomena, one must be able to recognize this peculiar

functional organization and distinguish it from the

products of chance.

Of course, the fact that living things are machines

organized to reproduce them- selves and their kin

does not mean that evolutionary functional anal-

ysis focuses narrowly on such issues as copula-

tion or pregnancy (things intuitively associated with

reproduction) over, say, taste preferences, vision,

emotional expression, social categorization, coali-

tion formation, or object recognition. A life his-

tory of successfully achieved reproduction (includ-

ing kin reproduction) requires accomplishing the en-

tire tributary network of preconditions for and facil-

itations of reproduction in complex ecological and

social environments. Of course, this includes all of

the information- gathering, inference, and decision-

making that these tasks entail. For this reason,

humans display a diverse range of adaptations de-

signed to perform a wide and structured variety of

subsidiary tasks, from solicitation of assistance from

one’s parents, to language acquisition, to modeling

the spatial distribution of local objects, to reading

the body language of an antagonist.

Finally, the behavior of individual organisms is

caused by the structure of their adaptations and

the environmental input to them; it is not indepen-

dently governed by the principle of individual ﬁt-

ness maximization. Individual organisms are best

thought of as adaptation-executers rather than as

ﬁtness-maximizers. Natural selection cannot directly

"see" an individual organism in a speciﬁc situation

and cause behavior to be adaptively tailored to the

functional requirements imposed by that situation.

To understand the role of selection in behavior, one

must follow out all steps in the chain: Selection act-

ing over evolutionary time has constructed the mech-

anisms we have inherited in the present, and it is

this set of mechanisms that regulates our behavior-

not natural selection directly. These mechanisms sit-

uated in particular individuals frequently-but by no

means always-bring about a functional coordination

between the adaptive demands of particular situa-

tions and associated behavioral responses.

Thus, the biological concept of functionality dif-

fers from the folk notion of functionality as goal-

seeking behavior. Although some of our evolved

psychological mechanisms probably operate through

goal-seeking, surely none of them has ﬁtness-

maximization as a mentally represented goal (see

Symons, this volume). Those goal-seeking mecha-

nisms that do exist most likely embody proximate

goals, such as "stay warm" or "protect your infant,"

rather than ultimate goals, such as "maximize your

ﬁtness" or "have as many offspring as possible." In-

deed, goals of the latter kind are probably impossible

to instantiate in any computational system (Symons,

1987, 1989, this volume; see also Barkow, 1989;

Cosmides & Tooby, 1987, 1992; Daly & Wilson,

1988; Irons, 1983, p. 200; Tooby & Cosmides,

1990b. For somewhat contrary views, see, e.g.,

Alexander, 1979, 1987 and Turke, 1990).

For this reason, an adaptationist approach does

not properly involve explaining or interpreting indi-

vidual [behavior in speciﬁc situations as "attempts"

to increase ﬁtness (Symons, 1989, this volume;

Tooby & Cosmides, 1990b). To make the distinc-

tion between these alternative views of evolution-

ary explanation clear-humans as ﬁtness- maximiz-

ers (ﬁtness-teleology) versus humans as adaptation-

executors (adaptation- ism)-a brief example will

serve. Fitness teleologists may observe a situation

and ask something like, "How is Susan increasing her

ﬁtness by salting her eggs?" An adaptationist would

ask, instead, "What is the nature of the evolved hu-

man salt preference mechanisms-if any-that are gen-

erating the observed behavior and how did the struc-

ture of these mechanisms mesh with the physiolog-

ical requirements for salt and the opportunities to

procure salt in the Pleistocene?" So, in viewing cases

of behavior, the adaptationist question is not, "How

does this or that action contribute to this particular

individual’s reproduction?" Instead, the adaptation-

ist questions are, "What is the underlying panhuman

psychological architecture that leads to this behav-

ior in certain speciﬁed circumstances?" and "What

are the design features of this architecture- if any-

that regulate the relevant behavior in such away

that it would have constituted functional solutions

to the adaptive problems that regularly occurred in

the Pleistocene?

What Adaptations Look Like.

For the reasons outlined above, the species-typical

organization of the psychology and physiology of

modern humans necessarily has an evolutionary ex-

planation and an evolutionarily patterned architec-

ture. This is not a vague speculation or an over-

reaching attempt to subsume one discipline inside

another, but constitutes as solid a fact as any in mod-

ern science. In fact, this conclusion should be a wel-

come one because it is the doorway through which a

very rich body of additional knowledge-evolutionary

biology-can be brought to bear on the study of psy-

chological architecture. At its core, the discovery of

the design of human psychology and physiology is

a problem in reverse engineering: We have working

exemplars of the design in front of us, but we need to

organize our sea of observations about these exem-

plars into a map of the causal structure that accounts

for the behavior of the system. Psychology has never

been limited by a lack of observations. Fortunately,

the knowledge that humans are the product of evolu-

tion supplies us with a powerful set of tools-the con-

cepts of evolutionary functional analysis-for organiz-

ing these observations into useful categories so that

the underlying systems of order can be discerned.

The most illuminating level of description for or-

ganizing observations about living species is usu-

ally in terms of their adaptations (and associated

evolutionary categories). This system of description

has some warrant on being considered a privileged

frame of reference because the complex functional

organization that exists in the design of organisms

was injected into them through the construction of

adaptations by natural selection. Adaptations are

the accumulated output of selection , and selection is

the single signiﬁcant anti-entropic or ordering force

orchestrating functional organic design (Dawkings,

1986). So if one is interested in uncovering intelligi-

ble organization in our species-typical psychological

architecture, discovering and describing its adapta-

tions is the place to begin.

To understand what complex adaptations "look

like," it will help to begin concretely with a standard

example, the vertebrate eye and its associated neural

circuitry . (For its role in understanding adaptations,

see Pinker & Bloom, this volume; for a discussion

of color vision, see Shepard, this volume.) The eye

consists of an exquisitely organized arrangement of

cells, structures, and processes, such as ( 1) a trans-

parent protective outer coating, the cornea; (2) an

opening, the pupil, through which light enters; (3)

an iris, which is a muscle that surrounds the pupil

and constricts or dilates the aperture, regulating the

amount of light entering the eye; (4) a lens, which

is both trans- parent and ﬂexible, and whose curva-

ture and thickness can be adjusted to bring objects of

varying distances into focus; (5) the retina, a light-

sensitive surface that lies in the focal plane of the

lens: this multilayered neural tissue lining the inside

back of the eye- ball is, in effect, a piece of the brain

that migrated to the eye during fetal development;

(6) classes of specialized cells (rods and cones) in

the retina that transform sampled properties of am-

bient light through selective photochemical reactions

into electro- chemical impulses; (7) the activation by

these electrochemical impulses of neighboring bipo-

lar cells, which, in turn, feed signals into neighbor-

ing ganglion cells, whose axons converge to form

the optic nerve; (8) the optic nerve, which carries

these signals out of the eye and to the lateral genicu-

late bodies in the brain; (9) the routing of these sig-

nals to the visual cortex, into a series of retinotopic

maps and other neural circuits, where they are fur-

ther analyzed by a formidable array of information-

processing mechanisms that also constitute crucial

parts of the visual system.

The dynamic regulatory coordination present in

the operation of the eye is also striking: The vari-

able aperture modulates the amount of light entering

the eye in coordination with ambient illumination;

the eyes are stereoscopically coordinated with each

other so that their lines of vision converge on the

same object or point of interest; the thickness and

curvature of the lens is modulated so that light from

the object being viewed is focused on the retina and

not in front of it or behind it; and so on. Through a

more detailed description, this list could easily be ex-

tended to include many thousands of specialized fea-

tures that contribute to the functionality of the sys-

tem (tear ducts, eyelids, edge detectors, muscle sys-

tems, speciﬁc photochemical reactions, and so on)

through the orchestrated arrangement of hundreds

of millions of cells. This is even more true if we

were to go beyond a taxonomically generalized de-

scription of the vertebrate eye and relate speciﬁc de-

sign features in particular vertebrates to the particu-

lar environments and visual tasks they faced. Frogs,

for example, have retinal "bug detectors"; the rab-

bit retina has a variety of specialized devices, includ-

ing a "hawk detector" (Marr, 1982, p. 32), and so

on. It is important to appreciate that this organi-

zation is not just macroscopic, but extends down to

the organized local relationships that subsets of cells

maintain with each other, which perform such com-

putations as edge detection and bug detection-and

beyond, down into the speciﬁc architecture of the

constituent cells themselves. Thus, rods and cones

have a distinctive design and layout that includes

specialized organelles that adjust the size and shape

of the photoreactive regions; they have membranes

that fold back on themselves to form sacs localizing

the photoreceptive pigments; they have specialized

chemistry so that light induces these pigments to un-

dergo chemical changes that ultimately result in a

change of membrane potential; they are arranged so

that this change of membrane potential effects the

release of neurotransmitters to neighboring bipolar

glion cells, and soon.

Thus, the eye is an extraordinarily complex ar-

rangement of specialized features that does some-

thing very useful for the organism. Moreover, this

structure was originally absent from the ancestral

design of the original single-celled founding organ-

ism, so the appearance of eyes in modern organ-

isms must be explained as a succession of modi-

ﬁcations across generations away from this initial

state. It is easy to see how selection, through re-

taining those accidental modiﬁcations that improved

performance, could start with an initial accidentally

light-sensitive nerve ending or regulatory cell and

transform it, through a large enough succession of

increasingly complex functional forms, into the su-

perlatively crafted modern eye (see, e.g., Dawkins,

1986). In fact, eyes (light-receptive organs) have

evolved independently over 40 times in the history

of animal life from eyeless ancestral forms (Mayr,

1982).

Of course, there are certainly many nonselection-

ist processes in evolution by which descendants are

modiﬁed away from ancestral forms-drift, macromu-

tation, hitchhiking, developmental by-product, and

so on. But selection is the only process that directs

change by retaining variants that are more func-

tional. Thus, selection is the only causal process

that has a systematic tendency to propel the system

in the direction of increasingly functional arrange-

ments, instead of into the immeasurably larger ar-

ray of nonfunctional arrangements that the system

could move to at each of the innumerable choice

points in the evolution of designs. In contrast, nonse-

lectional processes can produce functional outcomes

only by chance because a new modiﬁcation’s degree

of functionality plays no role in determining whether

nonselectional processes will cause it to be retained

or eliminated. For this reason, evolutionary pro-

cesses other than selection are properly classiﬁed as

"chance processes" with respect to the evolution of

adaptive complexity. It would be a coincidence of

miraculous degree if a series of these function-blind

events, brought about by drift, by-products, hitch-

hiking, and so on, just happened to throw together

a structure as complexly and interdependently func-

tional as an eye (Dawkins, 1986; Pinker & Bloom,

this volume). For this reason, nonselectionist mech-

anisms of evolutionary change cannot be seen as pro-

viding any reasonable alternative explanation for the

eye or for any other complex adaptation. Complex

functional organization is the signature of selection.

The eye is by no means a unique case. Immunol-

ogists, for example, have traced out a similar, im-

mensely articulated architecture of complexly inter-

related defenses (the blood monocytes, histiocytes,

free macrophages, T -lymphocytes, B-Iymphocytes,

spleen, thymus, and so on). In fact, virtually every

organ that has been examined so far betrays a com-

plex functionality unmatched as of yet by any system

engineered by humans. More than a century of re-

search and observation conﬁrms that selection builds

into organisms a complex functional organization of

an eye-like precision and quality.

Still, although many social and biological scien-

tists are willing to concede that the body is full of

the most intricately functional machinery, heavily or-

ganized by natural selection, they remain skeptical

that the same is true of the mind. Moreover, parti-

sans of the Standard Social Science Model insist on

the Cartesian distinction between the material world

of anatomy and physiology and the mental world of

psychology, vigorously resisting attempts to see them

as different descriptions of the same integrated sys-

tem, subject to the same organizing principles. Argu-

ments by Chomsky and others that our psychological

architecture should contain "mental organs" for the

same reasons that the rest of the body contains phys-

ical organs (i.e., that different tasks require func-

tionally different solutions) have yet to convince the

majority of psychologists out- side of perception and

language (Chomsky, 1975; Marshall, 1981).

Thus, Lewontin is expressing a thoroughly ortho-

dox SSSM skepticism toward the idea that the hu-

man psychological architecture is functionally orga-

nized when he suggests that human cognition may

have developed as the purely epiphenomenal con-

sequence of the major increase in brain size, which,

in fact, may have been selected for quite other rea-

sons" (Lewontin, 1990, p. 244). At least as nu-

merous are those researchers who detect in hu-

man thought and behavior something more than the

sheer accident that Lewontin sees, and yet who ask:

Aren’t psychological (or neurophysiological) mecha-

nisms expected to be less well-engineered than phys-

iological organs? High degrees of functionality are

all very well for eyes, intestines, and immune sys-

tems, but what about the constituent structures of

the human psychological architecture? Are there at

least any examples of well-engineered psychological

adaptations that might parallel physiological adapta-

tions?

What is most ironic about this question is that

perhaps the single most uncontroversial example of

an adaptation-an example that is conceded to be

well-engineered by even the most exercised of the

anti-adaptationist is a psychological adaptation: the

eye. As Epicharmus pointed out two and half mil-

lennia ago, "Only mind has sight and hearing; all

things else are deaf and blind. " The eye and the

rest of the visual system perform no mechanical or

chemical service for the body; it is an information-

processing adaptation. This information-processing

device is designed to take light incident on a

two-dimensional body surface and-through apply-

ing information-processing procedures to this two-

dimensional array-construct cognitive models of the

local three-dimensional world, including what ob-

jects are present, their shapes, their locations, their

orientations, their trajectories, their colors, the tex-

tures of their surfaces, as well as face recognition,

emotional expression recognition, and so on. In-

deed, for those committed to a Cartesian world view,

one could think of the eye as a tube that traverses

metaphysical realms, one end of which obtrudes into

the physical realm, the other into the mental. For

modern monists, however, these two realms are sim-

ply alternative descriptions of the same thing, con-

venient for different analytic purposes. The "mental"

consists of ordered relationships in physical systems

that embody properties typically running under la-

bels such as "information," "meaning," or regulation.

From this point of view, there is no Cartesian tube:

both ends of the visual system are physical and both

are mental.

Because psychologists as a community have been

resistant to adaptationist thinking, it was an enor-

mous (although nonaccidental) stroke of good for-

tune that the visual system extrudes a "physical end"

to the surface of the body and that this "physical

end" bears a remarkable resemblance to the camera,

a functional machine designed by humans. Selec-

tion has shaped the physical structure of the eye so

that it reﬂects and exploits the properties of light,

the geometry of the three-dimensional world, the re-

fracting properties of lenses, and so on; the camera

has a similar structure because it was designed by

human engineers to reﬂect and exploit these same

properties. These parallels between camera and eye

were clues that were so obvious and so leading that

it became a reasonable enterprise to investigate the

visual system from a functionalist perspective. Re-

searchers started with the physical end and followed

the "Cartesian

tube" upward and inward, so to speak, into the

mind. In so doing, they have discovered increas-

ingly complex and specialized computational ma-

chinery: edge detectors, motion detectors, shape de-

tectors, depth boundary detectors, bug detectors (in

the third neural layer of the retina of frogs), stereo-

scopic disparity analyzers, color constancy mecha-

nisms, face recognition systems, and on and on.

Hundreds of vision researchers, working over

decades, have been mapping this exquisitely struc-

tured information-processing adaptation, whose

evolutionary function is scene analysis-the recon-

struction of models of real-world conditions from

a two-dimensional visual array. As more and more

functional subcomponents are explored, and as arti-

ﬁcial intelligence researchers try to duplicate vision

in computational systems attached to electronic cam-

eras, four things have become clear (Marr, 1982;

Poggio, Torre, & Koch, 1985). The ﬁrst is that the

magnitude of the computational problem posed by

scene analysis is immensely greater than anyone had

suspected prior to trying to duplicate it. Even some-

thing so seemingly simple as perceiving the same

object as having the same color at different times

of the day turns out to require intensely specialized

and complex computational machinery because the

spectral distribution of light reﬂected by the object

changes widely with changes in natural illumination

(Shepard, this volume). The second conclusion is

that as a psychological adaptation (or set of adap-

tations, depending on whether one is a lumper or

splitter), our visual system is very well engineered,

capable of recovering far more sophisticated infor-

mation from two-dimensional light arrays than the

best of the artiﬁcially engineered systems developed

so far. The third is that successful vision requires spe-

cialized neural circuits or computational machinery

designed particularly for solving the adaptive prob-

lem of scene analysis (Marr, 1982). And the fourth

is that scene analysis is an unsolvable computational

problem unless the design features of this special-

ized machinery "assume" that objects and events in

the world manifest many speciﬁc regularities (Shep-

ard, 1981, 1984, 1987a; Marr, 1982; Poggio et

al., 1985). These four lessons-complexity of the

adaptive information-processing problem, well en-

gineered problem-solving machinery as the evolved

solution, specialization of the problem-solving ma-

chinery to ﬁt the particular nature of the problem,

and the requirement that the machinery embody "in-

nate knowledge" about the problem-relevant parts of

the world-recur throughout the study of the com-

putational equipment that constitutes human psy-

chology (Cosmides & Tooby, 1987, 1992; Tooby &

Cosmides, 1989a, 1990b;on language, see Chomsky,

1975; Pinker, 1989;onvision,seeMarr, 1982; Poggio

et al., 1985).

These discoveries of superlative "engineering" in

the visual system have been paralleled in the study

of the other sense organs, which are simply the

recognizable transducing ends of an intricate mass

of psychological adaptations that consist of increas-

ingly more complex and integrative layers of spe-

cialized neural processing. For a variety of reasons,

the information-processing adaptations involved in

perception have been the only psychological mech-

anisms that have been studied for decades and in

depth from a functionalist perspective. Contribut-

ing factors include the fact that their functionality

is obvious to all sensate humans and their scientiﬁc

study was rescued from the metaphysical doubt that

hangs over other psychological phenomena because

their associated physical transducing structures pro-

vided a "materialist" place to begin. Arguably the

most important factor, however, was that these were

the only, mechanisms for which psychologists had

any good standards of what counted as biologically

successful problem-solving. Unacquainted with evo-

lutionary biology, few psychologists know that there

are standards for successful problem-solving in other

realms as well, such as social behavior. Unless one

knows what counts as a biologically successful out-

come, one simply cannot recognize or investigate

complex functional design or assess the extent to

which a design is well engineered.

Consequently, at present it is difﬁcult to assess

how well psychological adaptations measure up

against the intricacy and functionality of other adap-

tations. We can only judge on the basis of the re-

stricted set that have already been studied exten-

sively from a functionalist perspective-the perceptual

mechanisms. Of course, because the paradigmatic

example of a well-engineered adaptation, used for

over 130 years in biology, is a psychological adap-

tation, we know that in at least some cases our

evolved information-processing machinery incorpo-

rates complex functional design of the highest order.

Indeed, when the eye does appear in debates over

Darwinism, it is usually used by anti-Oarwinians,

who insist that the eye is far too perfect a mech-

anism to have been constructed by natural selec-

tion. In general, whenever information-processing

mechanisms have been studied from an evolution-

ary functional perspective-human vision and audi-

tion, echolocation in bats, dead-reckoning in desert

ants, and so on- the results have indicated that

the brain/mind contains psychological adaptations

at least as intricately functional as anything to be

found in the rest of the body.

One could perhaps argue that perceptual mecha-

nisms are exceptional cases because they are evolu-

tionarily older than those psychological adaptations

that are distinctively human-speciﬁc, and so have

had more time to be reﬁned. There are many rea-

sons to suspect this is not the case. But, even if

it were, it would only suggest that purely human-

speciﬁc adaptive problems, such as extensive tool

use or extensive reciprocation, would have problem-

solving adaptations less exquisite than vision, hear-

ing, maternal care, threat-perception, motivational

arbitration, mate selection, foraging, emotional com-

munication, and many other problems that have

been with us for tens of millions of years. It would

not mean that we have no adaptations to human-

speciﬁc problems at all. One reason the case of lan-

guage is so illuminating is that it speaks to exactly

this issue of the potential for complex functional-

ity in human- speciﬁc adaptations. The language

faculty is the only human information-processing

system outside of perception that has been stud-

ied extensively with clear standards of what counts

as functional performance, and the facts of psy-

cholinguistics weigh in heavily against the hypoth-

esis that human-speciﬁc adaptations have had insuf-

ﬁcient time to evolve the same highly elaborated, in-

tricately interdependent functionality characteristic

of perceptual mechanisms (Pinker & Bloom, this vol-

ume). The language faculty has the same hallmarks

of overwhelmingly functional complex design that

the visual system does, and yet we know it is a recent

and human-speciﬁc adaptation that evolved after the

hominids split off from the (rest of the) great apes

(Pinker & Bloom, this volume). The claim that lan-

guage competence is a simple and poorly engineered

adaptation cannot be taken seriously, given the to-

tal amount of time, engineering, and genius that has

gone into the still unsuccessful effort to produce ar-

tiﬁcial systems that can remotely approach-let alone

equal-human speech perception, comprehension, ac-

quisition, and production.

Finally, behavioral scientists should be aware that

functional organic machines look very different from

the kinds of systems human engineers produce using

planning and foresight. Human engineers can start

with a clean drawing board, designing systems from

scratch to perform a function cleanly, using materials

selected particularly for the task at hand. Evolving

lineages are more like the proverbial ship that is al-

ways at sea. The ship can never go into dry dock

for a major overhaul; whatever improvements are

made must be implemented plank by plank, so that

the ship does not sink during its modiﬁcation. In

evolution, successive designs are always constructed

out of modiﬁcations of whatever preexisting struc-

tures are there-structures linked (at least in the short

run) through complex developmental couplings. Yet

these short-run limitations do not prevent the emer-

gence of superlatively organized psychological and

physiological adaptations that exhibit functionality

of the highest known order- higher, in fact, than hu-

man engineers have been able to contrive in most

cases. This is because the evolutionary process con-

tinues to operate over large numbers of individuals

and over enormous stretches of time, with selection

relentlessly hill-climbing. To anthropomorphize, se-

lection achieves its results through "tinkering," sav-

ing large numbers of frequently small and indepen-

dent improvements cumulatively over vast expanses

of time (Jacob, 1977). Thus, chains of successive

modiﬁcations may be very large indeed to arrive at

an increasingly sophisticated "solution" or problem-

solving mechanism. The fact that alternative modiﬁ-

cations are randomly generated-and that selection at

anyone time is limited to choosing among this ﬁnite

set of actual alternatives-means that the evolution-

ary process might by chance "overlook" or "walk by"

a speciﬁc solution that would have been obvious to

a human engineer, simply because the correct muta-

tions did not happen to occur. The fact that evolution

is not a process that works by "intelligence" cuts both

ways, however. Precisely because modiﬁcations are

randomly generated, adaptive design solutions are

not precluded by the ﬁnite intelligence of any engi-

neer. Consequently, evolution can contrive subtle so-

lutions that only a superhuman, omniscient engineer

could have intentionally designed.

So, although organisms are functionally designed

machines, they look very different from the ma-

chines that humans build. For this reason, the sci-

ence of understanding living organization is very dif-

ferent from physics or chemistry, where parsimony

makes sense as a theoretical criterion. The study of

organisms is more like reverse engineering, where

one may be dealing with a large array of very differ-

ent components whose heterogeneous organization

is explained by the way in which they interact to

produce a functional outcome. Evolution, the con-

structor of living organization, has no privileged ten-

dency to build into designs principles of operation

that are simple and general. Evolution operates by

chance-which builds nothing systematic into organ-

isms-and by selection-which cumulatively adds mod-

iﬁcations, regardless of whether they add complex-

ity. Thus, psychologists are not likely to ﬁnd a few

satisfying general principles like Maxwell’s equations

that unify all psychological phenomena, but instead

a complex pluralism of mechanisms. Satisfying gen-

eral principles will instead be found at the next level

up, in the principles of evolutionary functionalism

that explain the organization of these mechanisms.

At an engineering or mechanism level, knowledge

will have to be constructed mechanism by mech-

anism, with the organization of the properties of

each mechanism made intelligible by knowing the

speciﬁc evolved function of that mechanism. Thus,

the computational mechanisms that generate mater-

nal love, grammar acquisition, mate selection, kin-

directed assistance, or reciprocation can be expected

to parallel Ramachadran ’s characterization of per-

ception as,

Essentially a "bag of tricks;" that through

millions of years of trial and error, the vi-

sual system evolved numerous short-cuts,

rules-of-thumb and heuristics which were

adopted not for their aesthetic appeal or

mathematical elegance but simply because

they worked (hence the "utilitarian" the-

ory). This is a familiar idea in biology but

for some reason it seems to have escaped

the notice of psychologists, who seem to

forget that the brain is a biological organ

just like the pancreas, the liver, or any other

specialized organ (Ramachadran, 1990, p.

24 ).

Adaptations, By-products, and Random Effects

The most fundamental analytic tool for recog-

nizing an adaptation is its deﬁnition. Stripped of

complications and qualiﬁcations, an adaptation is

( 1) a system of inherited and reliably developing

properties that recurs among members of a species

that (2) became incorporated into the species’ stan-

dard design because during the period of their in-

corporation, (3) they were coordinated with a set

of statistically recurrent structural properties outside

the adaptation (either in the environment or in the

other parts of the organism), (4) in such away that

the causal interaction of the two (in the context

of the rest of the properties of the organism) pro-

duced functional outcomes that were ultimately trib-

utary to propagation with sufﬁcient frequency (i.e., it

solved an adaptive problem for the organism). (For

a more extensive deﬁnition of the concept of adap-

tation, see Tooby & Cosmides, 1990b). Adaptations

are mechanisms or systems of properties crafted by

natural selection to solve the speciﬁc problems posed

by the regularities of the physical, chemical, devel-

opmental, ecological, demographic, social, and in-

formational environments encountered by ancestral

populations during the course of a species’ or pop-

ulation’s evolution (for other discussions of adap-

tation, see Pinker & Bloom, this volume; Symons,

1989, this volume; R. Thornhill, 1991; Tooby & Cos-

mides, 1990a; Williams, 1966, 1985; see Dawkins,

1986, for his discussion of adaptations under the

name adaptive complexity).

Thus, chance and selection, the two components

of the evolutionary process, explain different types

of design properties in organisms, and all aspects

of design must be attributed to one of these two

forces. Complex functional organization is the prod-

uct and signature of selection. Reciprocally, the

species-typical properties of organisms attributable

to chance will be no more important, organized, or

functional than can be attributed to chance. The con-

spicuously distinctive cumulative impacts of chance

and selection allow the development of rigorous

standards of evidence for recognizing and estab-

lishing the existence of adaptations and distinguish-

ing them from the non- adaptive aspects of organ-

isms caused by the nonselectionist mechanisms of

evolutionary change (Pinker & Bloom, this volume;

Symons, this volume; R. Thornhill, 1991 ; Tooby

& Cosmides, 1990b; Williams, 1966, 1985). Com-

plex adaptations are usually species-typical (Tooby

& Cosmides, 1990a); moreover, they are so well-

organized and such good engineering solutions to

adaptive problems that a chance coordination be-

tween problem and solution is effectively ruled out

as a plausible explanation. Adaptations are recog-

nizable by "evidence of special design " (Williams,

1966); that is, by recognizing certain features of

the evolved species-typical design of an organism

”as components of some special problem-solving ma-

chinery" (Williams, 1985, p. 1) that solve an evolu-

tionarily long-standing problem. Standards for rec-

ognizing special design include factors such as econ-

omy, efﬁciency, complexity, precision, specialization,

and reliability, which, like a key ﬁtting a lock, render

the design too good a solution to a deﬁned adaptive

problem to be coincidence (Williams, 1966). Like

most other methods of empirical hypothesis testing,

the demonstration that something is an adaptation

is always, at the core, a probability assessment con-

cerning how likely a situation is to have arisen by

chance. The lens, pupil, iris, optic nerve, retina, vi-

sual cortex, and so on, are too well coordinated both

with each other and with environmental factors-such

as the properties of light and the reﬂecting properties

of surfaces-to have arisen by chance.

In addition to adaptations, the evolutionary pro-

cess commonly produces two other outcomes visi-

ble in the designs of organisms: ( 1) concomitants

or by-products of adaptations (recently nicknamed

"spandrels"; Gould & Lewontin, 1979); and (2) ran-

dom effects. The design features that comprise adap-

tations became incorporated into the standard de-

sign because they promoted their own frequency and

are, there- fore, recognizable by their organized and

functional relationships to the rest of the design and

to the structure of the world. In contrast, concomi-

tants of adaptations are those properties of the phe-

notype that do not contribute to functional design

per se, but that happen to be coupled to properties

that are, and so were dragged along into the organ-

ism’s design because of selection on the design fea-

tures to which they are linked. They may appear

organized, but they are not functionally organized.

The explanation for any speciﬁc concomitant or

spandrel is, therefore, the identiﬁcation of the adap-

tation or adaptations to which it is coupled, together

with the reason why it is coupled. For example,

bones are adaptations, but the fact that they are

white is an incidental by-product. Bones were se-

lected to include calcium because it conferred hard-

ness and rigidity to the structure (and was dietar-

ily available), and it simply happens that alkaline

earth metals appear white in many compounds, in-

cluding the insoluble calcium salts that are a con-

stituent of bone. From the point of view of func-

tional design, by-products are the result of "chance,"

in the sense that nothing in the process of how they

came to be incorporated into a design other than

sheer coincidence would cause them to be coordi-

nated solutions to any adaptive problem. For this

reason, by-products are expected not to contribute

to the solution of adaptive problems more often or

more effectively than chance could explain. Finally,

of course, entropic effects of many types act to in-

troduce functional disorder into the design of organ-

isms. They are recognizable by the lack of coordi-

nation that they produce within the architecture or

between it and the environment, as well as by the

fact that they frequently vary between individuals.

Classes of entropic processes include mutation, evo-

lutionarily unprecedented environmental change, in-

dividual exposure to unusual circumstances, and de-

velopmental accidents. Of course, one can decom-

pose organ- isms into properties ( or holistic rela-

tions) according to any of an inﬁnite set of alter- na-

tive systems. But, unless one applies a categorization

system designed to capture their functional designs

or adaptations, organisms will seem to be nothing

but spandrels, chemistry, and entropy.

Recognizing Psychological Adaptations: Evolu-

tionary, Cognitive, Neural, and Behavioral Levels

of Analysis

Capturing lnvariance in Functional Organization:

Behavioral, Cognitive, and Neuroscience Descrip-

tions

If the psychological architectures of organisms are

infused with complex functional organization, this is

not always easy to see. Precisely because functional

organization may be very complex, and embedded

in an even more bewildering array of variable and

intricate by-products, it may appear to the unaided

intellect to be in distinguish- able from chaos or poor

design. Unless one knows what to look for-unless,

at the very least, one knows what counts as func-

tional -one cannot recognize complex functionality

even when one sees its operation.

Sciences prosper when researchers discover the

level of analysis appropriate for describing and inves-

tigating their particular subject: when researchers

discover the level where in variance emerges, the

level of underlying order. What is confusion, noise,

or random variation at one level resolves itself into

systematic patterns upon the discovery of the level

of analysis suited to the phenomena under study.

How, then, should the psychological architectures

of organisms be described so as to capture a level

of underlying functional order? Three different

languages for describing psychological phenomena

are commonly used: The behavioral, the cogni-

tive, and the neurobiological. Each language has

strengths and weaknesses for different scientiﬁc pur-

poses. For the purpose of discovering, analyzing,

and describing the functional organization of our

evolved psychological architecture, we propose that

the information-processing language of cognitive sci-

ence is the most useful. In the ﬁrst place, this is be-

cause the evolutionary function of the brain is the

adaptive regulation of behavior and physiology on

the basis of information derived from the body and

from the environment. Alternative design features

are selected for on the basis of how well they solve

adaptive problems-problems whose solution affects

reproduction. How an organism processes informa-

tion can have an enormous impact on its reproduc-

tion. It is, therefore, meaningful to ask what kind

of cognitive design features would have constituted

good solutions to adaptive information-processing

problems that persisted over many generations. Evo-

lutionary biology and hunter- gatherer studies sup-

ply deﬁnitions of the recurrent adaptive problems

humans faced during their evolution, and cogni-

tive psychology describes the information-processing

mechanisms that evolved to solve them. By combin-

ing insights from these two ﬁelds, the functional or-

ganization of the mind can be brought into sharp re-

lief.

Second, adaptations are usually species-typical.

Consequently, to capture evolved functional organi-

zation, one needs a language that can describe what

is invariant across individuals and generations. This

process of description is key: By choosing the wrong

descriptive categories, everything about an organism

can seem variable and transitory to the extent that

"plasticity" or "behavioral variability" can seem the

single dominant property of an organism. In con-

trast, well-chosen categories can bring out the hid-

den organization that reappears from individual to

individual and that, consequently, allows psycholog-

ical phenomena to be described both economically

and precisely.

Purely behavioral categories are seldom able to

capture meaningful species-typical uniformity: Are

humans "aggressive" or "peaceful," "pair-bonding" or

"polygynous," "rational" or "irrational?" With much

justice, Geertz, echoing Kroeber, dismissed large and

vague behavioral universals, such as marriage and

religion, as "fake" (1973, p. 101). Human phenom-

ena accurately described and categorized solely in

terms of behavioral outcomes appear endlessly vari-

able; they seem to manifest a kaleidoscopic welter

of erratic and volatile phenomena, which makes any

underlying uniformity-let alone functional design-

difﬁcult to see. Exceptions, such as reﬂexes and ﬁxed

action patterns, occur in the very few cases where

the mapping between stimulus and behavior is sim-

ple and immediate. Behavioral characterizations of

anything much more complicated rapidly become so

watered down with exceptions that, at best, one ends

up with vague portrayals employing terms such as

"capacity," "predisposition," "urge," "potential," and

so on-things too murky to be helpful either in de-

scribing adaptations or in predicting behavior.

Perhaps more important, however, is that behavior

is not a phenomenon sui generis. It is the product of

mechanisms that process information. Mechanisms

that produce behavior can be usefully studied on a

variety of different descriptive and explanatory lev-

els. Neuroscientists describe the brain on a physical

level-as the inter- action of neurons, hormones, neu-

rotransmitters, and other organic aspects. In con-

trast, cognitive psychologists study the brain as an

information-processing system-that is, as a collection

of programs that process information-without refer-

ence to the exact neurophysiological processes that

perform these tasks. A cognitive description speci-

ﬁes what kinds of information the mechanism takes

as input, what procedures it uses to transform that

information, what kinds of data structures (repre-

sentations) those procedures operate on, and what

kinds of representations or behaviors it generates as

output. The study of cognition is the study of how

humans and other animals process information.

To understand subsequent arguments clearly, it is

important to keep in mind exactly what we mean by

the cognitive or information-processing level. Like

all words, "cognitive" is used to mean many differ-

ent things. For example, some researchers use it in

a narrow sense, to refer to so-called "higher men-

tal" processes, such as reasoning, as distinct from

other psychological processes, such as "emotion" or

"motivation"; that is, to refer to a concept that cor-

responds more or less to the folk notion of reason-

ing while in a calm frame of mind. In contrast, we

are using the word cognitive in a different and more

standard sense. In this chapter, we use terms such

as cognitive and information-processing to refer to

a language or level of analysis that can be used to

precisely describe any psychological process: Rea-

soning, emotion, motivation, and motor control can

all be described in cognitive terms, whether the pro-

cesses that give rise to them are conscious or uncon-

scious, simple or complex. In cognitive science, the

term mind refers to an information-processing de-

scription of the functioning of an organism’s brain

and that is the sense in which we use it. (For a more

detailed discussion of the nature of cognitive expla-

nations, see Block, 1980; Fodor, 1981; or pylyshyn,

1984.)

For example, ethologists have traditionally stud-

ied very simple cognitive pro- grams. A newborn

herring gull, for instance, has a cognitive program

that deﬁnes a red dot on the end of a beak as salient

information from the environment, and that causes

the chick to peck at the red dot upon perceiving

it. Its mother has a cognitive program that deﬁnes

pecking at her red dot as salient information from

her environment, and that causes her to regurgitate

food into the newborn’s mouth when she perceives

its pecks. These simple programs adaptively regu-

late how herring gulls feed their young. (If there is a

ﬂaw anywhere in these programs-i.e., if the mother

or chick fails to recognize the signal or to respond

appropriately-the chick starves. If the ﬂaw has a ge-

netic basis, it will not be passed on to future genera-

tions. By such feedback, natural selection shapes the

design of cognitive programs.)

These descriptions of the herring gull’s cognitive

programs are entirely in terms of the functional rela-

tionship among different pieces of information; they

describe two simple information-processing systems.

Moreover, precise descriptions of these cognitive pro-

grams can capture the way in which information is

used to generate adaptive behavior. Of course, these

programs are embodied in the herring gull’s neuro-

biological "hardware. " Knowledge of this hardware,

however, is not necessary for understanding the pro-

grams as information-processing systems. Presum-

ably, one could build a silicon-based robot, using

chemical processes completely different from those

present in the gull’s brain, that would produce the

same behavioral output (pecking at red dot) in re-

sponse to the same informational input (seeing red

dot). The robot’s cognitive programs would main-

tain the same functional relationships among pieces

of information and would, therefore, be, in an im-

portant sense, identical to the cognitive pro- grams

of the herring gull. But the physical processes that

implement these programs in the robot would be to-

tally different.

Although all information-processing mechanisms

operate by virtue of the physical processes that im-

plement them, cognitive descriptions and physicalist

ones are not equivalent, but complementary. They

cannot be reduced to each other. For this reason,

the information-processing descriptions of cognitive

science are not merely metaphors in which brains

are compared to computers. Their status as an in-

dependent level of psychological explanation can be

established by considering the fact that the same

information-processing relationships can be embod-

ied in many different physical arrangements. The

text-editing program Wordstar, for example, can run

on machines with many different kinds of physical

architectures, but it always has the same functional

design at an information-processing level-the same

key strokes will move the cursor, delete a word, or

move a block of text. And the robot "gull" will still

peck at a red dot, even though its programs are

embodied in silicon chips rather than in neurons.

These relationships can be described independently

of their physical instantiation in any particular com-

puter or organism, and can be described with pre-

cision. Thus, an information-processing program,

whether in an organism or in a computer, is a set

of invariant relationships between informational in-

puts and "behavioral" outputs. Moreover, from the

point of view of the adaptive regulation of behavior,

it is the cognitive system of relationships that counts.

Given that the correct information-processing steps

are carried out, selection pressures on psycholog-

ical mechanisms are "blind" to the speciﬁc physi-

cal implementation of their information- processing

structure (except insofar as different physical imple-

mentations may vary in factors such as metabolic

cost). Because the primary function of the brain

is the adaptive regulation of behavior and physiol-

ogy in response to information, natural selection re-

tains neural mechanisms on the basis of their abil-

ity to create functionally organized relationships be-

tween information and behavior (e.g., the sight of

a predator activates inference procedures that cause

the organism to hide or ﬂee) or between informa-

tion and physiology ( e.g., the sight of a predator in-

creases the organism ’s heart rate in preparation for

ﬂight). The mechanism is selected to create the cor-

rect information-behavior or information-physiology

relationship and, so long as the physical implemen-

tation produces this relationship, its particular form

is free to vary according to other factors. Indeed, at

certain points in development, injury to the human

brain can sometimes be "repaired" in the sense that

different neurons re-create the same ordered rela-

tionship between information and behavior that the

damaged ones had implemented prior to the injury

(Flohr, 1988). When "rewiring" of this kind occurs,

the information-processing relationship is preserved,

not its physical instantiation.

In short, it is primarily the information-processing

structure of the human psycho- logical architecture

that has been functionally organized by natural se-

lection, and the neurophysiology has been organized

insofar as it physically realizes this cognitive organi-

zation. Because the function of the brain is informa-

tional in nature, its richly organized functional struc-

ture is only visible when its properties are described

in cognitive terms. Much of great interest can be

learned by investigating the brain in neurobiological

terms, but its adaptive dimension will remain invisi-

ble unless and until its mechanisms are described in

a language that is capable of expressing its informa-

tional functions.

For these reasons, the invariant functional organi-

zation of complex psychological adaptations is more

likely to be captured by cognitive descriptions than

by neuroscience ones.4 Just as mathematics is an

indispensable language for describing many scien-

tiﬁc phenomena, information-processing language is

a precise descriptive vehicle for capturing how com-

plex systems functionally interact with complex en-

vironments. What mathematics is for physics, cogni-

tive descriptions can be for a science of psychology

and behavior.

The use of information-processing language is not

enough, however. Alone, it is no more useful for

discovering invariances in functional organization

than any other descriptive language. Unless one

knows what counts as functional, one cannot recog-

nize complex functional design even when one sees

its operation. Is friendship functional? Is anger? Is

joining a group? Is pregnancy sickness? Unless one

knows what adaptive problems a species encoun-

tered during its evolutionary history and what would

have counted as solutions to these problems, these

questions are unanswerable. To discover invariances

in the functional organization of the human mind,

the language and methods of cognitive science must

be used in concert with principles drawn from evo-

lutionary biology.

Where Evolutionary Biology and Cognitive Psy-

chology Meet.

Conceptual systems, models, and theories function

as organs of perception: They allow new kinds of evi-

dence and new relationships to be perceived (Popper,

1972). As Einstein remarked, "it is the theory which

decides what we can observe" (Heisenberg, 1971,

p. 63). The tools of evolutionary functional analy-

sis function as an organ of perception, bringing the

blurry world of human psychological and behavioral

pheno into sharp focus and allowing one to discern

the formerly obscured level of our richly organized

species-typical functional architecture.

Theories about selection pressures operating in

ancestral environments place important constraints

on-and often deﬁne-what can count as an adaptive

function. Indeed, many theories of adaptive function

deﬁne what would count as adaptive information-

processing. Consider, for example, Hamilton’s rule,

which describes the selection pressures operating

on mechanisms that generate behaviors that have

a reproductive impact on an organism and its kin

(Hamilton, 1964). The rule deﬁnes (in part) what

counts as biologically successful outcomes in these

kinds of situations. These outcomes often cannot be

reached unless speciﬁc information is obtained and

processed by the organism. In the simplest case of

two individuals, a mechanism that produces acts of

assistance has an evolutionary advantage over alter-

native mechanisms if it reliably causes individual i to

help relative j whenever C; < r;jBj. In this equation,

Ci = cost to i of rendering an act of assistance to j,

measured in terms of foregone reproduction, Bj =

beneﬁt to j of receiving that act of assistance, mea-

sured in terms of enhanced reproduction, and r;j =

the probability that a randomly sampled gene will be

present at the same locus in the relative due to joint

inheritance from a common ancestor.

Other things being equal, the more closely psycho-

logical mechanisms reliably pro- duce behavior that

conforms to Hamilton ’s rule, the more strongly they

will be selected for. Under many ecological condi-

tions, this selection pressure deﬁnes an information-

processing problem that organisms will be selected

to evolve mechanisms to solve.

Using this description of an adaptive problem as

a starting point, one can immediately begin to de-

ﬁne the cognitive subtasks that would have to be ad-

dressed by any set of mechanisms capable of pro-

ducing behavior that conforms to this rule. What

information-processing mechanisms evolved to re-

liably identify relatives, for example? What crite-

ria and procedures do they embody-for example, do

these mechanisms deﬁne an individual as a sibling if

that individual was nursed by the same female who

nursed you? What kind of information is processed

,to estimate rij, the degree of relatedness? Under an-

cestral conditions, did siblings and cousins co-reside,

such that one might expect the evolution of mecha-

nisms that discriminate between the two? After all,

rj full/ sib = 4ri. ﬁrst cousin. What kind of mech-

anisms would be capable of estimating the magni-

tudes of the consequences of speciﬁc actions on one’s

own and on others’ reproduction? How are these

various pieces of information combined to produce

behavior that conforms to Hamilton’s rule? And so

on.

This example highlights several points. First and

most important, it shows how knowledge drawn

from evolutionary biology can be used to discover

functional organization in our psychological archi-

tecture that was previously unknown. Hamilton’s

rule is not intuitively obvious; no one would look for

psychological mechanisms that are well-designed for

producing behavior that conforms to this rule unless

they had already heard of it. After Hamilton’s rule

had been formulated, behavioral ecologists began to

discover psychological mechanisms that embodied it

in nonhuman animals (Krebs & Davies, 1984). Un-

guided empiricism is unlikely to uncover a mecha-

nism that is well-designed to solve this kind of prob-

lem.

Second, this example illustrates that one can easily

use the deﬁnition of an adaptive problem to generate

hypotheses about the design features of information-

processing mechanisms, even when these mecha-

nisms are designed to produce social behavior. It

allows one to break the adaptive problem down

into cognitive subtasks, such as kin recognition and

cost/beneﬁt estimation, in the same way that know-

ing that the adaptive function of the visual system is

scene analysis allows one to identify subtasks such

as depth perception and color constancy.

Third, the example shows how knowing the an-

cestral conditions under which a species evolved can

suggest hypotheses about design features of the cog-

nitive adaptations that solve the problem. For ex-

ample, co-residence is a reliable cue of sib-hood in

some species, but other cues would have to be picked

up and processed in a species in which siblings and

cousins co-reside.

Fourth, Hamilton ’s rule provides one with a stan-

dard of good design for this particular problem. Such

standards are an essential tool for cognitive scien-

tists because they allow them to identify whether

a hypothesized mechanism is capable of solving the

adaptive problem in question and to decide whether

that mechanism would do a better job under an-

cestral conditions than alternative designs. This al-

lows one to apply the powerful methods of learnabil-

ity analysis outside of psycholinguistics, to adaptive

problems involving social behavior (see pp. 73- 77,

on evolutionary functional analysis).

Fifth, this example illustrates how insights from

evolutionary biology can bring functional organiza-

tion into clear focus at the cognitive level, but not

at the neurobiological level. Hamilton’s rule imme-

diately suggests hypotheses about the functional or-

ganization of mechanisms described in information-

processing terms, but it tells one very little about the

neurobiology that implements these mechanisms-it

cannot be straightforwardly related to hypotheses

about brain chemistry or neuroanatomy. Once one

knows the properties of the cognitive mechanisms

that solve this adaptive problem, however, it should

be far easier to discover the structure of the neural

mechanisms that implement them.

The intellectual payoff of coupling theories of

adaptive function to the methods and descriptive

language of cognitive science is potentially enor-

mous. By homing in on the right categories-

ultimately adaptationist categories-an immensely in-

tricate, functionally organized, species-typical archi-

tecture can appear, with perhaps some additional

thin ﬁlms of frequency-dependent or population-

speciﬁc design as well (e.g., McCracken, 1971). Just

as one can now ﬂip open Gray’s Anatomy to any page

and ﬁnd an intricately detailed depiction of some

part of our evolved species-typical morphology, we

anticipate that in 50 or 100 years one will be able

to pick up an equivalent reference work for psychol-

ogy and ﬁnd in it detailed information-processing de-

scriptions of the multitude of evolved species-typical

adaptations of the human mind, including how they

are mapped onto the corresponding neuroanatomy

and how they are constructed by developmental pro-

grams.

The Impact of Recurrent Environmental and

Organismic Structure on the Design of Adapta-

tions

Organisms transact the business of propagation in

speciﬁc environments, and the persistent character-

istics of those environments determine the dangers,

opportunities,

and elements the organism has to use and to cope

with in its process of propagation.

Consequently, the structure of the environment

causes corresponding adaptive organization to ac-

cumulate in the design of the organism (Shepard,

1987a; Tooby & Cosmides, 1990b ).For example, the

design of eyes reﬂects the properties of light, ob-

jects, and surfaces; the design of milk reﬂects the di-

etary requirements of infants (and what was dietar-

ily available to mothers); the design of claws reﬂects

things such as the properties of prey animals, the

strength of predator limbs, and the task of capture

and dismemberment. This functional organization in

the organism-its set of adaptations- is designed to ex-

ploit the enduring properties of the environment in

which it evolved (termed its environment of evolu-

tionary adaptedness, or EEA) and to solve the recur-

ring problems posed by that environment. Adapta-

tions evolve so that they mesh with the recurring

structural features of the environment in such a way

that reproduction is promoted in the organism or its

kin. Like a key in a lock, adaptations and particular

features of the world ﬁt together tightly, to promote

functional ends.

Moreover, from the point of view of any speciﬁc

design feature or adaptation, the rest of the encom-

passing organism itself constitutes an enduring envi-

ronmental structure as well. New adaptations or de-

sign features will be selected for on the basis of how

i well they productively coordinate with the persis-

tent characteristics of this internal V environment, as

well as with the external environment. This is why

adaptations evolve to ﬁt together with each other

within the organism so well. Thus, the adaptive

mesh between tendon, muscle, and bone is no dif-

ferent in principle than the adaptive mesh between

foraging mechanisms and the ecological distribution

of food and cues reliably correlated with its presence

(Real, 1991 ). Obviously, therefore, adaptations may

solve -endogenous adaptive problems and may im-

prove over evolutionary time without necessarily be-

ing driven by or connected to any change in the ex-

ternal environment.

Long-term, across-generation recurrence of

conditions-external, internal, or t their interaction-is

central to the evolution of adaptations, and it is easy

to see why. Transient conditions that disappear after

a single or a few generations may lead to e some

temporary change in the frequency of designs, but

the associated selection pressures will disappear

or reverse as often as conditions do. Therefore,

it is only those conditions that recur, statistically

accumulating across many generations that lead

to the construction of complex adaptations. As

corollary, anything that is recurrently true (as a

net statistical or structural matter) across large

numbers of generations could potentially come to

be exploited by an evolving adaptation to solve a

problem or to improve performance. For this reason,

a major part of adaptationist analysis involves sifting

for these environmental or organismic regularities

or invariances. For example, mental states, such

as behavioral intentions and emotions, cannot be

directly observed. But if there is a reliable correla-

tion over evolutionary time between the movement

of human facial muscles and emotional state or

behavioral intentions, then specialized mechanisms

can evolve that infer a person’s mental state from the

movement of that person’s facial muscles (Ekman,

1973, 1984; Fridlund, in press). Indeed, evidence

drawn from cognitive neuroscience indicates that

we do have mechanisms specialized for "reading"

facial expressions of emotion (Etcoff, 1983, 1986).

To begin with, a cognitive adaptation can, through

exploiting the world’s subtle statistical structure, go

far beyond the information it is given, and recon-

struct from fragmentary cues highly accurate mod-

els of local conditions by exploiting these relation-

ships (e.g., self-propelled motion is correlated with

the presence of an animal; a sharp discontinuity in

reﬂected light intensity is correlated with the pres-

ence of an edge). This evolutionary Kantian posi-

tion has already been richly vindicated in the ﬁelds

of perception and psychophysics (see, e.g., Marr,

1982; Shepard, 1981, 1984, 1987a, this volume),

where the representations that our evolved compu-

tational systems construct go far beyond what is "log-

ically" warranted solely by the sensory information

itself, usually settling on single preferred interpre-

tations. Our minds can do this reliably and validly

because this fragmentary information is operated on

by evolved procedures that were selected precisely

because they reﬂect the subtle relationships endur-

ingly present in the world (e.g., shading cues that are

correlated with shape and depth, time-location rela-

tionships that are correlated with the most probable

kinematic trajectories followed by natural objects).

These mechanisms supply a privileged organization

to the available sense data so that the interaction of

the two generates interpretations that usually corre-

spond to actual conditions in the external world. In

the absence of specialized mechanisms that assume

and rely on certain relationships being characteristic

of the world, recovering accurate models of the ex-

ternal world from sense data would be an insoluble

computational problem (Marr, 1982; Poggio et al.,

1985).

Parallel ideas form the centerpiece of Chomskyan

psycholinguistics: Children must be equipped with

specialized mechanisms ("mental organs") that are

function- ally organized to exploit certain grammat-

ical universals of human language. Other- wise,

language learning would be an unsolvable com-

putational problem for the child (Chomsky 1957,

1959, 1975, 1980; Pinker 1979, 1982, 1984, 1989;

Wexler & Culi- cover, 1980). The discovery and ex-

ploratory description of such universal subtle rela-

tionships present in the "world" of human language

is a primary activity of modern linguists and psy-

cholinguists. Proposed mechanisms for language

learning that do not include specialized procedures

that exploit these relationships have been repeatedly

shown to be inadequate (e.g., Pinker 1989, 1991;

Pinker & Prince, 1988). As in perception, adapta-

tions for grammar acquisition must mesh with the

enduring structure of the world. But in this case,

the recurrent structure to be meshed with is created

by the species-typical design of other (adult) human

minds, which produce grammars that manifest cer-

tain relationships and not others.

Due to common evolutionary ancestry, the liv-

ing world of plants and animals is structured into

species and other more inclusive units that share

large sets of properties in common: Wolves resem-

ble other wolves, mammals other mammals, and so

on. Living things occur in so-called natural kinds.

This is another enduring set of relation- ships in the

world that our minds evolved to exploit. Ethnobiolo-

gists and cognitive anthropologists such as Atran and

Berlin have shown that the principles humans spon-

taneously use in categorizing plants and animals re-

ﬂect certain aspects of this enduring structure, and

are the same cross-culturally as well (Atran, 1990;

Berlin, Breedlove, & Raven, 1973).

In the last decade, the ﬁeld of cognitive develop-

ment has been revolutionized by the discovery that

the principles of inference that infants and children

bring the tasks of learning are organized to reﬂect

the particular recurrent structure of speciﬁc problem

domains, such as object construal and motion, the

differences between artifacts and living kinds, phys-

ical causality, and so on (see, e.g., Carey & Gelman,

1991 ). These evolved, domain-speciﬁc cognitive

specializations have been shown to be specialized ac-

cording to topic and to develop in the absence of ex-

plicit instruction.

For example, contrary to the Piagetian notion that

infants must "learn " the object concept, recent re-

search has shown that (at least) as early as 10 weeks-

an age at which the visual system has only just

matured-infants already have a sensorily-integrated

concept of objects as entities that are continuous in

space and time, solid (two objects cannot occupy

the same place at the same time), rigid, bounded,

cohesive, and move as a unit ( e.g., Spelke, 1988,

1990, 1991 ). Indeed, when infants of this age are

shown trick displays that violate any of these as-

sumptions, they indicate surprise-one could almost

say in such cases that the object concept embodied

in their evolved mechanisms causes them to "disbe-

lieve" the evidence of their senses (Leslie, 1988).

By 27 weeks, infants already analyze the motion

of inanimate objects into sub movements and use

this parsing to distinguish causal from noncausal re-

lationships (Leslie, 1988; Leslie & Keeble, 1987).

Needless to say, these are all relationships that accu-

rately reﬂect the evolutionarily long-enduring struc-

ture of the world.

A. Brown ( 1990) has shown that early causal

principles such as "no action at a distance" guide

learning about tool use in children as young as 18

months; these children categorize tools for use ac-

cording to functional properties (e.g., has a hooked

end for pulling) over nonfunctional properties (e.g.,

color). In contrast, the same children have great dif-

ﬁculty learning how to use a tool when its mecha-

nism of action appears to violate one of their con-

cepts about physical causality-concepts that mirror

certain aspects of Newtonian mechanics.

Very young children also make sharp distinc-

tions between the animate and inanimate worlds.

Throughout our evolutionary history, being an ani-

mal has been reliably-if imperfectly-correlated with

self-generated motion, whereas inanimate objects

rarely move unless acted upon by an outside force.

Recent research suggests that young children use this

cue to distinguish the animate from the inanimate

worlds, and make very different inferences about the

two (Gelman, 1990b; Premack, 1990). More gener-

ally, experiments by Keil ( 1989) and others indicate

that the kind of infer- ences children spontaneously

make about "natural kinds," such as animals, plants,

and substances, differ sharply from those they are

willing to make about human-made artifacts. Natu-

ral kinds are viewed as having invisible "essences"

that bear a causal relation to their perceptual at-

tributes, whereas artifacts are deﬁned by how their

perceptual attributes subserve their (intended) func-

tion. In an important series of experiments, Gelman

and Markman ( 1986, 1987; Markman, 1989) found

that natural kinds were a powerful organizer of in-

ference in young children. In general, being a mem-

ber of a natural kind carries more inferential weight

than being perceptually similar. In addition, children

give more weight to natural kind membership when

reasoning about traits that actually are more likely to

vary as a function of membership in a natural kind,

such as breathing, than when reasoning about traits

that are more to vary as a function of perceptual sim-

ilarity, such as weight or visibility at night (for sum-

mary, see Markman, 1989).

These principles apply far beyond these few sim-

ple cases. The world is full of long enduring struc-

ture, and the mind appears to be full of correspond-

ing mechanisms that use these structural features

to solve a diverse array of adaptive problems: ge-

ometrical and physical relationships that shape the

probability of various trajectories (Shepard, 1984

), biomechanically possible and impossible motions

(Shiffrar & Freyd, 1990), momentum effects on tra-

jectories (Freyd, 1987), correlations between the in-

gestion of plant toxins and teratogenesis (Profet,

1988, this volume), privileged relationships between

the gravitational ﬁeld and the orientation of ob-

jects in the world (Triesman, 1977), and on and on.

It is only for expository convenience that we have

mostly focused on mechanisms bearing on catego-

rization and inference ("knowledge"), rather than on

motivation, emotion, and decision making ("value").

The structure of the world is reﬂected in the nature

of behavior-regulating systems as well because the

long-term statistical structure of the world system-

atically creates relationships between choices and

adaptive consequences. (For a discussion of how

emotional adaptations reﬂect the relationship be-

tween decisions and the detailed structure of an-

cestral conditions, see Tooby & Cosmides, 1990b.)

Mind/world relationships extend all the way from

the ease with which people acquire fears of spiders

and snakes (Marks, 1987; Seligman, 1971 ), to the

more subtle impact that aesthetic factors have on

habitat choice and wayﬁnding (Kaplan, this volume;

Orians & Heerwagen, this volume), to the relative

unwillingness of adults to have sex with people with

whom they co-resided for long periods during child-

hood (McCabe, 1983; Parker & Parker, 1986; Past-

ner, 1986; Shepher, 1983; Westermarck, 1891; Wolf,

1966, 1968; Wolf& Huang, 1980; N. W. Thornhill,

1991 ), to the intensity with which parents and chil-

dren may come to love each other (Bowlby, 1969), to

the often violent passions humans exhibit when they

discover the existence of spousal inﬁdelity (Daly &

Wilson, 1988; Wilson & Daly,

this volume). For those who study psycho-

logical adaptations, the long-enduring structure of

the world provides a deeply illuminating source of

knowledge about the evolved architecture of the

mind. As Shepard has so eloquently put it, there has

been the evolution of a mesh between the principles

of the mind and the regularities of the world, such

that our minds reﬂect many properties of the world

(Shepard, 1987a). Many statistical and structural

relationships that endured across human evolution

were "detected" by natural selection, which designed

corresponding computational machinery that is spe-

cialized to use these regularities to generate knowl-

edge and decisions that would have been adaptive

in the EEA. Because the enduring structure of ances-

tral environments caused the design of psychologi-

cal adaptations, the careful empirical investigation

of the structure of environments, from a perspective

that focuses on adaptive problems and outcomes,

can provide powerful guidance in the exploration

of the mind. The long-term structure of the ances-

tral world is worth knowing, worth studying, and

worth relating to psychology. This realization vastly

widens the scope of information that can be brought

to bear on questions in psychology: Evolutionary

biology, paleoanthropology, hunter-gatherer studies,

behavioral ecology, botany, medicine, nutrition, and

many other ﬁelds can be mined for information that

suggests speciﬁc hypotheses, guides one toward pro-

ductive experimentation, and informs one about the

broad array of functionally specialized mechanisms

that are likely to be present. The stuff of the mind is

the stuff of the world, and so the investigation of the

rich structure of the world provides a clearly observ-

able and empirically tractable-if not royal-road into

the hidden countries of the mind.

THE CENTRAL ELEMENTS OF EVOLUTIONARY

FUNCTIONAL ANALYSIS

Approaching the coordination between the struc-

ture of the ancestral world and the design features of

adaptations with an engineering sensibility is what

gives empirical speciﬁcity and inferential power to

evolutionary functional analysis. The following are

ﬁve structured components that can be ﬁt together

in such an analysis.

I. An adaptive target: a description of what counts

as a biologically successful out- come in a given situ-

ation. Out of the inﬁnite set of potential behavioral

sequences, which small subset would count as a so-

lution to the adaptive problem? Here, one wants to

know which behavioral outcomes will have the prop-

erty of enhancing the propagation of the psychologi-

cal designs that gave rise to them. For example, out

of all the substances in the world, which should the

organism eat and which should it avoid? With whom

should the organism mate? How much parental care

should it devote to each offspring? When should the

organism join a coalition? What inferences should

be drawn on the basis of the retinal display about

the location of various surfaces? In deﬁning an adap-

tive target, the goal is to ascertain whether the pro-

posed behavioral outcome, in combination with all

the other activities and outcomes produced by the

organism, will enhance design propagation under

ancestral conditions.

2. Background conditions: a description of the re-

current structure of the ancestral world that is rel-

evant to the adaptive problem. One wants to know

what features of the ancestral world were sufﬁciently

stable to support the evolution of a design that could

produce an adaptive targe1. This could be apart of

the external environment, another part of the stan-

dard design of the organism, or a combination of the

two. This includes the information available to solve

the problem, the environmental and endogenous ob-

stacles to solving the problem, and so on. So, for ex-

ample, the regular spatial orientation of human eyes

with respect to each other, the face, and the ground

constitute background conditions for the evolution

of face recognition mechanisms in infants. Often,

but not always, the ancestral world will be similar

to the modern world ( e.g., the properties of light

and the laws of optics have not changed). However,

one needs to know something about hunter-gatherer

studies and paleoanthropology to know when an-

cestral conditions germane to the adaptive problem

diverge from modern conditions. Of course, when

there is a difference between the two, ancestral con-

ditions are the applicable ones for the purpose of

analyzing the functional design of an adaptation be-

cause they are the cause of that design. Modern envi-

ronments are relevant to the analysis of the ontogeny

of mechanisms and their calibration. It is important

to keep in mind that a mechanism that was capable

of producing an adaptive target under ancestral con-

ditions may not be capable of doing so under modern

ones. Our visual system fails to maintain color con-

stancy under sodium vapor lamps in modern park-

ing lots (Shepard, this volume), and attempting to

understand color constancy mechanisms under such

unnatural illumination would have been a major im-

pediment to progress.

I 3. A design: a description of the articulated or-

ganization of recurrent features in the organism that

together comprise the adaptation or suspected adap-

tation. A design description of the eye, for example,

would include a speciﬁcation of its species-typical

parts and the manner in which they interact to pro-

duce an adaptive target.

The design-or even the existence-of a proposed

information-processing mechanism is frequently un-

known. Indeed, an appropriate functional descrip-

tion of a design is often what one is trying to dis-

cover. When this is the case, this step in an evolution-

ary functional analysis would be the construction of

a hypothesis about the existence and design features

of a psychological adaptation. This might include

what environmental cues the mechanism monitors,

what information it draws from other mechanisms,

how it categorizes and represents this information,

what procedures or decision rules transform the in-

formational input, what kinds of representations or

behaviors it produces as output, which mechanisms

use its output for further processing, how its output

is used by other mechanisms to generate behavior,

and so on. The more causally explicit one can make

the design description at the cognitive level, the bet-

ter. Eventually, one hopes to have a description of the

neurobiological implementation of the adaptation as

well.

4. A performance examination: a description of

what happens when the proposed adaptation mech-

anistically interacts with the world. What range of

outcomes does the design actually produce? Like

putting a new aircraft prototype in a wind tunnel,

what one is looking for is a good causal or "engineer-

ing" analysis of how the proposed design actually

performs under conditions that are representative of

situations our ancestors routinely faced, and how it

performs under present conditions as well. For a pro-

posed language acquisition device, for example, one

wants to model how its information- processing pro-

cedures perform when they encounter normal lin-

guistic environments, in order to see whether the in-

teraction of procedures and environment assembles

an increasingly elaborated computational system ca-

pable of producing intelligible and grammatical sen-

tences. Similarly, one wants to model how psycho-

logical mechanisms in women or men interact with

their social and informational environments to pro-

duce mating preferences. We want to emphasize that

we are looking here for a mechanistic or causal de-

scription of how the system generates output given

input. statements like, "the human child learns its

culture through imitation and generalization " are

not models of how input generates output. They are

too unspeciﬁed to qualify as hypotheses or explana-

tions; we should have ceased treating them as such

a long time ago.

5. A performance evaluation: a description or

analysis of how well (or how poorly) the design, un-

der circumstances paralleling ancestral conditions,

managed to produce the adaptive target (the set of

biologically successful outcomes). The better the

mechanism performs, the more likely it is that one

has identiﬁed an adaptation.

It is just as important, however, to see whether

the proposed mechanism produces the behaviors one

actually observes from real organisms under mo-

dem conditions. If it does, this suggests that the

researcher is converging on a correct description of

the design of the mechanisms involved, whether they

are producing behavior that is currently adaptive or

not. The Westermarck incest avoidance mechanism,

for example, passes both tests. It produces adaptive

outcomes under ancestral (and many modem) con-

ditions (e.g., distaste for sex between siblings who

co-resided as children), and it also explains the non-

adaptive outcomes that are observed under certain

modem conditions ( e.g., distaste for sex between

kibbutz creche mates who co-resided as children

[Shepher, 1983]; distaste for sex with spouses who

were adopted into one’s family at a young age and

with whom one was raised [Wolf & Huang, 1980)).

In short, an evolutionary functional analysis con-

sists of asking a series of engineering questions:

Would the proposed design have interacted with

properties of the ancestral world to produce target

adaptive outcomes? Does the proposed design inter-

act with properties of the modern world to produce

outcomes that one actually observes in real organ-

isms, whether these outcomes are adaptive or not? Is

there an alternative design that is better able to gen-

erate adaptive targets under ancestral conditions? If

so, then are there any background conditions that

one has overlooked that would have prevented the

alternative design from evolving? And so on.

Natural selection is the process that shapes biolog-

ical form to match function, and this link between

form and function has been a critically illuminat-

ing relationship in thousands of applications. Ever

since Harvey’s question about why there were valves

in the veins led him to discover the circulation of

the blood, functional questions about organismic de-

sign have been a powerful engine for the discovery of

new knowledge (Mayr, 1983). Those even distantly

connected to organismic biology have become aware

of the spectacular functionalist revolution that has

transformed the ﬁeld over the last 30 years, placing

adaptationism on anew and far more rigorous foun-

dation (Hamilton, 1964; Maynard Smith, 1982; see,

especially, Williams, 1966). The reason why Lewon-

tin and Gould’s accusation (famous among social sci-

entists) that adaptationism consists of post hoc sto-

rytelling has so resoundingly failed to impress prac-

ticing evolutionary biologists is that they saw on a

daily basis that adaptationism was anything but post

hoc (Gould & Lewontin, 1979; for discussion, see

Pinker & Bloom, this volume). Simply put, an expla-

nation for a fact by a theory cannot be post hoc if

the fact was unknown until after it was predicted by

the theory and if the reason the fact is known at all

is because of the theory. Functionalist analysis in bi-

ology has motivated thousands of predictions about

new and critical phenomena, whose subsequent dis-

covery conﬁrmed the productivity of the emerging

paradigm. Lewontin and Gould’s critique has pri-

marily impressed those outside of evolutionary and

organismic biology who have not been exposed on a

professional basis to the ﬂood of new ﬁndings that

were both generated and economically organized by

the newly emerging functionalist principles.

When they are linked together, the ﬁve compo-

nents outlined above not only pro- vide a frame-

work for the explanation of facts that are already

known; they also form a powerful heuristic system

for the generation of new knowledge. Depending

on which questions you need answered and what in-

formation you already have, you can put these re-

lationships to a number of richly productive alter-

native uses. For example, if you are trying to dis-

cover the structure of unknown psychological mech-

anisms, you ﬁrst need to integrate steps land 2 to-

gether into a deﬁnition of an adaptive problem (what

Marr called a computational theory or task analysis;

Marr, 1982). You need to determine things such as

what information was routinely available in the en-

vironment and in the organism to solve the problem

(step 2), and what outcomes constituted a success-

ful solution to the problem (step 1 ). From this,

you can begin to develop hypotheses about the na-

ture of the information-processing mechanisms that

might have evolved to solve the problem, and then

empirically test for their presence. (For a discussion

of this approach, see Marr, 1982, and Cosmides &

Tooby, 1987. For some applications of this approach

to speciﬁc psychological problems: on vision, see

Marr, 1982; on mechanisms specialized for reason-

ing about social exchange, see Cosmides, 1989, Cos-

mides & Tooby, 1989, this volume, and Gigerenzer

& Hug, in press; on mechanisms regulating parental

solicitude, see Mann, this volume, and Daly & Wil-

son, 1988). In short, by using steps 1 and 2, one can

create a hypothesis about function that leads to the

discovery of form. This use of the elements of evolu-

tionary functional analysis guides the researcher step

by step from a deﬁnition of an adaptive problem to

the discovery and mapping of the mechanisms that

solve it.

An alternative starting point is step 3: a well-

speciﬁed candidate hypothesis about the structure of

an information-processing mechanism. So, for ex-

ample, you might hypothesize that operant condi-

tioning explains the acquisition of natural language

grammars. To proceed with an evolutionary func-

tional analysis, you would then need to develop a

description of the relevant environmental features

(step 2) and deﬁne what counts as a successful out-

come (step 1 ). You would then proceed to steps

4 and 5-performance examination and evaluation.

If your hypothesis about design is correct, then the

step 4 performance examination will reveal that the

design’s interaction with the relevant environment

features is at least capable of producing a success-

ful outcome. The performance evaluation of step 5

will allow you to determine whether the design hy-

pothesized in step 3 is better at producing adaptive

outcomes than alter- native designs.

We will refer to the application of steps 4 and 5

as the solvability criterion: To be correct, a cognitive

adaptation must be capable of solving the proposed

problem or generating behavior that we know hu-

mans routinely perform and of doing so given the

relevant background conditions. Although this may

seem like an obvious step, theories in psychology

are rarely evaluated in this way, which has allowed

entire research communities to labor under the im-

pression that, say, associationism or imitation con-

stitute effective explanations of the phenomena they

studied. Such tests of computational performance-or

learnability analyses as they are called when applied

to learning tasks-were pioneered in psycholinguistics

by Pinker and colleagues ( 1979, 1984, 1989, 1991;

Pinker & Prince, 1988; Wexler & Culicover, 1980) in

order to evaluate which theories of language acqui-

sition could actually account for the fact that chil-

dren learn the language of their local community. By

using this method one can, in fact, rule out entire

classes of theories as inadequate, without having to

empirically test each one of an inexhaustible set of

trivial variants. Because there are an inﬁnite num-

ber of alternative theories, empirical falsiﬁcation is

not by itself a practical research strategy; it must be

combined with other sources of valid inference if one

is to be able to draw larger and more interesting con-

clusions. For psychologists, the analysis of computa-

tional performance is one way of doing this.

Yet another approach to evolutionary functional

analysis begins with noting the existence of a com-

plexly articulated and recurrent phenotypic pattern-

for example, eyes, teeth, pregnancy sickness, or

sexualjealousy-and investigating whether it might be

the expression of an adaptation (Williams, 1966, p.

10). In such cases, one is following the logic in yet

another direction: Given a known phenotypic struc-

ture (step 3), one dissects the environment (step 2)

and the requirements for reproduction (step 1 ), to

ﬁnd out whether they compose a well-deﬁned adap-

tive problem for which the reliable outcomes of the

design (step 4) constitute a well-engineered solution

(step 5). Profet’s proposal that pregnancy sickness

constitutes an adaptation to limit maternal ingestion

ofteratogens during the most vulnerable phases of

embryogenesis is an excel- lent application of this ap-

proach (Profet, this volume). It should be stressed

that this is the only type of functionalist analysis

to which Gould and Lewontin’s accusation of post

hoc storytelling could possibly apply, even in prin-

ciple, since it is the only one that works backward

from known facts about phenotypic design. Yet, even

here, the critique could only apply if all facts about

the environment, the other parts of the organism,

and the structure believed to be an adaptation were

known in advance. In practice, this is never the

case. This form-to-function approach is just as pro-

ductive as the others because it leads to the predic-

tion and organization of previously unknown facts,

usually about additional design features of the or-

ganism as well as about the recurrent structure of

the world. For example, the study of the visual sys-

tem has proﬁted immensely from the fact that sci-

entists knew that the eye exists and that the visual

system’s function is to perform scene analysis given

data transduced by the eye. Indeed, the functional-

ist approach to the study of vision has generated one

of the most sophisticated and least ad hoc bodies of

knowledge in psychology. As Mayr put it, summariz-

ing the historical record in response to accusations

that adaptationist research was simply post hoc sto-

rytelling, "The adaptationist question, ’What is the

function of a given structure or organ?’ has been for

centuries the basis for every advance in physiology"

( 1983, p. 32). Adaptationist principles can provide

equally powerful guidance for research in psychol-

ogy as well.

Even if every aspect of a mechanism were already

known, examining the detailed transactions between

selected features of the environment and selected

parts of the mechanism would clarify many features

of its functional organization, such as which aspects

of the design perform the work ( e.g., which aspects

of pregnancy sickness cause the mother to avoid in-

gesting teratogens) and which are functionless or

even harmful side effects (such as calorie reduction

during the ﬁrst trimester). Naturally, the form- to-

function approach does include the risk of answering

the post hoc "why" question that Gould and Lewon-

tin so disdain; that is, of explaining why already

known features of biological designs came to be as

they are. But even physics and geology run the "risk"

of addressing such Kiplingesque post hoc questions

as why Mercury has an orbit that deviates from the

predictions of Newtonian mechanics, why Asia has

the Himalayas, or why the universe has its present

set of four interactions, temporal asymmetry, back-

ground radiation, and particle distribution. In sci-

ence, this is usually called "explanation."

TOWARD A POST-STANDARD MODEL VIEW OF

DEVELOPMENT

Development from an Adaptationist Perspec-

tive

The recognition that organisms are integrated col-

lections of problem-solving mechanisms organized

to propagate their designs brings with it an adap-

tationist framing of development. An adaptation is,

by its nature, an improbably good organization of

elements and so will not often spontaneously come

into existence merely by chance. Instead, for adap-

tations to exist, they must be speciﬁcally constructed

from the materials present in evolutionarily normal

environments. Accordingly, the developmental pro-

grams and machinery responsible for assembling an

adaptation correctly are also adaptations. As adapta-

tions, they themselves have complex structures that

assume and require recurrent features of the world,

and that interact with this recurrent structure to pro-

duce biologically functional targeted outcomes.

Hence, the primary function or target of develop-

mental adaptations (which include the genes) is to

reconstruct in offspring the evolved functional or-

ganization that was present in their parents, which

is predominantly species-typical design. The genes

and the mechanisms of genetic transmission are, of

course, adaptations to the problem of faithfully repli-

cating into the offspring critical information neces-

sary to reconstruct this design. The genes come

embedded in a matrix of cellular and developmen-

tal machinery constituting an additional set of adap-

tations that use the genetic structure as regulatory

elements to institute and to guide the bryogenesis

and subsequent development along species-standard

pathways. For this reason, it is useful to think of

the genes together with the developmental machin-

ery as one integrated suite of adaptations-the devel-

opmental programs-and to distinguish the minor id-

iosyncratic features of an individual’s genes and zy-

gotic machinery from the recurrent or species-typical

dimensions that have endured long enough to have

been organized by natural selection. The latter spec-

ify the species-standard physiological and psycholog-

ical architecture visible in all humans raised in nor-

mal environments, whereas the former specify the

usually minor perturbations within that architecture

(Tooby & Cosmides, 1990a).

Why do we so often connect complex adapta-

tions or evolved architectures with concepts such as

species-typical, human universal, species-standard,

recurrent, and so on? This is because when humans

are described from the point of view of their complex

adaptations, differences tend to disappear, and a uni-

versal architecture stands out in stark relief. This is

both empirically the case (nearly everyone has two

eyes, two hands, the same sets of organs, and so

on) and theoretically expected to be the case if or-

ganisms are primarily collections of complex adap-

tations. The logic is straightforward (Tooby & Cos-

mides, 1990a; see also Tooby, 1982):

1. A species is a group of organisms with a com-

mon history of interbreeding and a continuing ability

to interbreed to form offspring who can typically re-

produce at least as well as their parents.

2. To survive and reproduce in a complex world,

organisms need complex problem-solving machinery

(complex adaptations).

3. Complex adaptations are intricate machines

that require complex "blueprints" at the genetic level.

This means that they require coordinated gene ex-

pression, involving hundreds or thousands of genes

to regulate their development.

4. Sexual reproduction automatically breaks apart

existing sets of genes and randomly generates in the

offspring new, never before existing combinations of

genes at those loci that vary from individual to indi-

vidual.

5. If genes differed from individual to individual

in ways that signiﬁcantly impacted the developed de-

sign of the component parts of complex adaptations,

then existing genetic combinations whose developed

expressions had ﬁt together into complex adapta-

tions would be pulled apart by sexual recombina-

tion. Equally, new combinations would be thrown

randomly together, resulting in phenotypes whose

parts were functionally incompatible. This is because

parts in any complex machine are functionally inter-

dependent: If you tried to build anew car engine out

of a mixture of parts from a Ronda and a Toyota, the

parts would not ﬁt together. To build anew engine

whose component parts ﬁt together, you would have

to salvage parts from two "parents" that were of the

same make and model.

6. Because sexual recombination is a random pro-

cess, it is improbable that all of the genes necessary

for a complex adaptation would be together in the

same individual if the genes coding for the compo-

nents of complex adaptations varied substantially be-

tween individuals.

7. Therefore, it follows that humans, and other

complex, long lived, outbreeding organisms, must be

very nearly uniform in those genes that underlie our

complex adaptations.

8. By the same token, sexually reproducing pop-

ulations of organisms freely tolerate genetic varia-

tion to the extent that this variation does not impact

the complex adaptive organization shared across in-

dividuals. To return to our car engine example, the

color of the parts is functionally irrelevant to the op-

eration of the car and so can vary arbitrarily and su-

perﬁcially among cars of the same make and model;

but the shapes of the parts are critical to functional

performance and so cannot vary if the "offspring" de-

sign is to function successfully.

These constraints on variation apply with equal

force to psychological adaptations: Even relatively

simple cognitive programs, "mental organs," or neu-

rological structures must contain a large number of

interdependent processing steps, limiting the nature

of the variation that can exist without violating the

functional integrity of the psychological adaptation.

The psychic unity of humankind-that is, a universal

and uniform human nature-is necessarily imposed

to the extent and along those dimensions that our

psychologies are collections of complex adaptations.

Therefore, it is selection interacting with sexual re-

combination that tends to impose near uniformity at

the functional level in complex adaptive designs (as

well as in whatever is develop- mentally coupled to

complex functional structure). It is selection that is

responsible for what we have been calling our uni-

versal evolved psychological and physiological archi-

tecture.

There is no small irony in the fact that Stan-

dard Social Science Model hostility to adaptation-

ist approaches is often justiﬁed through the accu-

sation that adaptationist approaches purportedly at-

tribute important differences between individuals,

races, and classes to genetic differences. In actual-

ity, adaptationist approaches offer the explanation

for why the psychic unity of humankind is genuine

and not just an ideological ﬁction; for why it applies

in a privileged way to the most signiﬁcant, global,

functional, and complexly organized dimensions of

our architecture; and for why the differences among

humans that are caused by the genetic variability

that geneticists have found are so overwhelmingly

peripheralized into architecturally minor and func-

tionally superﬁcial properties. If the antiadaptation-

ists were correct (e.g., Gould & Lewontin, 1979) and

our evolved architectures were not predominantly

sets of complex adaptations or properties develop-

mentally coupled to them, then selection would not

act to impose cross-individual uniformity, and indi-

viduals would be free to vary in important ways and

to any degree from other humans due to genetic dif-

ferences. If the world were, in fact, governed by

nonselectionist forces, then the psychic unity of hu-

mankind would simply be a ﬁction.

Modern geneticists, through innovative molecular

genetic techniques, have certainly discovered within

humans and other species large reservoirs of genetic

variability(Hubby & Lewontin, 1966; Lewontin &

Hubby, 1966; see reviews in Ayala, 1976, and Nevo,

1978). But it is only an adaptationist analysis that

predicts and explains why the impact of this variabil-

ity is so often limited in its scope to micro-level bio-

chemical variation, instead of introducing substan-

tial individuating design differences. The study of

the operation of selection on complex mechanisms

makes it difﬁcult to see how more than a tiny frac-

tion of this variation could be constitutive of complex

psychological or physiological adaptations.

Thus, human design resolves itself into two pri-

mary tiers: First, an encompassing functional su-

perstructure of virtually universal, complexly artic-

ulated, adaptively organized developmental, physio-

logical, and psychological mechanisms, resting on a

universally shared genetic basis; and, second, low

level biochemical variation creating usually slight

individuating perturbations in this universal design

due to the existence of a reservoir of genetic vari-

ability in the species. There may also be some thin

ﬁlms of population-speciﬁc or frequency-dependent

adaptive variation on this intricate universal struc-

ture (see, e.g., Durham, 1991; McCracken, 1971 ),

but for a number of reasons these will be very small

in magnitude next to the complex structure of a uni-

versal human nature (for discussion, see Tooby &

Cosmides, 1990a, 1990b). The primary function of

developmental adaptations is to reconstruct in each

new individual this complex, functional architecture,

and the primary focus of adaptationists is the study

of this universal structure.

The fact that humans in ordinary environments re-

liably develop a clearly recognizable species-typical

architecture should in no way be taken to imply that

any developed feature of any human is immutable

or impervious to modiﬁcation or elimination by sufﬁ-

ciently ingenious ontogenetic intervention. Nothing

about humans could possibly be immune from devel-

opmental intervention, simply because we are physi-

cal systems open to contact and manipulation by the

rest of the world; we are not some- thing made un-

alterable by inexorable supernatural predestination.

People frightened of the myth that biology is destiny

can be reassured (just as others may be alarmed) by

the fact that there are no limits to what could be

done, especially by evolutionarily novel measures:

Deliver the right quanta to the right ribosomes or

other locations at the right times and anyone or any-

thing could be successively modiﬁed into a water-

melon or an elephant. In contrast, Standard So-

cial Science Model advocates, such as Gould, tend to

equate evolved biological design with immutability

without any logical or empirical warrant. As Gould

expresses his rather magical belief, "If we are pro-

grammed to be what we are, then these traits are

ineluctable. We may, at best, channel them, but we

cannot change them either by will, education, or cul-

ture" (Gould, 1977c, p. 238).

In actuality, the very openness of development to

intervention poses a critical set of adaptive problems

for developmental adaptations. Their primary func-

tion is to successfully reconstruct each functionally

necessary detail of our species-typical architecture,

including the tens or hundreds of thousands of spe-

ciﬁc components and arrangements that endow us

with a lens, a retina, an optic nerve, language, ma-

ternal attachment, emotions, retinotopic maps, ten

ﬁngers, a skeleton, color constancy, lungs, a rep-

resentational system embodying the implicit theory

that others have minds, an ability to cooperate, spa-

tial cognition, and so on. Each of these adaptations

constitutes a very narrow target of improbably good

functional organization. Because the world is full

of potential disruptions, there is the perennial threat

that the developmental process may be perturbed

away from the narrow targets that deﬁne mechanis-

tic workability, producing some different and non-

functional outcome. Develop- mental adaptations

are, therefore, intensely selected to evolve machin-

ery that defends the developmental process against

disruption (Waddington, 1962). Profet (this vol-

ume) provides an elegant analysis of a psychologi-

cal adaptation designed to defend against just such

threats to adaptive development, protecting embryo-

genesis from the potentially disruptive plant tox-

ins in the mother’s diet through modifying her di-

etary decisions during pregnancy. More generally

developmental programs are often designed to re-

spond to environmentally or genetically introduced

disorder through feedback-driven compensation that

redirects development back toward the successful

construction of adaptations. Thus, developmental

processes have been selected to defend themselves

against the ordinary kinds of environmental and ge-

netic variability that were characteristic of the envi-

ronment of evolutionary adaptedness, although not,

of course, against evolutionarily novel or unusual

manipulations.

Of course, unlike human-built machines that have

astatic architecture until they break down, organ-

isms are systematically transformed by developmen-

tal adaptations over their life histories from zygote

to senescence. Thus, the task facing developmen-

tal adaptations is not to assemble a machine of ﬁxed

design, but rather to assemble and modify the set

of expressed adaptations according to a moving tar-

get of age, sex, and circumstance-dependent design

speciﬁcations. For example, adaptive problems are

often speciﬁc to a particular life stage, and so the or-

ganism must be developmentally timed to have the

necessary adaptations for that stage, regardless of

whether, as aside effect, they happen to appear be-

fore or persist after they are needed ( e.g., the pla-

centa, fetal hemoglobin, the sucking reﬂex, the abil-

ity to digest milk, the fear of strangers, ovulation, the

ability to be sexually aroused, milk production, and

so on).

Hence, the Standard Model assumption-critical

to its logic-that the mental organization present in

adults but absent from newborns must be "acquired"

from the social world has no conceptual foundation

and is, in many cases, known to be empirically false.

In the worldview of the SSSM, biological construc-

tion goes on in the uterus, but at birth the child

is "biologically complete" except for growth; at this

point, it is surrendered into the sole custody of social

forces, which do the remainder of the construction of

the individual. This, of course, reﬂects folk biology,

captured in the two dictionary deﬁnitions of innate

as "present from birth" and as "intrinsic." Social con-

structivist arguments frequently take the form that

because thus-and-such is absent at birth, or doesn?t

appear until after age seven, or until after puberty,

it is obviously "learned" or "socially constructed." As

a result, a common, but generally irrelevant feature

of "nativist" versus "environmentalist" debates is over

what is "present from birth." This confuses (among

other things) the question of whether something is

expressed at the time of birth with whether there ex-

ists in the individual evolved developmental mecha-

nisms that may activate and organize the expression

of an adaptation at some point in the life cycle. De-

velopmental processes continue to bring additional

adaptations on line (as well as remove them) at least

until adulthood, and there is an increasing amount of

evidence to suggest that age-driven adaptive changes

in psycho- logical architecture continue throughout

adulthood (see, e.g., Daly & Wilson, 1988). Thus,

just as teeth and breasts are absent at birth and

develop later in an individual’s life history, percep-

tual organization, domain-speciﬁc reasoning mecha-

nisms, the language acquisition device, motivational

organization, and many other intricate psychologi-

cal adaptations mature and are elaborated in age-

speciﬁc fashions that are not simply the product

of the accumulation of ?experience." Consequently,

psychological adaptations may be developmentally

timed to appear, disappear, or change operation to

mesh with the changing demands of different age-

speciﬁc tasks, such as parenting, emotional decoding

of the mother’s voice, language acquisition, species-

appropriate song learning, and so on (Daly & Wilson,

1988; Femald, this volume; Marler, 1991; Newport,

1990).

Equally, although most human psychological and

physiological adaptations appear to be sexually

monomorphic, some are obviously sexually differ-

entiated to address those adaptive problems whose

task demands were recurrently disparate for females

and males over evolutionary time ( e.g., Buss, 1987,

1989, 1991, this volume; Daly & Wilson, 1988; Ellis,

this volume; Silverman & Eals, this volume; Symons,

1979; Wilson & Daly, this volume ). For any partic-

ular gender difference, many psychologists are in-

terested in whether it was caused ( 1) by sexu-

ally monomorphic psychologies encountering differ-

ential treatment by the social world, or (2) by sex-

ually differentiated developmental mechanisms en-

countering treatment from the social world, whether

that treatment was uniform or differential. As inter-

esting as this question may be, however, the fact that

an expressed gender difference may ﬁrst appear af-

ter birth, or even late in life, is evidence neither for

nor against either of these views.

For these reasons, one needs to distinguish an or-

ganism’s evolved design or species- typical architec-

ture from its "initial state" (Carey, 1985a); that is, its

state at whatever point in development one chooses

to deﬁne as "initial" (birth, conception, fetus prior to

gonadal or neural sexual differentiation, puberty, or

whatever). Not all features of evolved human de-

sign are or can be present at anyone time in anyone

individual. Thus, the genetically universal may be

developmentally expressed as different maturational

designs in the infant, the child, the adolescent, and

the adult; in females and males; or in individuals

who encounter different circumstances. Pregnancy

sickness is arguably a feature of our evolved uni-

versal design, but it does not appear in males, chil-

dren, or women who have never become pregnant;

it is only present in sexually mature women while

they are pregnant. Thus, when we use terms such

as "evolved design," "evolved architecture," or even

"species-typical," "species-standard," "universal," and

"panhuman," we are not making claims about every

human phenotype all or even some of the time; in-

stead, we are referring to the existence of evolution-

arily organized developmental adaptations, whether

they are activated or latent. Adaptations are not nec-

essarily expressed in every individual. They only

need to have been expressed often enough in our

evolutionary history to have been targets of selec-

tion, and, hence, to have been organized by se-

lection so that they reliably develop under appro-

priate circumstances. For this reason, adaptations

and adaptive architecture can be discussed and de-

scribed at (at least) two levels: ( 1) the level of

reliably achieved and expressed organization (as,

for example, in the realized structure of the eye),

and (2) at the level of the developmental programs

that construct such organization. To avoid cumber-

some expressions, we do not usually bother to ter-

minologically distinguish successfully assembled ex-

pressed adaptive architecture from the more fun-

damental developmental adaptations that construct

them. Context usually makes obvious which is being

discussed.

Selection Regulates How Environments Shape

Organisms

Many social and biological scientists have labored

under the false impression that only certain things

are under the "control, " or "inﬂuence," or "deter-

mination " of the genes or of biology. According

to this view, evolutionary approaches are only ap-

plicable to those traits under such "genetic control,

" and the greater the environmental inﬂuence or

control, the smaller the domain of things for which

evolutionary analyses properly apply (e.g., Sahlins,

1976a; Gould, 1977a, 1977b, 1977c; note, espe-

cially, Gould’s Standard Model contrast of "genetic

control" with the "purely cultural"). In this du-

alistic conception, the genes are "biological" and

evolved, while "the environment"- including the so-

cial environment-is nonbiological and nonevolved.

In consequence, the environment is held to be some-

thing that not only can attenuate, nullify, or even re-

verse "genetic forces" but may break the causal chain

entirely, liberating human affairs from the causal

patterning of evolution. For proponents of the SSSM,

it is self- evident that the causal forces of evolu-

tion and "biology" are located solely inside the or-

ganism and are expressed in an unadulterated form

only at birth, if then. In contrast, the causal forces

of the environment are seen as external to the or-

ganism, as having their own independent causal his-

tory, and as having no particular reason to act on

the organism in such away as to preserve or elab-

orate the organism’s initial biological organization.

In short, the environment is conceptualized as ob-

viously nonbiological in character. Development is

consequently portrayed as a process in which the

new- born organism-usually seen as a passive clay-

like object with some initial biologically given form-

is pounded or sculpted by the active and nonbiologi-

cal environment according to its accidents, structure,

or agenda. It follows from this view that biology can

only express itself in human life if it is unalterable

or at least rigid enough to resist the pounding forces

of the environment-a bombardment that begins at

birth. One might think of the stubbornly biological

aspects of human life as the hardened part of the

clay, while the more plastic parts are easily shaped

by the environment and quickly lose their initial bio-

logical form. Consequently, even if advocates of the

SSSM do not want to dichotomize traits cleanly into

two sets (e.g., hardened versus wet clay), they could

array them by this criterion as more or less biolog-

ically determined; that is, as more or less environ-

mentally inﬂuenced.

Despite its tenacity in the social sciences at large,

this Standard Model view of development has been

abandoned by many cognitive scientists and by biol-

ogists because it rests on a series of fallacies and mis-

conceptions. To begin with, despite the routine use

of such dualistic concepts and terms by large num-

bers of researchers throughout the social and bio-

logical sciences, there is nothing in the real world

that actually corresponds to such concepts as "ge-

netic determination " or "environmental determina-

tion." There is nothing in the logic of development

to justify the idea that traits can be divided into ge-

netically versus environmentally controlled sets or

arrayed along a spectrum that reﬂects the relative

inﬂuence of genes versus environment. And, most

critically, the image of "the environment" as a "non-

biological" causal inﬂuence that diminishes the "ini-

tial" evolved organization of humans rests on the fail-

ure to appreciate the role that the evolutionary pro-

cess plays in organizing the relationship between our

species-universal genetic endowment, our evolved

developmental processes, and the recurring features

of developmental environments.

In the ﬁrst place, every feature of every pheno-

type is fully and equally codetermined by the interac-

tion of the organism’s genes (embedded in its initial

package of zygotic cellular machinery) and its onto-

genetic environments-meaning everything else that

impinges on it. By changing either the genes or the

environment any outcome can be changed, so the in-

teraction of the two is always part of every complete

explanation of any human phenomenon. As with all

interactions, the product simply can- not be sensibly

analyzed into separate genetically determined and

environmentally determined components or degrees

of inﬂuence. For this reason, everything, from the

most delicate nuance of Richard Strauss’s last per-

formance of Beethoven’s Fifth Sym. phony to the

presence of calcium salts in his bones at birth, is to-

tally and to exactly that same extent genetically and

environmentally codetermined. "Biology" cannot be

segregated off into some traits and not others.

Nevertheless, one could understand and acknowl-

edge that all human phenomen2 are generated by

gene-environment interactions, yet believe that the

existence and participation of the environment in

such interactions insulates human phenomena from

interesting evolutionary patterning. After all, if only

our genes evolved, whereas the form of the environ-

ment is generated by other processes (such as ge-

ology, cultural transmission, epidemiology, and me-

teorology) then the gene-environment interaction

seems to blunt the organizing effects evolution might

otherwise have on human life. Although this view

seems quite reasonable, a close examination of how

natural selection actually adaptively organizes gene-

environment interactions over time leads to a very

different conclusion, which might be summed up by

the counterintuitive claim that "the environment" is

just as much the product of evolution as are the

genes.

To understand why this is so, one needs to dis-

tinguish "the environment" in the sense of the real

total state of the entire universe-which, of course,

is not caused by the genes or the developmental

mechanisms of any individual-from "the environ-

ment" in the sense of those particular aspects of the

world that are rendered develop- mentally relevant

by the evolved design of an organism ’s developmen-

tal adaptations. It is this developmentally relevant

environment-the environment as interacted with by

the organism-that, in a meaningful sense, can be said

to be the product of evolution, evolving in tandem

with the organism ’s organized response to it. The

confusion of these two quite distinct senses of "envi-

ronment" has obscured the fact that the recurrent or-

ganization of the environment contributes a biolog-

ical inheritance parallel to that of the genes, which

acts co-equally with them to evolutionarily organize

the organism throughout its life.

The assumption that only the genes are evolved

reﬂects a widespread misconception about the way

natural selection acts. Genes are the so-called

units of selection, which are inherited, selected, or

eliminated, and so they are indeed something that

evolves. But every time one gene is selected over

another, one design for a develop- mental program

is selected over another as well; by virtue of its

structure, this developmental program interacts with

some aspects of the environment rather than others,

rendering certain environmental features causally

relevant to development. So, step by step, as natu-

ral selection constructs the species’ gene set (chosen

from the available mutations), it constructs in tan-

dem the species’ developmentally relevant environ-

ment (selected from the set of all properties of the

world). Thus, both the genes and the developmen-

tally relevant environment are the product of evolu-

tion.

Even more crucially, by selecting one developmen-

tal program over another, the evolutionary process

is also selecting the mechanisms that determine how

the organ- ism will respond to environmental input,

including environmental input that varies. A devel-

opmental mechanism, by virtue of its physical de-

sign, embodies a speciﬁcation for how each possible

state of the developmental environment is to be re-

sponded to, if encountered. This is a central but little

understood point: There is nothing "in " the environ-

ment that by itself organizes or explains the devel-

opment, psychology, morphology, or behavior of any

organism. "The" environment affects different organ-

isms in different ways. We ﬁnd the smell of dung re-

pellent; dung ﬂies are attracted to it. Temperature

at incubation determines the sex of an alligator, but

not of a human (Bull, 1983). A honeybee larva that

is fed Royal Jelly will become a queen bee rather

than a sterile worker, but Royal Jelly will not have

this effect on a human baby. Many bats navigate by

sound echoes that humans cannot even hear. Rats

have an elaborate sense of smell, which their food

choice mechanisms use, but their navigation mecha-

nisms ignore smell cues entirely in favor of geomet-

ric cues (Gallistel, 1990). Indeed, this last exam-

ple shows that the developmentally relevant environ-

ment is not just organism-speciﬁc, it is mechanism-

speciﬁc. In other words, the actual relationship be-

tween environmental conditions and developmental

outcomes is created by the design of the develop-

mental procedures that exist in the organism and,

within the limit of the physically possible, mecha-

nisms could be designed into the system to create a

causal relationship between any imaginable environ-

mental input and any imaginable output. In princi-

ple, genetic engineers could build honeybee larvae

that develop into workers if, and only if, they are ex-

posed to recitations of Allen Ginsberg’s "Howl."

Aside from physical necessity, then, it is the

evolved design of the organism that decides what or-

ganized consequences the environment can have on

it. The rules that govern how environments impact

the developing organism have themselves evolved

and have been shaped by selection. Consequently,

the evolutionary process deter- mines how the en-

vironment shapes the organism. Over evolutionary

time, genetic variation in developmental programs

(with selective retention of advantageous variants),

explores sampled properties out of the total environ-

ment potentially available to be interacted with. This

process discovers which recurrent features are useful

in the task of organizing and calibrating psycholog-

ical adaptations and which recur- rent features are

unreliable or disruptive. It renders the latter irrele-

vant to development.

A natural response is to claim that although the

genes are highly stable, replicated with few muta-

tions from generation to generation, the environ-

ment is volatile, rendering any developmental pro-

cess coordinating the two ineffectual. Once again,

how- ever, our intuitions are not a privileged per-

spective from which one can declare the world to be

either stable or variable. Whether the world is "sta-

ble" or "variable" depends on the categorization sys-

tem used or, to put it another way, on which parts of

the world are selected to be processed by a mecha-

nism.

Consider, for example, the following thought ex-

periment. Imagine that an identical pool shot is

set up every generation on a rather odd pool table.

Three of the four cushions wobble continuously and

unpredictably, but one happens to be stable. The

"genes" determine the exact direction the cue ball

is hit each time, while the "environment" (i.e., the

angle of the cushions when struck) determines how

the shot will be reﬂected back. Whether a particu-

lar shot successfully sinks the target ball in a pocket

(i.e., whether it achieves the adaptive target) is de-

termined by the interaction of the direction of the

shot and the orientation of the cushion at the time

the ball hits it (i.e., the interaction of genes and envi-

ronment determine the outcome). Assume also that

there is variation in the direction of the shot (i.e., in

the "genes") and that successful shots cause genes to

be retained.

Over the long run, feedback-driven selection will

come to determine which direction the ball is hit.

In determining this direction, it will also end up se-

lecting the stable cushion for the bank shot, and not

the wildly oscillating ones. It will end up directing

the shot at exactly that spot along the stable cushion

from which the shots are stably successful.

Similarly, selection will design developmental

adaptations that respond to those aspects of the

world that have a relatively stable recurrent struc-

ture, such that the mesh between the two will re-

liably produce design-propagating outcomes. Just

as selection has acted on genetic systems to keep

mutations to tolerable levels, selection has acted to

"choose" the more stable parts of the environment

to render developmentally relevant, such that these

aspects of the environment stably mesh with devel-

opmental pro- grams to produce reliably developing

adaptive architectures.

The Standard Model framing says that the world

pre-exists and is not caused by the organism, so

that the world’s effect on the organism will have no

particular tendency to organize the developing or-

ganism according to any evolved or adaptive pat-

tern. Equally, the pool-table cushion pre-existed each

shot and was not created by them, and the laws

of physics determined how each shot would be re-

ﬂected back. So, in a static SSSM analysis, it is self-

evident that the outcome is the mixture of two fac-

tors, one "biological" and one "nonbiological," with

the nonbiological diluting, obliterating, or even re-

versing the biological. In contrast, an evolutionary

analysis points out that the shot, through its careful

targeting, picked out the particular cushion hit and

the exact location hit. Over time, the selective reten-

tion of successful shots will organize the effect that

the pre-existing environment had on the trajectories

of the shots and the outcome of the game. The pre-

existing structure of the world was exploited to im-

pose an organization on the outcome that it would

not otherwise have had.

In this same fashion, the evolutionary process ex-

plores and sifts the environment for aspects that

will usefully organize the developing organism. The

evolutionary process puts to work sources of or-

ganization and information anywhere they are un-

earthed, whether in the genes or in the environ-

ment, in a mother’s smile or in a companion’s ex-

pression of surprise. Selection has crafted the de-

sign of the develop- mental programs so that or-

ganisms tap into these reservoirs of information or

hook themselves to environmental forces that help

to construct them. Thus, the genomes of organisms

have evolved to "store" organization and informa-

tion that is necessary or helpful for development in

the structure of the world itself. For example, for

a developing child, the information in the minds of

other humans, properly used, is a very useful source

of information to use in their development, as are

the linguistic patterns encountered in the local lan-

guage community and patterns in local social behav-

ior. Natural selection has intricately orchestrated de-

velopmental mechanisms so that things in the de-

velopmentally relevant world have been assigned an

appropriate causal role, from gravity, plants, and

three-dimensionality, to language, mothers, and so-

cial groups. Evolution shapes the relationship be-

tween the genes and the environment such that they

both participate in a coordinated way in the con-

struction and calibration of adaptations. Thus, evo-

lutionarily patterned structure is coming in from the

environment, just as much as it is coming out from

the genes.

Accordingly, "biology" is not some substance that

is segregated or localized inside the initial state of

the organism at birth, circumscribing the domain to

which evolutionary analyses apply. It is also in the

organization of the developmentally relevant world

itself, when viewed from the perspective imposed by

the evolved developmental mechanisms of the or-

ganism. Thus, nothing the organism interacts with

in the world is nonbiological to it, and so for hu-

mans cultural forces are biological, social forces are

biological, physical forces are biological, and so on.

The social and cultural are not alternatives to the bi-

ological. They are aspects of evolved human biology

and, hence, they are the kinds of things to which

evolutionary analysis can properly be applied.

Social scientists need to recognize that humans

have evolved to expect, rely on, and take advan-

tage of the richly structured participation of the

environment-including the human social and cul-

tural environment-in the task of adaptive develop-

ment. Our developmental and psychological pro-

grams evolved to invite the social and cultural

worlds in, but only the parts that tended, on balance,

to have adaptively useful effects. Programs gov-

erning psychological development impose concep-

tual frame- works on the cultural and social worlds;

choose which parts of the environment are moni-

tored; choose how observations and interactions are

categorized, represented, and interrelated; decide

what entities to pursue interactions with; and, most

importantly, determine what algorithms or relation-

ships will organize environmental input into devel-

opmental change or psychological output. Conse-

quently, the study of developmental adaptations is

a central branch of evolutionary psychology. Under-

standing these adaptations will make visible the sub-

tly stable structure of the developmentally relevant

world and illuminate the evolutionary patterning in

how human beings respond to smiles, to language,

and to the cultural knowledge in others’ minds. Each

human, by expressing his or her species-typical ar-

chitecture, contributes to the environmental regular-

ities that others inhabit and rely on for their devel-

opment.

For these reasons, it is a complete misconception

to think that an adaptationist perspective denies or

in the least minimizes the role of the environment

in human development, psychology, behavior, or so-

cial life. Environmentalists have been completely

correct about the importance of environmental in-

put in the explanation of human behavior. Humans

more richly and complexly engage the variable fea-

tures of the environment than any other species we

know of. It is this perception that has maintained en-

vironmentalism as the predominant viewpoint in the

social sciences, despite its crippling inadequacies as

an analytic framework. The terms "culture," "social-

ization," "intelligence," and "learning" are labels for

poorly understood families of processes that reﬂect

this complex and overwhelming human engagement

with environ- mental inputs. Any viable theory of

the evolved architecture of humans must reﬂect

l this reality and must be environmentalist in

this sense. As discussed, the incoherence of Stan-

dard Model environmentalism stems from ( I) the

widespread failure to recognize that environmental

responsiveness requires a complex evolved design

(expressible as either a set of developmental adap-

tations or as a reliably developing psychological ar-

chitecture; (2) the refusal to investigate or specify

the nature of this architecture or these programs;

and (3) the failure to recognize that the regulatory

structure of these programs speciﬁes the relationship

between environmental input and behavioral, devel-

opmental, or psychological output.

f For social scientists, of course, this recognition

requires a radical change in practice: Every "en-

vironmentalist" explanation about the inﬂuence of

a given part of the environment on humans will-

if it is to be considered coherent-need to be ac-

companied by a speciﬁc "nativist" hypothesis about

the evolved developmental and psychological mech-

anisms that forge the relationship between the en-

vironmental input and the hypothesized psychologi-

cal output. All "environmentalist" theories necessar-

ily depend upon and invoke "nativist" theories, ren-

dering environmentalism and nativism interdepen-

dent doctrines, rather than opposed ones. For post-

Standard Model researchers, these incoherent tradi-

tional dichotomies (genetic/environmental, biologi-

cal/social, nativist/environmental) are being aban-

doned, as is reﬂected, for example, in the title of

a recent article, "Learning by instinct" (1. Gould &

Marler, 1987).

The Impact of the Recurrent Structure of Hu-

man Life and Human Culture on the Design of

Psychological Adaptations

The evolved mesh between the information-

processing design of human psychological adapta-

tions, their developmentally relevant environments,

and the stably recurring structure of humans and

their environments is pivotal to understanding how

an evolutionary psychological approach to culture

differs from that of the Standard Social Science

Model. For traditional anthropologists, cultures vary

from place to place, and there is nothing privi-

leged about a conceptual framework that catego-

rizes human thought and action so as to capture

underlying patterns of cross-cultural uniformity, as

against the inﬁnite class of perspectives by which hu-

man thought and behavior appear everywhere dif-

ferent (Geertz, 1973, 1983, 1984; see D. E. Brown,

1991, for a critique of this view). Nevertheless,

from the "point of view" of natural selection, such

uniformities-however subtle and unimportant to pro-

fessionally neutral anthropological minds-are indeed

privileged, and for a very simple reason. However

varig’Qle cultures and habitats may have been dur-

ing human evolution, selection would have sifted hu-

man social and cultural life (as well as everything

else) for obvious or subtle statistical and structural

regularities, building psychological adaptations that

exploited some subset of these regularities to solve

adaptive problems. (As we will discuss, one of the

problems that had to be solved using regularities was

the problem of learning "culture" itself.)

Thus, Geertz’s starting point, that humans have

evolved to use culture, is obviously true (although

not in the slavish sense he envisions). But the next

step in his logic- that humans don’t have general cul-

tures, only particular ones, and so evolved to real-

ize themselves only through cultural particularity-is

the error of naive realism. No instance of anything

is intrinsically (much less exclusively) either "gen-

eral" or "particular"-these are simply different lev-

els at which any system of categorization encoun-

ters the same world. When you meet Roger Shep-

ard you are, at one and the same time, meeting

both a particular (and distinctive) individual and

a manifestation of humanity in general, embody-

ing innumerable species-typical characteristics. So

it is with cultures. Selection operated across ances-

tral hominid populations according to what were,

in effect, systems of categorization, screening cross-

cultural variability for any recurrent relationships

that were relevant to the solution of adaptive prob-

lems. To be thoroughly metaphorical, natural se-

lection scrutinized the structure of human cultural

and social environments, searching for regularities

that could be used to engineer into our evolved ar-

chitecture effective techniques for adaptive problem-

solving. Thus, the issue is: During the Pleistocene,

were there any statistical and structural uniformi-

ties to human life from culture to culture and habitat

to habitat, from any perspective-no matter how sub-

tle or abstract or unobservable-that could have been

used by species-typical problem-solving machinery

for the adaptive regulation of behavior and physiol-

ogy? Geertz sees (modern) cultures as irredeemably

particularized, conﬁdently dismissing talk of mean-

ingful human universals as nearly vacuous. Did nat-

ural selection "see" the human world the same way?

The answer is obvious, once the question is asked.

Anthropological orthodoxy to the contrary, human

life is full of structure that recurs from culture to

culture, just as the rest of the world is. (Or, if

one prefers, there are innumerable frames of refer-

ence within which meaningful cross-cultural unifor-

mities appear, and many of these statistical unifor-

mities and structural regularities could potentially

have been used to solve adaptive problems.) Exactly

which regularities are, in fact, part of the develop-

mentally relevant environment that is used by our

universal architectures is a matter to be empirically

determined on a mechanism-by-mechanism, case-

by-case basis. Such statistical and structural reg-

ularities concerning humans and human social life

are an immensely and indeﬁnitely large class (D. E.

Brown, 1991): adults have children; humans have

a species-typical body form; humans have character-

istic emotions; humans move through a life history

cued by observable body changes; humans come in

two sexes; they eat food and are motivated to seek it

when they lack it; humans are born and eventually

die; they are related through sexual reproduction

and through chains of descent; they turn their eyes

toward objects and events that tend to be informa-

tive about adaptively consequential issues; they of-

ten compete, contend, or ﬁght over limited social or

subsistance resources; they express fear and avoid-

ance of dangers; they preferentially associate with

mates, children, and other kin; they create and main-

tain enduring, mutually beneﬁcial individuated rela-

tionships with nonrelatives; they speak; they create

and participate in coalitions; they desire, plan, de-

ceive, love, gaze, envy, get ill, have sex, play, can be

injured, are satiated; and on and on. Our immensely

elaborate species-typical physiological and psycho-

logical architectures not only constitute regularities

in themselves but they impose within and across cul-

tures all kinds of regularities on human life, as do

the common features of the environments we inhabit

(see D. E. Brown, 1991, for an important exploration

of the kinds and signiﬁcance of human universals).

Human developmental mechanisms have been

born into one cultural environment or another hun-

dreds of billions of times, so the only truly long-term

cumulatively directional effects of selection on hu-

man design would have been left by the statistical

commonality that existed across cultures and habi-

tats. Consequently, the sustained impact of these

cross-culturally recurrent relationships sculpted the

problem-solving mechanisms of the human mind to

expect and exploit the common structure of human

cultures and human life; that is, natural selection

constructed adaptations specialized to mesh with the

detailed structural regularities common to our an-

cestral cultural environments. For this reason, not

only does natural selection privilege frames of ref-

erence that reveal patterns of universality in human

life but our evolved psychological architecture does

also. Embedded in the programming structure of

our minds are, in effect, sets of assumptions about

the nature of the human world we will meet during

our lives. So (speaking metaphorically) we arrive

in the world not only expecting, Geertzian fashion,

to meet some particular culture about whose specif-

ically differentiated peculiarities we can know noth-

ing in advance. We also arrive expecting to meet, at

one and the same time, and in one and the same em-

bodiment, the general human culture as well-that is,

recognizably human life manifesting a wide array of

forms and relations common across cultures during

our evolution ( or at least some set out of the super-

set). Thus, human architectures are "pre-equipped"

(that is, reliably develop) specialized mechanisms

that "know" many things about humans, social rela-

tions, emotions and facial expressions, the meaning

of situations to others, the underlying organization

of contingent social actions such as threats and ex-

changes,

t language, motivation, and so on. ! To take only

one example, humans everywhere include as part of

their standard conceptual equipment the idea that

the behavior of others is guided by invisible inter-

nal entities, such as "beliefs" and "desires"-reﬂecting

what Dennett calls "the intentional stance" ( 1987).

Of course, this way of thinking seems so natural to us

that it is difﬁcult to see that there is anything to ex-

plain: It is tempting to think that beliefs and desires

are "real" and that, therefore, humans everywhere

simply learn to see the world as it really is. Side-

stepping the complex question of whether this pan-

human folk psychology is an accurate way of captur-

ing "real" human psychology (i.e., whether it is a true

or a complete description), we simply want to point

out that things such as beliefs and desires are inher-

ently unobservable hidden variables used to explain

observations that could be explained by any of an

inﬁnite set of alternative theories (in fact, psycholo-

gists have come up with many such theories). There-

fore, a belief in beliefs and desires cannot be justiﬁed

by observations alone, so the fact that it is conven-

tional among humans to "theorize" about others in

this fashion is not inexorably mandated by their ex-

perience or otherwise required by the structure of

the external world. For the same set of nonman-

dated ideas to have emerged everywhere on earth,

our developmental programs or cognitive architec-

tures must impose this way of interpreting the world

of other humans on us.

In fact, an intensive research effort in the ﬁeld

of cognitive development has recently provided sub-

stantial support for the hypothesis that our evolved

psychological architecture includes procedures that

cause very young children to reliably develop a

belief-desire folk psychology-a so-called "theory of

mind" (e.g, Astington, Harris, & Olson, 1988; Leslie,

1987, 1988; Perner, 1991; Wellman, 1990; Wim-

mer & Perner, 1983). Developmental psychologists

have been ﬁnding that even 2- and 3-year-olds make

different inferences about "mental entities" (dreams,

thoughts, desires, beliefs) than about "physical enti-

ties." Moreover, children typically "explain" behavior

as the conﬂuence of beliefs and desires (e.g., Why

has Mary gone to the water fountain? Because she

has a desire for water (i.e., she is thirsty) and she be-

lieves that water can be found at the water fountain).

Such inferences appear to be generated by a domain-

speciﬁc cognitive system that is sometimes called a

"theory of mind" module (Leslie, 1987). This module

consists of specialized computational machinery that

allows one to represent the notion that "agents" can

have "attitudes" toward "propositions" (thus, "Mary"

can "believe" that "X," "Mary" can "think" that "X," and

so on). Between the ages of 3 and 5 this domain-

speciﬁc inferential system develops in a characteris-

tic pattern that has been replicated cross-culturally

in North America, Europe, China (Flavell, Zhang,

Zou, Dong & Qui, 1983), Japan (Gardner, Harris,

Ohmoto & Hamazaki, 1988), and a hunter-gatherer

group in Camaroon (A vis & Harris, in press). More-

over, there is now evidence suggesting that the neu-

rological basis of this system can be selectively dam-

aged; indeed, autism is suspected to be caused by a

selective neurological impairment of the "theory of

mind" module (Baron-Cohen, Leslie, & Frith, 1985;

Leslie, 1987, 1988; Leslie & Thaiss, 1990).

This research indicates that a panhuman "theory

of mind" module structures the folk psychology that

people develop. People in different cultures may

elaborate their folk psychologies in different ways,

but the computational machinery that guides the de-

velopment of their folk notions will be the same, and

some of the notions developed will be the same as

well. Humans come into the world with the ten-

dency to organize their understanding of the ac-

tions of others in terms of beliefs and desires, just

as they organize patterns in their two-dimensional

retinal array under the assumption that the world is

three-dimensional and that objects are permanent,

bounded, and solid.

Thus, not only do evolved mechanisms assume

certain things will tend to be true of human life but

these specialized procedures, representational for-

mats, cues, and categorization systems impose-out of

an inﬁnite set of potential alternatives-a detailed or-

ganization on experience that is shared by all normal

members of our species. There is certainly cultural

and individual variability in the exact forms of adult

mental organization that emerge through develop-

ment, but these are all expressions of what might be

called a single human metaculture. All humans tend

to impose on the world a common encompassing

conceptual organization, made possible by universal

mechanisms operating on the recurrent features of

human life. This is a central reality of human life

and is necessary to explain how humans can com-

municate with each other, learn the culture they are

born into, understand the meaning of others’ acts,

imitate each other, adopt the cultural practices of

others, and operate in a coordinated way with others

in the social world they inhabit. By metaculture, we

mean the system of universally recurring relation-

ships established and constituted by ( 1) our univer-

sal evolved species-typical psychological and physio-

logical architectures, (2) the interaction of these ar-

chitectures with each other in populations, (3) their

interaction with the developmentally relevant recur-

rent structure of human natural and cultural envi-

ronments, and (4) their patterned standard impact

on human phenomena.

Social scientists have traditionally considered

there to be a tension or explanatory competition

between human universals and transmitted cultural

variability: the more of one, the less of the other (0.

E. Brown, 1991 ). However, careful causal analysis

of the information-processing tasks required to learn

transmitted culture leads to what is

f very nearly the opposite conclusion. In fact, it

is only the existence of this common f metacultural

structure, which includes universal mechanisms spe-

cialized to mesh with ! the social world, that makes

the transmission of variable cultural forms possible.

To make this clear, consider the question of how

it is possible for preIinguistic children to deduce the

meanings of the words they hear when they are in

the process of learning their local language for the

ﬁrst time. The child’s task of discovering the mean-

ings of words involves isolating, out of an inﬁnite

set of possible meanings, the actual meanings in-

tended by other speakers (e.g., Carey, 1982, 1985a;

Quine, 1960). Children can infer the meanings of

messages in the local, but unknown language only

because they, like cryptographers, have a priori sta-

tistical knowledge about likely messages, given the

situational context. To solve the problem of refer-

ential ambiguity, the child’s procedures for seman-

tic analysis must depend on the fact that our uni-

versal evolved psychological architectures impose on

the world enough standard and recur- rent interpre-

tations between speaker and listener to make the

deduction of a core lexicon possible. Since the in-

fant is new to the culture and ignorant of it, these

shared interpretations cannot be supplied by the cul-

ture itself, but must be supplied by the .1 human

universal metaculture the infant or child shares with

adults by virtue of their common humanity. (In con-

trast, the Standard Model’s initially content-free gen-

eral process child mind would share no common

interpretations with local adults and could rely on

no necessary imposition of common event constru-

als by both speaker and listener.) Thus, the sys-

tem for assigning correct semantic meanings to cul-

turally arbitrary signs necessarily relies on the pres-

ence of species-typical cognitive adaptations and on

the nonarbitrariness of meaning systems that in-

habit these cognitive adaptations. These mecha-

nisms reliably identify evolutionarily recurrent situ-

ations (such as threat, play, or eating) in such away

that the participants have similar construals of the

situation and responses to it, including things likely

to be said about it.

For example, children who are just learning their

local language interpret novel words using Mark-

man’s "whole object assumption" and her "taxonomic

assumption." The whole object assumption causes

them to interpret the novel word "cup" as referring

to a whole cup, and not to its handle, the porce-

lain it is made of, a cup on saucer, a cup of

tea (and so on); the "taxonomic assumption" causes

them to interpret "cup" as referring to all objects

of the same type, and not to the particular cup be-

ing pointed to at that moment (Markman, 1989;

Markman & Hutchinson, 1984). O course, the op-

eration of these assumptions depends, in turn, on

interpretations generated by the kinds of domain-

speciﬁc inferential systems discussed earlier, which

deﬁne what entities and relations count as whole

objects, animals, plants, people, natural kinds, ar-

tifacts, taxonomic categories, and so on (Carey &

Gelman, 1991). Still other domain-speciﬁc reason-

ing procedures may privilege certain interpretations

01 social relations. Thus, social contract algorithms

have both intrinsic deﬁnitions for those terms used

by their procedures and cues for recognizing which

elements in recurrent situations correspond to those

terms (Cosmides, 1989; Cosmides & Tooby, 1989,

this volume). Consequently, these evolved reason-

ing specializations may sometimes function as nu-

clei around which semantic inference is conducted.

Emotional expression also function as metacultural

cues that assign standardized meanings to the con-

tingent elements of situations (see Fernald, this vol-

ume; Tooby & Cosmides, 1990b). Fol example, if

someone reacts with fear, others interpret this as a

reaction to danger and attempt to identify in the situ-

ation what the dangerous entity is, re-evaluating var-

ious stimuli. They may scan the local environment,

organizing their search by a categorization system

that privileges some things (e.g., snakes) over others

(e.g., ﬂowers; (Cook, Hodes, & Lang, 1986).

Thus, we have the surprising result that it is the

shared species-typical mechanism! and common

metacultural framings that make it possible for a

child to learn what is culturally variable: in this case,

the meanings of words in the local language. This ar-

gument, in fact, generalizes beyond language: The

variable features of culture can be learned solely be-

cause of the existence of an encompassing univer-

sal human metaculture. The ability to imitate the

relevant parts of others’ actions (Meltzoff, 1988),

the ability to reconstruct the representations in their

minds, the ability to interpret the conduct of others

correctly, and the ability to coordinate one’s behav-

ior with others all depend on the existence of human

metaculture. Sperber and Wilson ( 1986) have writ-

ten at length about how, for successful communica-

tion to be possible, both sender and receiver must

share a great many assumptions about the world.

The less they mutually assume, the more difﬁcult it is

to communicate until, in the limiting case, they can-

not communicate at all. The child arrives in the cul-

ture free of any knowledge about its particularities,

and so the only way the child initially can be commu-

nicated with is through what is mutually manifest

between the child and the adults by virtue of their

common humanity (e.g., Fernald, this volume). The

same is true, as Sperber ( 1982) concisely points out,

of ethnographers: The best refutation of cultural rel-

ativity is the activity of anthropologists themselves,

who could not understand or live within other hu-

man groups unless the inhabitants of those groups

shared assumptions that were, in fact, very similar

to those of the ethnographer. Like ﬁsh unaware of

the existence of water, interpretativists swim from

culture to culture interpreting through universal hu-

man metaculture. Metaculture informs their every

thought, but they have not yet noticed its existence.

So the beginning of this section, in which we dis-

cussed how natural selection sifted cultural variabil-

ity throughout the Pleistocene for uniformities, gave

only a one-sided analysis of how, despite cultural

variability, universals still existed. It is even more

important to realize that contentful human univer-

sals make possible the very existence of transmit-

ted cultural variability (what is usually called "cul-

ture"), which would otherwise be impossible. There-

fore, the development of increasing cultural varia-

tion throughout the Pleistocene was made possible

by the evolution of psycho- logical specializations

that exploited the regularities of human metaculture

in order to learn the variable features of culture. To

return to a position William James stated a century

ago, to behave ﬂexibly, humans must have more "in-

stincts" than other animals, not fewer.

THE TRANSITION TO POST -STANDARD

MODEL PSYCHOLOGY

The Decline of Standard Model Psychology

The progression from Standard Model psychol-

ogy to post-Standard Model psychology was driven

largely by the emergence of new and more rigorous

standards that psychological theories are now ex-

pected to meet. As the ﬁeld grew more sophisticated,

various communities of psychologists began insisting

on causal accounts of how hypothesized Standard

Model mechanisms produced their effects: What are

the networks of cause and effect that, step by step,

lead from input to output? In the social sciences,

no model of the human psychological architecture

seemed impossible when its proponents didn’t have

to specify by what methods it generated human be-

havior. The cognitive revolution, with its emphasis

on formal analysis, made clear that theories needed

to be made causally explicit to be meaningful, and

it supplied psychologists with a far more precise lan-

guage and set of tools for analyzing and investigating

complexly contingent, information-responsive sys-

tems. When examined from this perspective, most

traditional theories turned out to be both incomplete

and incapable of accounting for large classes of ob-

served phenomena. Indeed, most no longer seemed

to qualify as hypotheses at all. For example, "learn-

ing" ceased to be seen as an explanation for behavior,

but instead was recognized as a label for a loosely

deﬁned class of phenomena generated by as yet un-

known procedures. For modern psychologists the

key question became: What is the explicit descrip-

tion of these procedures?

Over the last three decades, the hard work of dis-

covering procedures that could actually account in

detail for observed behavior and competences has

led to the wide- spread conclusion that our evolved

psychological architecture must include a large set

of mechanisms of a very different character than

Standard Model psychologists had envisioned. The

most fundamental shift from Standard Model to

post-Standard Model psychology has been the aban-

donment of the axiom that evolved psychological

mechanisms must be largely-or exclusively-general-

purpose and free of any contentful structure not put

there by experience (e.g., Carey & Gelman, 1991;

Chomsky, 1975; Cosmides & Tooby, 1987, 1992; Gal-

listel, 1990; Gigerenzer, 1991b; Gigerenzer

&Murray, 1987; Herrnstein, 1977; Pinker, 1984;

Rozin, 1976; Rozin&Schull, 1988; Shepard, 1984,

1987a; Symons, 1987). Many psychologists have

been forced by their data to conclude that both

human and nonhuman minds contain-in addition

to whatever general-purpose machinery they may

have-a large array of mechanisms that are (to list

some of the terms most frequently used) functionally

specialized, con- tent-dependent, content-sensitive,

domain-speciﬁc, context-sensitive, special-purpose,

adaptively specialized, and so on. Mechanisms that

are functionally specialized have been called (with

some differences in exact deﬁnition) adaptive spe-

cializations by Rozin ( 1976), modules by Fodor (

1983), and cognitive competences or mental organs

by Chomsky (1975, 1980).

Consequently, the core of the debate is not re-

ally about whether the reliably developing design of

the mind evolved-the answer to that question can

only be yes. The debate is, instead, over whether

our evolved psychological architecture is predom-

inantly domain-general (Symons, 1987). Did the

human mind evolve to resemble a single general-

purpose computer with few or no intrinsic content-

dependent programs (e.g., Gould, 1979)? Or does

its evolved architecture more closely resemble an

intricate network of functionally dedicated comput-

ers, each activated by different classes of content

or problem, with some more general-purpose com-

puters embedded in the architecture as well (e.g.,

Chomsky, 1975, 1980; Cosmides & Tooby, 1987; Gal-

listel, 1990; Gazzaniga, 1985; Rozin, 1976; Symons,

1987)? In other words, does the human mind come

equipped with any procedures, representational for-

mats, or content-primitives that evolved especially

to deal with faces, mothers, language, sex, food, in-

fants, tools, siblings, friendship, and the rest of hu-

man metaculture and the world?

Solvability and the Formal Analysis of Natural

Competences.

Thirty years ago, Noam Chomsky inaugurated

anew era in the behavioral sciences when he be-

gan to explore psychological questions by analyz-

ing the capacities of well- speciﬁed computational

systems (Chomsky 1957, 1959). His approach was

distinctive. To evaluate existing psychological the-

ories, he ﬁrst made their underlying assumptions

about computational mechanisms explicit. He then

tested the ability of these computational mechanisms

to solve real, natural problems that humans were

known to be able to solve. In his ﬁrst application of

this method, he attempted to evaluate the adequacy

of behaviorist accounts of language, particularly as

presented in Skinner’s then recently published book,

Verbal Behavior ( 1957). When Chomsky examined

the behaviorist account of language in the light of

these criteria, he found that it suffered from a series

of difﬁculties that precluded it from being a persua-

sive explanation for human linguistic competence.

Chomsky’s research program brought the serious

deﬁciencies of the Standard Model into plain view

because it combined two key ingredients: (1) the

study of tasks related to a natural, complex, real-

world competence that humans were known to have,

and (2) the use of formal solvability analyses to ex-

plore the actual computational capacities of mech-

anisms hypothesized to generate explicitly deﬁned

outcomes. A theory about the design of a mecha-

nism cannot be correct if, under the relevant condi-

tions, that design cannot solve the problem or gen-

erate the performance that the theory claims it can;

this can be determined using a solvability analysis,

as outlined in pp. 73-77.

Language was a pivotal choice for a test

of domain-general accounts of behavior because

language-particularl syntax- involved complex but

clearly speciﬁable patterns of behavior that humans

were already known to be able to produce under nat-

ural conditions without elaborate experimental ma-

nipulations. Within this domain, one could precisely

and unambiguously deﬁne criteria for recognizing

what behavioral patterns humans could and did rou-

tinely produce (grammatical versus ungrammatical

sentences). Therefore, one could deﬁne what output

any mechanism hypothesized to account for these

behavioral patterns had to produce as well. In con-

trast, no one could tell whether associationist mech-

anisms or general-purpose symbol-processing mech-

anisms could account for phenomena such as "reli-

gion," "marriage," or "politics" because no one had

an unambiguous empirical deﬁnition of human per-

formance in these spheres.

By specifying what counts as the production of

grammatical utterances or the acquisition of the

grammar of a human language, psycholinguists

working within the Chomskyan research tradition

have been using solvability analyses to show that

a task routinely mastered by four-year-old children

is too richly structured to be accounted for by any

known general-purpose mechanism operating in real

time (Chomsky, 1975, 1980; Pinker, 1979, 1984,

1989, 1991; Pinker & Bloom, this volume; Wexler

& Culi- cover, 1980). Despite three decades of in-

tensive efforts by Standard Model psychologists to

get general-purpose cognitive machinery to learn

grammar, their theories have fared no better than

did their behaviorist predecessors. To take a re-

cent example, through careful solvability analyses,

Pinker and Prince were able to show that newly

proposed domain-general connectionist and associ-

ationist models were computation- ally insufﬁcient

to solve even so narrow a problem as the acquisi-

tion of the past tense in English (Pinker, 1991; Pinker

& Prince, 1988). These mechanisms failed precisely

because they lacked computational machinery spe-

cialized for the acquisition of gram- mar.

Thirty years of such ﬁndings have forced many

cognitive psychologists, against their inclination,

to accept domain-speciﬁc hypotheses about lan-

guage learning-to conclude that humans have as

part of their evolved design a language acqui-

sition device (LAD), which incorporates content-

dependent procedures that reﬂect in some form "uni-

versal grammar" (Chomsky, 1975, 1980; Pinker,

1979,1984,1989,1991; Wexler & Culicover, 1980).

In this view, the architecture of the human mind

contains content-specialized mechanisms that have

evolved to exploit the subtle cross-culturally re-

curring features of the grammars of human lan-

guage communities-one facet of human metaculture

(Pinker & Bloom, this volume).

The introduction of solvability analyses and the

increasing demand for well-speciﬁed information-

processing models have exposed the deﬁciencies of

Standard Model theories in other areas of psychol-

ogy as well (see, e.g., Carey & Gelman, 1991; Cos-

mides, 1989; Cosmides & Tooby, 1989, this volume;

Gelman & Markman, 1986, 1987; Keil, 1989; Leslie,

1987; Markman, 1989). Standard Model theories

are usually so underspeciﬁed that one cannot make

their underlying assumptions about computational

mechanisms procedurally explicit. To the extent that

they can be evaluated, however, when they are faced

with real world tasks that humans routinely solve,

they consistently perform poorly or not at all.

In fact, the large-scale theoretical claims of Stan-

dard Model psychology never had a strong empiri-

cal base. Limited empirical support could be pro-

duced for Standard Model domain-general theories,

but only so long as research was conﬁned to the

investigation of experimenter-invented, laboratory

limited, arbitrary tasks. The occasional matches be-

tween domain-general theories and data sets have

been chronically weak and experimentally fragile.

These restricted empirical successes depended on

carefully picked experimental venues, such as pi-

geons isolated from conspeciﬁcs pecking for food

in stimulus-depauperated environments or humans

learning lists of nonsense syllables. Standard Model

theories of mechanisms have maintained themselves

as empirically credible primarily through pretheoret-

ical decisions concerning what kinds of experiments

were considered meaningful and through assump-

tions imposed a priori on the class of hypotheses

that would be entertained. For humans and nonhu-

mans alike, exposure to biologically signiﬁcant stim-

uli and natural tasks elicits complexly patterned per-

formances that Standard Model theories are unable

to predict or explain. So, to keep behavioral phe-

nomena in line with theory, Standard Model psychol-

ogists had to keep humans and other species outside

of ecologically valid circumstances, away from any

biologically signiﬁcant stimuli, and test them on ar-

tiﬁcial problems that subjects would not have had

to solve in their environment of evolutionary adapt-

edness (for discussion, see Beach, 1950; Breland &

Breland, 1961; Hermstein, 1977; Lockard, 1971 ).

Although these weaknesses have now mostly been

abandoned in the study of other species, they unfor-

tunately remain endemic in many areas of human

psychology.

Once animal behavior researchers let the pigeon

out of its barren artiﬁcial cage, a rich ﬂock of be-

havioral phenomena appeared, and questions in-

evitably arose about the mechanisms that guide the

animal to do all the different things it needs to

do in natural environments to survive and repro-

duce. Thus, ethology (or behavioral ecology, sociobi-

ology, or animal behavior) played an important cor-

rective role by providing examples of the tasks or-

ganisms solve and the complex performances they

exhibit in more natural conditions (Daly & Wilson,

1984b; Krebs & Davies, 1984; Lorenz, 1965; Rozin

& Schull, 1988; Tinbergen, 1951; Wilson, 1975).

These ﬁelds carefully documented functionally inter-

pretable behaviors that lie far outside anything that

Standard Model psychology and a short list of drives

could explain. Researchers investigating the now

well-known selection pressure expressed by Hamil-

ton’s rule (see pp. 67-68) documented an enor-

mous array of kin-directed assistance in nonhuman

animals-behaviors completely undreamed of in Stan-

dard Model psychology (Hamilton, 1964; Williams

& Williams, 1957; for review, see Krebs & Davies,

1984). Infant macaques become emotionally at-

tached to immobile cloth ﬁgures even though they

nurse ("are reinforced") on another structure entirely

(Harlow, Harlow, & Suomi, 1971 ). There are re-

ports from an entire range of species-from langurs to

lions to rodents-of newly resident males killing the

unweaned infants of their predecessors, thereby ac-

celerating ovulation in their new mates (Hrdy, 1977;

Hausfater & Hrdy, 1984 ). Ring doves may expend

considerable effort to monitor the sexual behavior

of their mates (Erickson & Zenone, 1976). There

was the discovery of the complex pat- terns of food

reciprocation in vampire bats-phenomena difﬁcult

to account for using traditional notions of general-

purpose cognition, conditioning, and drive reduction

(e.g., Wilkinson 1988, 1990). From echolocation to

parental care, to celestial navigation, to courtship,

to coalitional action in chimpanzees, to seasonal mi-

gration, to decoying predators away from nests, to

communication in bees, to "friendship" and domi-

nance in baboons, nonhuman behavior is full of tasks

and organized behaviors that do not remotely ﬁt

into Standard Model psychology. This burgeoning

body of phenomena caused many animal behavior

researchers to break away from the narrow experi-

mental paradigms and narrow questions of the Stan-

dard Social Science Model.

In human psychology, the observational basis for

Standard Model theories was equally circumscribed,

but escape from its narrow experimental paradigms

has been more difﬁcult than in nonhuman psychol-

ogy. Standard Model psychologists had no salient

reason for suspecting that different psychological

mechanisms would be activated by different kinds of

tasks. Human activities appeared to be so variable-

both between cultures and among individuals within

a culture-that the notion that some tasks and prob-

lems might be more "natural" than others did not

seem conspicuously sensible. Although most psy-

chologists were faintly aware that hominids lived

for millions of years as hunter-gatherers or foragers,

they did not realize that this had theoretical impli-

cations for their work. More to the point, however,

the logic of the standard Social Science Model in-

formed them that humans were more or less blank

slates for which no task was more natural than any

other. Until the emergence of a community of Chom-

skyan psycholinguists, mainstream psychology had

been overwhelmingly dominated by general-purpose

learning and cognitive theories. In consequence, the

same processes were assumed to account for learn-

ing and action in all domains of human activity, from

suckling at the breast to incest avoidance, language

learning, and alliance negotiations among Oani war-

riors.

By questioning the assumption that all tasks were

created equal, Chomsky exposed how narrowly cho-

sen Standard Model research topics had actually

been and how over- reaching the extrapolation had

been from these topics to the rest of human thought

and action. The rise of Chomskyan psycholinguis-

tics constituted a decisive turning point in the devel-

opment of human psychology because it introduced

the subversive idea that some tasks might awaken

associated competences that were more "natural"

than others: more functionally specialized, more

complex, more reliably developing, more species-

differentiated, and, therefore, more worthy of de-

tailed exploration (Marr & Nishihara, 1978).

The Rise of Domain-Speciﬁc Psychology

The Chomskyan revolution in the study of lan-

guage slowly began to legitimize the exploration

of models of our evolved psychological architecture

that did not assume a priori that all tasks are solved

by the same set of content-independent processes.

In diverse subcommunities, the gradually expand-

ing freedom to consider domain-speciﬁc hypotheses

alongside more orthodox ones has led to their in-

creasing acceptance. Performance in virtually ev-

ery kind of experimental situation is sensitive to

the content and context of the task, and domain-

speciﬁc hypotheses tend to organize, account for,

and predict this performance better than their Stan-

dard Model predecessors. Although social and be-

havioral scientists outside of cognitive, comparative,

and physiological psychology still routinely assume a

domain-general human mind, within the community

of psychologists who rigorously study mechanisms

this view is in retreat and disarray. Standard Model

psychology has been able to persist only in those re-

search communities that avoid formal analysis en-

tirely or that avoid using it to study performance on

ecologically valid, natural tasks.

Thus, researchers who ask hard questions about

how organisms actually solve problems and who fo-

cus on the real performance of organisms on nat-

ural tasks have had to abandon the idea that the

mind is free of content-specialized machinery. Re-

searchers who study color vision, visual scene anal-

ysis, speech perception, conceptual development in

children, mental imagery, psychophysics, locomo-

tion, language acquisition, motor control, anticipa-

tory motion computation, face recognition, bio- me-

chanical motion perception, emotion recognition, so-

cial cognition, reasoning, and the perception and

representation of motion, for example, cannot ac-

count for the Psychological phenomena they study by

positing computational mechanisms that are solely

domain-general and content-independent (see, e.g.,

Bizzi, Mussa-Ivaldi, & Gisz- ter, 1991; Carey &Gel-

man, 1991; Etcoff, 1986; Freyd, 1987; Kosslyn,

1980; Liberman & Mattingley, 1985, 1989; Lind-

blom, 1986, 1988; Maloney & Wandell, 1985; Marr,

1982; Pinker, 1984, 1989; Poggio et al., 1985; Prof-

ﬁtt & Gilden, 1989; Shepard, 1981, 1984, 1987a;

Shiffrar & Freyd, 1990; Spelke, 1988, 1990). In

fact, the reality has always been that every ﬁeld

of psychology bristles with observations of content-

dependent phenomena. Freedom from the axiom

that all psychological phenomena must be explained

by content-independent machinery has allowed psy-

chologists to move ahead to explore-and to view

as meaningful-the rich content-sensitive effects that

permeate psychological phenomena (e.g., Astington

et al., 1988; A. Brown, 1990; Carey, 1985b; Carey &

Gelman, 1991; Cosmides, 1989; Cosmides & T ooby

, this volume; Gelman & Markman, 1986, 1987;

Gigerenzer& Hug, in press; Gigerenzer & Murray,

1987; Keil, 1989; Manktelow & Over, 1991). For-

merly, these omnipresent content effects were con-

sidered an embarrassment to be explained away or

else dismissed as noise. Now they are considered to

be primary data about the structure of the mind.

Outside of cognitive psychology, the emergence of

post-Standard Model approaches derived their im-

petus from branches of evolutionary biology. In the

1950s and 1960s, the successful application of evo-

lutionary approaches to animal behavior in ethol-

ogy and its successor disciplines provided evidence

of domain-speciﬁc mechanisms that was difﬁcult to

ignore (e.g., attachment, emotion, phobias, mating,

for- aging, navigation). This trend was accelerated

by the rapid advances in evolutionary biology over

the last three decades, which made the previously

clouded connection between evolution and behav-

ior somewhat clearer. These advances included ( 1)

more coherent approaches to the nature-nurture is-

sue, (2) a more rigorous foundation for the theory of

natural selection (Williams, 1966), (3) formal anal-

yses of what behaviors would be favored by selec-

tion in a variety of newly explored domains (e.g.,

Charnov, 1976; Hamilton, 1964; Maynard Smith,

1982; Stephens & Krebs, 1986; Trivers, 1971, 1972,

1974; Williams, 1966), and (4) a cascade of success-

ful applications of these theories to animal behavior

(Alexander, 1974; Daly & Wilson, 1984b; Krebs &

Davies, 1984; Wilson, 1975).

Just as in the case of nonhuman behavior, evo-

lutionarily informed studies of human choice, mo-

tivation, emotion, and action also bristle with doc-

umented phenomena that cannot be accounted for

with content-independent architectures and a short

list of drives, rewards, or reinforces (the chapters

in this volume are a small sampling of such cases).

For example, ever since the Harlows demolished the

myth that an infant’s love for its mother was a con-

ditioned response to food rewards, the rich collec-

tion of co-adapted mechanisms in the mother and

infant has been a productive focus of psychological

investigation (e.g., Bowlby, 1969). Profet (this vol-

ume) identiﬁes a maternal psychological adaptation

for the protection of the fetus during embryo- gene-

sis. Fernald (this volume) explores the communica-

tive adaptations mothers have to the infant’s percep-

tual limitations. Moreover, cross-cultural regulari-

ties in fall-rise patterns of maternal fundamental fre-

quency provide an elegant illustration that a child

and adult initially communicate by virtue of what

they share through their common human metacul-

ture. Communication through such human univer-

sals is a pre- condition for the child’s acquisition of

the culturally speciﬁc. Facial expressions of emotion

represent another evolved modality through which

humans communicate situation-construals, and the

cross-culturally stable features of emotional expres-

sion provide another critical foundation for human

metaculture. Ekman and his colleagues have estab-

lished one of the earliest and most sophisticated tra-

ditions of evolutionary psychological research, and

these studies of emotional expression represent a

major achievement in modern psychology (e.g., Ek-

man, 1973, 1982, 1984; Ekman & Friesen, 1975; Ek-

man, Levenson, & Friesen, 1983). Etcoff( 1986) has

marshalled substantial neuropsychological evidence

that humans have mechanisms specialized for the

identiﬁcation of emotional expression-an adaptation

to an important, stable feature of ancestral social and

cultural environments.

Indeed, ever since Darwin ( 1871, 1872), emo-

tions have been seen as the product of the evolu-

tionary process and usually, although not always,

as functional adaptations(e.g.,Arnold, 1960, 1968;

Chance, 1980; Daly,etal., 1982; Eibl-Eibesfeldt,

1975; Ekman, 1982; Frijda, 1986; Hamburg, 1968;

Izard, 1977; Otte, 1974; Plutchik, 1980; Tomkins,

1962, 1963; Tooby & Cosmides, 1990b; and many

others). Functional or not, the emotions collec-

tively provide a dense and pervasive network of

domain-speciﬁc phenomena that have consistently

resisted assimilation into any Standard Model the-

ory. However, in contrast to their Standard Model

reputation as crude and indiscriminate responses, on

close scrutiny each speciﬁc emotion appears to be

an intricately structured information-sensitive regu-

latory adaptation. In fact, the emotions appear to

be designed to solve a certain category of regula-

tory problem that inevitably emerges in a mind full

of disparate, functionally specialized mechanisms-

the problem of coordinating the menagerie of mech-

anisms with each other and with the situation be-

ing faced (Tooby, 1985; Tooby & Cosmides, 1990b;

Nesse, 1990).

Daly and Wilson have been exploring the evolved

complexity and functional subtlety of the human mo-

tivational system. They have produced a substan-

tial body of ﬁndings supporting speciﬁc hypotheses

they derived from a broad array of adaptation- ist

theories (Daly & Wilson, 1981, 1982, 1984b, 1987a,

1987b, 1988; Daly, et al., 1982; Wilson & Daly, 1985,

1987, this volume). Their particular interest has

been the evolved motivational systems that regulate

parental care, spousal relations, sexual jealousy, sex-

ual proprietariness, and risk-taking. By using behav-

ioral phenomena such as violence and homicide as

dependent measures, they have been able to inves-

tigate many aspects of these evolved motivational

systems-including how their operation is affected by

factors such as gender, age, kinship, reproductive

value, number of children, and other situational vari-

ables (see also Mann, this volume). Similarly, in the

area of human mate choice and sexuality, the work

of Symons, Buss, and many others shows that the

construct of a "sex drive" is completely inadequate to

cope with the structured richness of the situational

factors processed by the differentiated sexual psy-

chologies of men and women across cultures (e.g.,

Buss, 1987, 1989, 1991, this volume, in prep.; El-

lis, this volume; Sadalla, Kenrick, & Vershure, 1987;

Symons, 1979; Townsend, 1987). These studies in-

dicate that existing theories of motivation will have

to be replaced with theories positing a far more elab-

orate motivational architecture, equipped with an

extensive set of evolved information-processing al-

gorithms that are contingently sensitive to a long list

of situational contents and contexts.

Thus, the examination of even a small sampling of

non-Standard Model behavioral studies by a handful

of researchers such as Bowlby, Daly and Wilson, Ek-

man, Femald, Marks, Buss, and Symons leads to the

conclusion that the human mind contains evolved

emotional and motivational mechanisms that are

speciﬁcally targeted to address adaptive problems in-

volved in parenting, emotional communication with

infants and adults, kinship, mate choice, sexual at-

traction, aggression, the avoidance of danger, mate

guarding, effort allocation in child care, and so on.

That is, humans have psychological adaptations that

contain contentful structure speciﬁcally "about" their

mothers, "about" their children, "about" the sexual

behavior of their mates, "about" those identiﬁed by

cues as kin, "about" how much to care for a sick child,

and so on, and these contents are not derived exclu-

sively from either a short list of drive or from cultur-

ally variable, socially learned "values."

In short, the central tenets of Standard Model psy-

chology are contradicted b results from a large and

rapidly growing body of research on humans and

nonhuman: from the cognitive community, from the

evolutionary community, from the behavioral ecol-

ogy community, and from other research commu-

nities as well (for example much of psychobiology,

comparative psychology, and neuroscience). Content

independent mechanisms simply cannot generate or

explain the richly patterned behavior and knowledge

structures that appear when one’s research focus is

widened beyond arbitrary laboratory tasks to include

the complex performances orchestrated by natural

competences on real world tasks. Moreover, un-

like most Standard Model theories and results, these

kinds of studies and hypotheses withstand cross-

cultural scrutiny and indicate that a great deal of

the substance of social life attributed to "culture"

around the world is in fact caused by the operation

of contingently responsive domain- speciﬁc mecha-

nisms. These converging results are accumulating

into a strikingly different picture than that provided

by the Standard Social Science Model. They indi-

cate that a universal, evolved psychological architec-

ture that is ﬁlled with contingently responsive mech-

anisms infuses distinctively human patterns into the

life of every culture.

The Frame Problem and the Weakness of

Content-Independent, Domain General Mecha-

nisms

From Flexibility to Adaptive Flexibility

In the passage from Standard Model to post-

Standard Model psychology it seems fair to say

that the greatest reversal lay in how content-

independence and domain-generality came to be

regarded. Many modern researchers recognize

that content-independent, general problem-solvers

are inherently weak in comparison to content-

specialized mechanisms. From a traditional point of

view, however, it seemed sensible to regard general-

ity as an enhancement of the capacity of a system:

The system is not prevented from assuming certain

states or kept from doing what is adaptive (or desir-

able) by a "rigid" or "biased" architecture. General-

ity of application seems like such an obvious virtue

and content-independence seems like such an obvi-

ous road to ﬂexible behavior, what could possibly be

wrong with them? As Marvin Harris puts this line of

reasoning, "Selection in the main has acted against

genetically imposed limitations on human cultural

repertoires" ( 1979, p. 136). Why rigidly prevent

the system from engaging in certain behaviors on

those occasions when they would be advantageous?

Moreover, why not have an "unbiased" architecture

in which the actual structure of local circumstances

impresses a true picture of itself in a free, objec-

tive, and unconstrained way? In this view, content-

speciﬁcity in evolved psychological design is imbued

with all the legendary attributes of "biology"-rigidity,

inﬂexibility, and constraint. It is viewed as prevent-

ing the system from achieving advantageous states

that would otherwise naturally come about.

So what, after all, is so wrong with domain-

general systems? Why do cognitive psychologists

and artiﬁcial intelligence researchers consistently

ﬁnd them too weak to solve virtually any complex

real world task? Why isn’t "ﬂexibility" in the form

of con- tent-independence a virtue? The answers to

these questions emerge from one clariﬁcation and

from two basic facts. The two facts have already

been touched on many times: (I) possibilities are in-

ﬁnite; and (2) desirable outcomes-by any usual hu-

man, evolutionary, or problem-solving standard-are

a very small subset of all possibilities.

The clariﬁcation concerns the kind of plasticity

and ﬂexibility that are implicitly being referred to.

Literally, plasticity, or ﬂexibility, is the simple capac-

ity to vary in some dimension. The more dimensions

of possible variation, the greater the "plastic- ity."

Hence, a lump of clay is very plastic with respect to

shape (although not with respect to substance, den-

sity, and so on). Similarly, there is an inﬁnite num-

ber of ways that humans and other animals could

potentially act. The difﬁculty lies in the fact that

the overwhelming majority of behavioral sequences

would be lethal in a few hours, days, or weeks. The

set of behaviors that leads even to temporary indi-

vidual survival- let alone to reproduction or design-

propagation-constitutes an extremely miniscule sub-

set of all possible behavioral sequences. Thus, the

property of freely varying behavior in all dimensions

independent of conditions is not advantageous: It is

evolutionarily and individually ruinous.

Accordingly, to be endowed with broad behavioral

plasticity unconnected to adaptive targets or envi-

ronmental conditions is an evolutionary death sen-

tence, guaranteeing that the design that generates

it will be removed from the population. Designs

that produce "plasticity" can be retained by selection

only if they have features that guide behavior into

the inﬁnitesimally small regions of relatively success-

ful performance with sufﬁcient frequency. In real-

ity, terms such as ﬂexibility or plasticity are implic-

itly used to mean something very different from the

simple "capacity to vary." They are implicitly used to

mean the capacity to adjust behavior (or morphol-

ogy) as a coordinated response to the speciﬁcs of

local conditions so that the new behavior is partic-

ularly appropriate to or successful in the speciﬁc cir-

cumstances faced.

This narrowly specialized form of ﬂexibility re-

quires three components: (1) a set of mechanisms

that deﬁne an adaptive target (such as ﬁnding food,

ﬁnding home, or ﬁnding a mate); (2) a set of mecha-

nisms that can compute or otherwise determine what

responses are most likely to achieve the adaptive tar-

get in each speciﬁc set of circum- stances that one is

likely to encounter; and (3) the ability to implement

the speciﬁc response once it is determined. Plasticity

in the "lump of clay /capacity to vary" sense refers

only to the third component: If an organism has cor-

rectly computed what it is advantageous to do, then

(and only then) is it disadvantageous to be inﬂexi-

bly pre- vented from implementing those changes by

some ﬁxed element of the system.

In fact, plasticity (e.g., variability) tends to be in-

jurious everywhere in the architecture except where

it is guided by well-designed regulatory mechanisms

that improve outcomes or at least do no harm.

It would be particularly damaging if these regula-

tory mechanisms were themselves capriciously "plas-

tic," instead of rigidly retaining those computational

methods that produce advantageous responses to

changing conditions. Thus, plasticity is only ad-

vantageous for those speciﬁc features of the organ-

ism that are governed by procedures that can com-

pute the speciﬁc changes or responses that will be,

on average, more successful than a ﬁxed pheno-

type. Adaptive ﬂexibility requires a "guidance sys-

tem" (Cosmides & Tooby, 1987; Tooby, 1985).

The most important conclusion to be derived from

this line of reasoning is that adaptive ﬂexibility can

only evolve when the mechanisms that make it pos-

sible are embedded within a co-evolved guidance

system. Consequently, the expansion of behavioral

and cognitive ﬂexibility over evolutionary time de-

pended acutely on how well-designed these compu-

tational guidance systems became. There is nothing

in the ability to vary per se that naturally leads sys-

tems to gravitate toward producing successful per-

formances. It is the guidance system itself that is

doing the bulk of the interesting regulation of out-

comes, with the "potential to vary" component ex-

plaining very little about the situation. Thus, Gould’s

( 1977a, 1979) faith in the explanatory power of the

SSSM concept of generalized human "biological po-

tential" depends either on (I) an unjustiﬁed teleolog-

ical panglossianism (e.g., unguided processes, such

as accidental brain growth, just happen to "work out

for the best, " giving humans the desire to care for

their children, to defend themselves when attacked,

to cooperate; the ability to recognize faces, to ﬁnd

food, to speak a language. ..), or (2) the unacknowl-

edged existence of co-evolved cognitive adaptations

that guide behavioral plasticity toward the achieve-

ment of adaptive targets.

It is the necessary existence of these co-evolved

guidance systems that has, for the most part, es-

caped the attention of Standard Model advocates.

In fact, the SSSM ediface is built on the conﬂation

of two distinct notions of ﬂexibility: ( I) ﬂexibility

as the absence of any limits on responses, and (2)

ﬂexibility as the production of contextually appropri-

ate responses. Advocates of the SSSM imagine that

ﬂexibility in the ﬁrst sense-an absence of limits on

variation-is easy to computationally arrange (just re-

move all "constraints"). But they also assume this is

the same as-or will automatically produce-ﬂexibility

in the second sense: adaptive, successful, or con-

textually appropriate behavior. Post-Standard Model

psychology rests on the recognition that ﬂexibility in

this second sense is not something that is teleologi-

cally inevitable once constraints are removed, but is,

instead, something very improbable and difﬁcult to

achieve, requiring elaborate functionally organized

machinery.

The Weakness of Content-independent Archi-

tectures

If the doors of perception were cleansed ev-

erything would appear to man, as it is, inﬁ-

nite. -WILLIAM BLAKE

If plasticity by itself is not only useless but inju-

rious, the issue then becomes, what kind of guid-

ance systems can propel computational systems suf-

ﬁciently often toward the small scattered islands of

successful outcomes in the endless expanse of al-

ternative possibilities? Attempts over the last three

decades to answer this question have led directly to

two related concepts, called by artiﬁcial intelligence

researchers and other cognitive scientists combina-

torial explosion and the frame problem.

Combinatorial explosion is the term for the fact

that with each new degree of freedom added to a sys-

tem, or with each new dimension of potential vari-

ation added, or with each new successive choice in

a chain of decisions, the total number of alternative

possibilities faced by a computational system grows

with devastating rapidity. For example, if you are

limited to emitting only one out of 100 alternative

behaviors every successive minute (surely a gross un-

derestimate: raise arm, close hand, toss book, ex-

tend foot, say "Ravel," etc.), after the second minute

you have 10,000 different behavioral sequences from

which to choose, a million by the third minute, a tril-

lion by six minutes, and 1012

r possible alternative

sequences after only one hour-a truly unimaginable

number. Every hour, each human is surrounded by

anew and endless expanse of behavioral possibility.

Which leads to the best outcome? Or, leaving aside

optimality as a hopelessly utopian luxury in an era of

diminished expectations, which sequences are non-

fatal? The system could not possibly compute the

anticipated out- come of each alternative and com-

pare the results, and so must be precluding without

complete consideration the overwhelming majority

of branching pathways. What are the principles that

allow us to act better than randomly?

Combinatorial explosion attacks any system that

deals with alternatives, which means any system that

is ﬂexible in response or has decisions to make. The

more ﬂexible the system, the greater the problem.

Even worse, knowledge acquisition is impossible for

a computational system equipped only with the lim-

ited information it can gain through its senses; this

is because the number of alternative states of affairs

in the world that are consistent with its sense data is

inﬁnite. For example, if cognitive mechanisms are at-

tempting to infer the meaning of an unknown word,

there is an inﬁnite set of potential meanings. If per-

ceptual mechanisms are trying to construct a three

dimensional model of the local world from a visual

array, there is an inﬁnite number of different ways to

do it that are all consistent with the array. For any ﬁ-

nite sample of sentences encountered, there exists an

inﬁnite number of alternative grammars that could

have generated them. If one is making a decision

about how to forage there is, practically speaking,

an inﬁnite number of possibilities. Moreover, ran-

dom choice is not a general solution to the problem

because for most adaptive or humanly deﬁned prob-

lems the islands of success are inﬁnitesimal next to

the illimitable seascapes of failure. And for biologi-

cal systems, success and failure are not arbitrary. The

causal world imposes a nonarbitrary distinction be-

tween detecting in one’s visual array the faint outline

of a partly camouﬂaged stalking predator and not

detecting it because of alternative interpretive pro-

cedures. Nonpropagating designs are removed from

the population, whether they believe in naive realism

or that everything is an arbitrary social construction.

The inexhaustible range of possibilities latent in

behavior, categorization, interpretation, decision,

and so on, is a not just an abstract philosophi-

cal point. It is an implacable reality facing every

problem-solving computational system. Each prelin-

guistic child trying to learn her own language or to

induce new knowledge about the world is faced with

this problem; so is every artiﬁcial intelligence sys-

tem. In artiﬁcial intelligence research, it is called the

"frame problem " (Boden, 1977); in linguistics, this

problem is called the "poverty of the stimuli" (Chom-

sky, 1975); in semantics, it is called the problem of

"referential ambiguity" (Gleitman & Wanner, 1982);

in developmental psychology, it is called the "need

for constraints on induction" (Carey, 1985a); in per-

ception, they say that the stimulus array "underde-

termines" the interpretation. Any design for an or-

ganism that cannot generate appropriate decisions,

inferences, or perceptions because it is lost in an

ocean of erroneous possibilities will not propagate,

and will be removed from the population in the next

generation. As selection pressures, combinatorial ex-

plosion and the frame problem are at least as merci-

less as starvation, predation, and disease.

With this as background, the converging results

from artiﬁcial intelligence, perception, cognitive de-

velopment, linguistics, philosophy, and evolution-

ary biology about the weaknesses of domain-general

content-independent mechanisms are not difﬁcult to

fathom. One source of difﬁculty can be sketched out

quickly. If a computational system, living or elec-

tronic, does not initially know the solution to the

problem it faces then its procedures must operate to

ﬁnd a solution. What methods do content indepen-

dent systems bring to problem-solving? To describe

a system as domain-general or content-independent

is to say not what it is but only what it lacks: It

lacks any speciﬁc a priori knowledge about the re-

current structure of particular situations or problem-

domains, either in declarative or procedural form,

that might guide the system to a solution quickly. It

lacks procedures that are specialized to detect and

deal with particular kinds of problems, situations,

relationships, or contents in ways that differ from

any other kind of problem, situation, relationship,

or content. By deﬁnition, a domain-general system

takes a "one size ﬁts all" approach.

To understand the importance of this, consider the

deﬁnition of an adaptation. An adaptation is a re-

liably developing structure in the organism, which,

because it meshes with the recurrent structure of the

world, causes the solution to an adaptive problem.

It is easy to see how a speciﬁc structure, like a bug

detector in a frog’s retina, in inter- action with bug

trajectories in the local environment, solves a feed-

ing problem for the frog. It is easy to see how the

Westermarck sexual disinterest mechanism combines

with the co-residence cue to diminish the probability

of sex between close relatives (Shepher, 1983; Wolf

& Huang, 1980). When the class of situations that

a mechanism is designed to solve is more narrowly

deﬁned, then ( I) the situations will have more re-

current features in common, and therefore (2) the

mechanism can "know" more in advance about any

particular situation that is a member of this class.

As a result, (3) the mechanism’s components can

embody a greater variety of problem-solving strate-

gies. This is because mechanisms work by meshing

with the features of situations and, by deﬁnition;

narrowly deﬁned situations have more features in

common. Our depth perception mechanism has this

property, for example: It works well because it com-

bines the output of many small modules, each sensi-

tive to a different cue correlated with depth. In ad-

dition, (4) the narrower the class, the more likely

it is that a good, simple solution exists-a solution

that does not require the simultaneous presence of

many common features. The frog can have a simple

"bug detector" precisely because insects share fea-

tures with one another that are not shared by many

members of more inclusive classes, such as "animals"

or "objects."

In contrast, the more general a problem-solving

technique is, the larger the range of situations across

which the procedure must successfully apply itself.

When the class of situations that a mechanism must

operate over is more broadly deﬁned, then (I) the

situations will have fewer recurrent features in com-

mon, therefore (2) the mechanism can "know" less in

advance about any particular situation that is a mem-

ber of this class. Because (3) broadly deﬁned situa-

tions have so few features in common for a mecha-

nism to mesh with, there exist fewer strategies capa-

ble of solving the problem.

This result is logically inevitable. Every kind of

problem-solving strategy that applies to a more in-

clusive class also applies to every subset within it;

but not every strategy that applies to a narrowly de-

ﬁned class will apply to the larger classes that con-

tain it (e.g., all insects are objects, but not all ob-

jects are insects). By identifying smaller and smaller

problem domains on the basis of an increasing set

of recurrent similarities, more and more problem-

solving strategies can be brought to bear on that set.

Conversely, by widening the problem domain that a

mechanism must address, strategies that worked cor-

rectly on only a subset of problems must be aban-

doned or subtracted from the repertoire because

they give incorrect answers on the newly included

problems in the enlarged domain. As problem do-

mains get larger and more broadly deﬁned, a smaller

and smaller set of residual strategies is left that

remains applicable to the increasingly diverse set

of problems. At the limit of perfect generality, a

problem- solving system can know nothing except

that which is always true of every situation in any

conceivable universe and, therefore, can apply no

techniques except those that are applicable to all

imaginable situations. In short, it has abandoned vir-

tually anything that could lead it to a solution.

This weakness of domain-general architectures

arises not because all relatively general problem-

solving techniques are useless; indeed, many are

very useful-the ability to reject propositions because

they are contradicted, the ability to associate, and

the ability to recalibrate based on the consequences

of actions, for example. The weakness arises because

content-sensitivity and specialization are eliminated

from the architecture. By deﬁnition, a content-

independent architecture does not distinguish be-

tween different problem-domains or content classes;

therefore, it is restricted to employing only general

principles of problem-solving that can apply to all

problems.

In contrast, a content-dependent domain-speciﬁc

architecture does identify situations as members of

speciﬁc problem domains and content classes. Be-

cause of this, it can maintain a repertoire of special-

ized problem-solving techniques that are only acti-

vated when they encounter the delimited domains to

which they are applicable (e.g., snakes, sex with kin,

grammar, falling in love, faces). At the same time,

a pluralistic architecture can simultaneously activate

every other problem-solving technique appropriate

to the larger and more inclusive classes that contain

the problem encountered (for faces: face recogni-

tion, object recognition, association formation, and

so on). Thus, a domain-speciﬁc architecture can de-

ploy every general problem-solving technique at the

disposal of a domain-general architecture and a mul-

titude of more speciﬁc ones as well. This sensible ap-

proach to organizing a problem-solving architecture

is exactly what is ruled out by SSSM advocates of a

content-independent mind whose procedures oper-

ate uniformly over every problem or domain.

To put it in adaptationist terms, what does the

work of adaptive problem-solving for organisms is (

I) the recurrent structure of the world relevant to the

problem, in interaction with (2) the recurrent struc-

ture of the adaptation. The more broadly deﬁned the

problem domain is ( I) the less recurrent structure

can be supplied by the world (because more diverse

situations have less recurrent structure in common),

and (2) the less recurrent structure can be supplied

by the adaptation in the form of problem-solving

procedures that are solution-promoting across a di-

verse class of situations. The erosion of both sets

of problem-solving structures-those in the adapta-

tion and those in the world-increasingly incapaci-

tates the system. This can sometimes be compen-

sated for, but only through a correspondingly costly

increase in the amount of computation used in the

attempt to solve the problem. The less the system

knows about the problem or the world to begin with,

the more possibilities it must contend with. Permuta-

tions being what they are, alternatives increase expo-

nentially as generality increases and combinatorial

explosion rapidly cripples the system. A mechanism

unaided by domain-speciﬁc rules of relevance, spe-

cialized procedures, "preferred" hypotheses, and so

on could not solve any biological problem of routine

complexity in the amount of time the organism has

to solve it, and usually could not solve it at all.

It is the perennial hope of SSSM advocates

within the psychological community that some

new technology or architecture (wax impressions,

telephone switching, dig- ital computers, symbol-

processing, recursive programming languages, holo-

grams, non-von Neumann architectures, parallel-

distributed processing-a new candidate emerges ev-

ery decade or so), will free them to return to

empiricism, associationism, domain-generality and

content-independence (where the SSSM tells them

they should go). Nevertheless, the functional neces-

sity of content-speciﬁcity emerges in every technol-

ogy because it is a logical inevitability. Most recently,

researchers are establishing this all over again with

connectionism (e.g., Jacobs, Jordan, & Barto, 1990;

Miller & Todd, 1990; Pinker & Prince, 1988; Todd

& Miller, 1991a, 199Ib). Combinatorial explosion

and the frame problem are obstacles that can only

be overcome by endowing computational architec-

tures with contentful structure. This is because the

world itself provides no framework that can decide

among the inﬁnite number of potential category di-

mensions, the inﬁnite number of relations, and the

inﬁnite number of potential hypotheses that could

be used to analyze it.

The Necessity of Frames.

Artiﬁcial intelligence research is particularly il-

luminating about these issues because explicitness

is demanded in the act of implementing as pro-

grams speciﬁc hypotheses about how problems can

be solved. By the program’s operation, one can tell

a great deal about the adequacy of the hypothesis.

Moreover, artiﬁcial intelligence researchers became

interested in getting computers and robots to per-

form real world tasks, where, just as in evolutionary

biology, action is taken in a real, structured, and con-

sequential environment. As a result, artiﬁcial intel-

ligence researchers can tell unambiguously whether

the decisions the system makes are a success or a fail-

ure. The range of problems studied in artiﬁcial intel-

ligence widened beyond cognitive psychology’s more

traditional, philosophy-derived concerns, to include

problems such as the regulation of purposive action

in a three-dimensional world.

To their great surprise, artiﬁcial intelligence re-

searchers found that it was very difﬁcult to dis-

cover methods that would solve problems that hu-

mans ﬁnd easy, such as seeing, moving objects or

even tying shoelaces. To get their programs to han-

dle even absurdly simpliﬁed tasks (such as moving

a few blocks around), they were forced to build in

substantial "innate knowledge" of the world. As a

practical matter, this "knowledge" was either in the

form of(l) content-dependent procedures matched

closely to the structural features of the task do-

main within which they were designed to operate,

or (2) representations (data structures) that accu-

rately reﬂected the task domain (i.e., "knowledge

of the world"). To move an object, make the sim-

plest induc- tion, or solve a straightforward prob-

lem, the computer needed a sophisticated model of

the domain in question, embodied either in proce-

dures or representations. Artiﬁcial intelligence re-

search demonstrated in a concrete, empirical form,

the long-standing philosophical objections to the tab-

ula rasa (e.g., Hume, 1977/1748; Kant, 1966/1781;

Popper, 1972; Quine, 1960,1969). These demonstra-

tions have the added advantage of bracketing just

how much "innate" structure is necessary to allow

learning to occur. ) Artiﬁcial intelligence researchers

call the speciﬁc contentful structures that problem-

solving systems need to be endowed with frames.

For this reason, the consistent inability of systems

without sufﬁciently rich and specialized frames to

solve real problemsis called the frameproblem (e.g.,

Boden, 1977; F. M. Brown, 1987; Fodor, 1983). A

frame provides a "world-view": It carves the world

into deﬁned categories of entities and properties, de-

ﬁnes how these categories are related to each other,

suggests operations that might be performed, de-

ﬁnes what goal is to be achieved, provides methods

for interpreting observations in terms of the problem

space and other knowledge, pro- vides criteria to dis-

criminate success from failure, suggests what infor-

mation is lacking and how to get it, and so on. For

example, one might apply a spatial/object frame to a

situation. In such a frame, the local world is carved

into empty space and objects, which are cohesive,

have boundaries deﬁned by surfaces, and move as a

unit. They have locations and orientations with re-

spect to each other. They have trajectories and, if

solid, cannot pass through one another (unless they

change the shape of the object passed through), and

so on. In such a framing, humans are simply objects

like any other and are not expected to pass through

other solid objects. Alternatively, one might have a

coalitional framing (present, for example, in a foot-

ball game or a war), in which humans are a relevant

and differentiated entity and are construed as ani-

mate goal- seeking systems that are members of one

of two mutually exclusive social sets; the members

of each set are expected to coordinate their behavior

with each other to reach goals that cannot be mutu-

ally realized for both sets; the goal of each set is to

thwart the purposes of the other, and so on. In our

own work, we have attempted to sketch out some of

the framing necessary for humans to engage in so-

cial exchange (Cosmides, 1989; Cosmides & Tooby,

1989, this volume). Very general mechanisms have

frames as well: In the formal logic of the propo-

sitional calculus, the problem-space is deﬁned syn-

tactically in terms of sets of propositions, truth val-

ues, and rules of inference such as modens ponens

and modus tollens. In this frame, the content of the

propositions is irrelevant to the operation of the rules

of inference.

The solution to the frame problem and combina-

torial explosion is always the same whether one is

talking about an evolved organism or an artiﬁcial in-

telligence system. When the information available

from the world is not sufﬁcient to allow learning to

occur or the problem to be solved, it must be sup-

plied from somewhere else. Because the world can-

not supply to the system what the system needs ﬁrst

in order to learn about the world, the essential ker-

nals of content-speciﬁc framing must be supplied ini-

tially by the architecture. For an artiﬁcial intelligence

system, a programmer can sup- ply it. For organisms,

however, it can only be supplied through the pro-

cess of natural selection, which creates reliably de-

veloping architectures that come equipped with the

right frames and frame-builders necessary to solve

the adaptive problems the species faced during its

evolutionary history.

Because of their survival into the present, we know

for a fact that living species can reliably solve an

enormous array of problems necessary to consis-

tently reproduce across thousands of generations in

natural environments. Moreover, the signal lesson

of modern evolutionary biology is that this adaptive

behavior requires the solution of many information-

processing problems that are highly complex-far

more complex than is commonly supposed (Cos-

mides & Tooby, 1987, 1989). If one bothers to an-

alyze virtually any adaptive problem human hunter-

gatherers solve, it turns out to require an incredi-

ble amount of evolved specialization (see, e.g., Cos-

mides & Tooby, 1989, this volume). Given (I)

the complexity of the world, (2) the complexity

of the total array of adaptive tasks faced by liv-

ing organisms, and (3) the sensitive frame- depen-

dence of problem-solving abilities, the psycholog-

ical architecture of any real species must be per-

meated with domain-speciﬁc structure to cause re-

liable reproduction in natural environments. Cur-

rent research in cognitive psychology and artiﬁcial

intelligence indicates that Standard Model theories

are far too frame-impoverished to solve even artiﬁ-

cially simpliﬁed computational problems (e.g., iden-

tifying and picking up soda cans in the MIT artiﬁ-

cial intelligence laboratory), let alone the complex

information-processing problems regularly imposed

by selective forces operating over evolutionary time.

Our minds are always automatically applying a

rich variety of frames to guide us through the world.

Implicitly, these frames appear to us to be part of the

world. For precisely this reason, we have difﬁculty

appreciating the magnitude, or even the existence

of, the frame problem. Just as the effortlessness of

seeing led artiﬁcial intelligence researchers to under-

estimate the complexity of the visual system, the au-

tomatic and effortless way in which our minds frame

the world blinds us to the computational complexity

of the mechanisms responsible. When anthropolo-

gists go to other cultures, the experience of variation

awakens them to things they had previously taken

for granted in their own culture. Similarly, biolo-

gists and artiﬁcial intelligence researchers are "an-

thropologists" who travel to places where minds are

far stranger than anywhere any ethnographer has

ever gone. We cannot understand what it is to be

human until we learn to appreciate how truly differ-

ent nonhuman minds can be, and our best points of

comparison are the minds of other species and elec-

tronic minds. Such comparisons awaken us to an en-

tire class of problems and issues that would escape

us if we were to remain "ethnocentrically" focused

on humans, imprisoned by mistaking our mentally

imposed frames for an exhaustive demarcation of re-

ality.

When we examine electronic minds that truly have

no frames and then try to give them even a few of

our own real world capacities, we are made force-

fully aware of the existence of the immensely intri-

cate set of panhuman frames that humans depend on

to function in the world, to communicate with one

another, and to acquire additional frames through

social inference from others (i.e., to "learn culture").

Geertz’s (1973) studies in Bali acquainted him with

some of the culturally variable frames that differ be-

tween Bali and the United States, but, as his writ-

ings make clear, they left him oblivious to the en-

compassing, panhuman frames within which these

variable elements were embedded (0. E. Brown,

1991). If he had widened his scope to include other

animal species, he would have been made strongly

aware of this dense level of universal and human-

speciﬁc metacultural frames-a level that should in-

terest every anthropologist because it permeates and

structures every aspect of human life. Indeed, if

Geertz had widened his scope still further to in-

clude electronic minds as points of comparison, he

might have come to realize the sheer magnitude of

what must be supplied by evolution to our psycho-

logical architectures for us to be recognizably hu-

man. Per- haps he might also have come to recog-

nize that he and the "simple" Balinese ﬁghting cocks

he watched even shared many frames lacking from

artiﬁcial intelligence systems (about things such as

space, motion, vision, looming threats, pain, hunger,

and, per- haps, conﬂict, rivalry, and status changes

after ﬁghts). Biology, cognitive psychology, and ar-

tiﬁcial intelligence research comprise a new form of

ethnography, which is revealing the previously invis-

ible wealth of evolved frames and specialized frame-

builders that our evolved psychological architecture

comes equipped with.

The Evolvability Criterion and Standard Model

Architectures

In a solvability analysis, the researcher asks

whether a proposed architecture is capable of gen-

erating a behavior that we know humans (or the rel-

evant species) regularly engage in, whether adaptive

or not. But one can also evaluate a proposed archi-

tecture by asking how it would fare in solving the

actual adaptive problems a species is known to have

regularly confronted and solved during its evolution-

ary history .Because non- human and human minds-

i.e., the computational systems responsible for reg-

ulating behavior-were produced by the evolutionary

process operating over vast expanses of time, tenable

hypotheses about their design must be drawn from

the class of designs that evolution could plausibly-

or at least possibly-have produced. To be adequate,

proposed designs must be able to account for the so-

lution of the broad array of distinct problems inher-

ent in reliable reproduction over thousands of gen-

erations under ancestral conditions. In short, a can-

didate design must satisfy the evolvability criterion.

In essence, designs that are more plausible according

to criteria drawn from evolutionary biology are to be

preferred over designs that are less plausible.

Some rules for evaluating hypotheses by the evolv-

ability criterion are as follows:

I. Obviously, at a minimum, a candidate architec-

ture must be able to perform all of the tasks and

subtasks necessary for it to reproduce. It can have

no properties that preclude, or make improbable, its

own reproduction across multiple generations in nat-

ural environments. Just by itself, this is a difﬁcult cri-

terion to meet. No known Standard Model psycho-

logical architecture can solve all or even very many

of the problems posed by reproduction in natural en-

vironments.

2. Given that human minds evolved out of prehu-

man primate minds, a hypothesis should not entail

that an architecture that is substantially inferior at

promoting its own propagation (its inclusive ﬁtness)

replace an architecture that was better designed to

promote ﬁtness under ancestral conditions. There

is now known to be an entire range of competences

and specialized design features that enhance propa-

gation in a large array of other species. A candidate

architecture should be at least roughly comparable

to them in their ability to solve the classes of adap-

tive problems humans and other primate species mu-

tually faced. For this reason, it is not sufﬁcient to in-

corporate into a general-purpose system a few drives

that account for why the organism does not die of

thirst or hunger in a few days. Even though some

psychological architectures of this kind might con-

ceivably manage their own reproduction under arti-

ﬁcially protected circumstances, they contain noth-

ing that would solve other obvious propagation-

promoting tasks that have called forth adaptive spe-

cializations in innumerable other species. Thus, to

be plausible, a proposed human architecture should

cause individuals to help relatives more or less ap-

propriately, to defend sexual access to their mates,

to for- age in a relatively efﬁcient way, and so on.

The SSSM view that human evolution was a pro-

cess of erasing "instincts" violates the evolvability cri-

terion unless it can be shown that for each puta-

tively "erased" adaptive specialization, the general-

purpose mechanism that is proposed to have re-

placed it would have solved the adaptive problem

better (Tooby, 1985; Tooby & DeVore, 1987). To our

knowledge, no general mechanism operating under

natural circumstances has ever been demonstrated

to be superior to an existing adaptive specialization.

3. A candidate architecture should not require

the world to be other than it really is. For exam-

ple, models of grammar acquisition that assume that

adults standardly correct their children ’s grammati-

cal errors do not meet this condition (Pinker, 1989).

Nor do socialization models that require children to

be taught where their own interests lie by individu-

als with conﬂicting interests-for many domains, this

class even includes the child’s own parents and sib-

lings (e.g., Hamilton, 1964; Trivers, 1974). An archi-

tecture that was completely open to manipulation by

others, without any tendency whatsoever to modify

or resist exploitive or damaging social input would

be strongly selected against. For this reason, cogni-

tive architectures that are passive vehicles for arbi-

trary semiotic systems are not plausible products of

the evolutionary process.

4. In a related vein, a candidate theory should not

invoke hypotheses that require assumptions about

the coordinated actions of others (or any part of the

environment) unless it explains how such coordina-

tion reliably came about during Pleistocene hunter-

gatherer life. For example, if the model proposes

that people acquire certain adaptive skills or infor-

mation from others through, say, imitation or conver-

sation, that model needs to explain how these oth-

ers reliably obtained the (correct) information and

where the information originated. If the blind lead

the blind, there is advantage to imitation. Conse-

quently, acceptable models should not employ shell

games, such as the venerable "adaptive knowledge

comes from the social world."

5. A candidate model must not propose the ex-

istence of complex capacities in 1 human psycho-

logical architecture unless these capacities solve or

solved adaptive (design-propagative) problems for

the individual. That is, social scientists should ex-

tremely uneasy about positing an improbably com-

plex structure in the system with the capacity to

serve nonbiological functional ends, unless that ca-

pacity is a by-product of functionality that evolved

to serve adaptive ends. Selection builds adaptive

functional organization; chance almost never builds

complex functional organization. So positing com-

plex designs that serve the larger social good, or that

complexly manipulate symbolic codes to spin webs

of meaning, or that cause one to maximize mone-

tary proﬁt, all violate the evolvability criterion unless

it can be shown that these are side effects of what

would have been adaptive functional organization in

the Pleistocer Similarly, one should not posit the ex-

istence of complex functional designs that evolved to

solve adaptive problems that emerged only very re-

cently. Complex functionality requires time to evolve

and, therefore, can arise only in response to long

standing adaptive problems (Dawkins, 1982, 1986;

Tooby & Cosmides, 1990a).

Over the course of their evolution, humans reg-

ularly needed to recognize object avoid predators,

avoid incest, avoid teratogens when pregnant, re-

pair nutritional deﬁciencies by dietary modiﬁcation,

judge distance, identify plant foods, capture animal

acquire grammar, attend to alarm cries, detect when

their children needed assistance, be motivated to

make that assistance, avoid contagious disease, ac-

quire a lexicon, I motivated to nurse, select con-

speciﬁcs as mates, select mates of the opposite sex,

select mates of high reproductive value, induce po-

tential mates to choose them, choose productive ac-

tivities, balance when walking, avoid being bitten by

venomous snake understand and make tools, avoid

needlessly enraging others, interpret social situa-

tions correctly, help relatives, decide which foraging

efforts have repaid the energy expenditure, perform

anticipatory motion computation, inhibit one’s mate

from conceiving children by another, deter aggres-

sion, maintain friendships, navigate, recognize faces,

recognize emotions, cooperate, and make effective

trade-offs among man of these activities, along with

a host of other tasks. To be a viable hypothesis about

human psychological architecture, the design pro-

posed must able to meet both solvability and evolv-

ability criteria: It must be able to solve the problems

that we observe modern humans routinely solving

and it must solve all the problems that were neces-

sary for humans to survive and reproduce in ances-

tral environments. No existing version of Standard

Model psychology can remotely begin to explain how

human perform these tasks.

Over the course of this chapter, we have touched

on how domain-speciﬁc mechanisms are empiri-

cally better supported than domain-general mecha-

nisms, on why domain-general mechanisms cannot

give rise to routinely observable behavioral perfor-

mances, on why domain-speciﬁc architectures are

usually more functional than domain-general archi-

tectures, and, especially, on why it is implausible or

impossible for predominantly content-independent,

domain-general computational systems t< perform

the tasks necessary for survival and reproduction in

natural environments The main arguments that we

have reviewed here (and elsewhere; see Cosmides &

Tooby, 1987, 1992; Tooby & Cosmides, 1990b) are

as follows:

1. In order to perform tasks successfully more of-

ten than chance, the architecture must be able to

discriminate successful performance from unsuccess-

ful performance. Because a domain-general archi-

tecture by deﬁnition has no built-in content-speciﬁc

rules for judging what counts as error and success

on different tasks, it must have a general rule. Un-

fortunately, there is no useable general cue or cri-

terion for success or failure that can apply across

domains. What counts as good performance for

one task (e.g., depth perception) is completely dif-

ferent from what counts as good performance for

other tasks (e.g., incest avoidance, immune regu-

lation, avoiding contagion, imitating, eating). The

only unifying element in discriminating success from

failure is whether an act promotes ﬁtness (design-

propagation). But the relative ﬁtness contribution

of a given decision cannot be used as a criterion for

learning or making choices because it is inherently

unobservable by the individual (for discussion, see

Cosmides & Tooby, 1987, 1992; Tooby & Cosmides,

1990b). Consequently, our evolved psychological ar-

chitecture needs substantial built-in content-speciﬁc

structure to discriminate adaptive success from fail-

ure. There needs to be at least as many different

domain-speciﬁc psychological adaptations as there

are evolutionarily recurrent functional tasks with dif-

ferent criteria for success.

2. As discussed at length, domain-general,

content-independent mechanisms are inefﬁcient,

handicapped, or inert compared to systems that also

include specialized techniques for solving particular

families of adaptive problems. A specialized mech-

anism can make use of the enduring relationships

present in the problem-domain or in the related fea-

tures of the world by reﬂecting these content-speciﬁc

relationships in its problem-solving structure. Such

mechanisms will be far more efﬁcient than general-

purpose mechanisms, which must expend time, en-

ergy, and risk learning these relationships through

"trial and possibly fatal error" (Shepard, 1987a).

3. Many problems that humans routinely solve are

simply not solvable by any known general problem-

solving strategy, as demonstrated by formal solvabil-

ity analyses on language acquisition ( e.g., Pinker,

1979, 1984, 1989, 1991; Pinker & Prince, 1988). We

think that the class of such problems is large and, as

discussed above, includes at a minimum all motiva-

tional problems.

4. Different adaptive problems are’ often incom-

mensurate. They cannot, in principle, be solved by

the same mechanism (Chomsky, 1980). To take a

simple example, the factors that make a food nutri-

tious are different from those that make a human a

good mate or a savannah a good habitat. As Sherry

and Schacter point out, "functional incompatibility

exists when an adaptation that serves one function

cannot, because of its specialized nature, effectively

serve other functions. The speciﬁc properties of the

adaptation that make it effective as a solution to

one problem also render it incompatible with the de-

mands of other problems" ( 1987, p. 439).

5. Many adaptive courses of action can be nei-

ther deduced nor learned by general criteria alone

because they depend on statistical relationships that

are unobservable to the relevant individual. For a

content-independent system to learn a relationship,

all parts of the relationship must be perceptually de-

tectable. This is frequently not the case. Natural se-

lection can "observe" relationships that exist between

a sensory cue, a decision rule, and a ﬁtness out-

come that is inherently unobservable to the individ-

ual making the decision (Tooby & Cosmides, 1990b),

as in the case of pregnancy sickness (Profet, 1988,

this volume) or the Westermarck incest avoidance

mechanism (Shepher, 1983; Wolf & Huang, 1980).

This is because natural selection does not work by

inference or computation: It took the real problem,

"ran the experiment," and retained those designs

whose information-processing procedures led over

thousand of generations to the best outcome. Natu-

ral selection, through incorporating content speciﬁc

decision rules, allows the organism to behave as if

it could see and be guide by relationships that are

perceptually undetectable and, hence, inherently un-

learnable by any general-purpose system.

6. As discussed, the more generally framed prob-

lems are, the more computational systems suffer

from combinatorial explosion, in which proliferating

alternatives choke decision and learning procedures,

bringing the system to a halt. If it were true that, a

Rindos ( 1986, p. 315) puts the central tenet of the

Standard Social Science Model, "the speciﬁcs that we

learn are in no sense predetermined by our genes,"

then we could learn nothing at all.

7. Everything a domain-general system can do

can be done as well or better by a system that also

permits domain-speciﬁc mechanisms because selec-

tion can incorporate any successful domain-general

strategies into an architecture without displacing

its existing repertoire of domain-speciﬁc problem-

solvers.

Without further belaboring the point, there is a

host of other reasons why content free, general-

purpose systems could not evolve, could not manage

their own reproduction, and would be grossly inef-

ﬁcient and easily outcompeted if they did. Equally

important, these arguments apply not simply to

the extreme limiting case of a completely content-

free, domain-general architecture but to all Standard

Model architectures, as conventionally presented.

The single criterion that any proposed human psy-

chological architecture must solve all the problems

necessary to cause reliable reproduction under nat-

ural conditions is decisive. When taken seriously

and considered carefully, it leads to the conclusion

that the human psychological architecture must be

far more frame-rich and permeated with content-

speciﬁc structure than mos1 researchers (including

ourselves) had ever suspected.

The Content-Speciﬁc Road to Adaptive Flexibil-

ity.

The ability to adjust behavior ﬂexibly and appro-

priately to meet the shifting demands of immediate

circumstances would, of course, be favored by selec-

tion, other things being equal. What organism would

not be better off if it could solve a broader array of

problems? Moreover, the psychologies of different

species do differ in the breadth of situations to which

they can respond appropriately, with humans act-

ing ﬂexibly to a degree that is zoologically unprece-

dented. Humans engage in elaborate improvised be-

haviors, from composing symphonies to piloting air-

craft to rice cultivation, which collectively indicate a

generality of achieved problem-solving that is truly

breathtaking. Although many human acts do not

successfully solve adaptive problems, enough do that

the human population has increased a thousandfold

in only a few thousand years. If general-purpose

mechanisms are so weak, how can the variability of

observed human behavior be reconciled with its level

of functionality?

As discussed above, there is little in content-

independent, domain-general strategies of problem-

solving that by themselves can account for functional

behavior, whether it is ﬂexible or not. In contrast,

specialized mechanisms can be very successful and

powerful problem-solvers, but they achieve this at

the price of addressing a narrower range of prob-

lems than a more general mechanism. If these were

the only two alternatives, organisms would be lim-

ited to being narrow successes or broad failures, and

the human case of broad adaptive ﬂexibility could

not be accounted for.

The solution to the paradox of how to create an

architecture that is at the same time both powerful

and more general is to bundle larger numbers of spe-

cialized mechanisms together so that in aggregate,

rather than individually, they address a larger range

of problems. Breadth is achieved not by abandon-

ing domain speciﬁc techniques but by adding more

of them to the system. By adding together a face

recognition module, a spatial relations module, a

rigid object mechanics module, a tool-use module, a

fear module, a social-exchange module, an emotion-

perception module, a kin-oriented motivation mod-

ule, an effort allocation and recalibration module,

a child-care module, a social-inference module, a

sexual-attraction module, a semantic-inference mod-

ule, a friendship module, a grammar acquisition

module, a communication-pragmatics module, a the-

ory of mind module, and so on, an architecture gains

a breadth of competences that allows it to solve a

wider and wider array of problems, coming to re-

semble, more and more, a human mind. The more

a system initially "knows" about the world and its

persistent characteristics, and the more evolution-

arily proven "skills" it starts out with, the more it

can learn, the more problems it can solve, the more

it can accomplish. In sharp contrast to the Stan-

dard Model, which views an absence of con- tent-

speciﬁc structure as a precondition for richly ﬂexi-

ble behavior, the analysis of what computational sys-

tems actually need to succeed suggests the opposite:

that the human capacity for adaptive ﬂexibility and

powerful problem-solving is so great precisely be-

cause of the number and the domain-speciﬁcity of

the mechanisms we have. Again, this converges on

William James’s argument that humans have more

"instincts" than other animals, not fewer (James,

1892; Symons, 1987).

Moreover, there are many reasons to think that

the number of function-general mechanisms and

function-speciﬁc mechanisms in an architecture are

not inversely related in a zero-sum relationship,

but are positively related (Rozin, 1976). Content-

specialized mechanisms dissect situations, thereby

creating a problem space rich with relevant rela-

tionships that content-independent mechanisms can

exploit (e.g., Cosmides & Tooby, under review;

Gigerenzer, Hoffrage & Kleinbolting, 1991; Shepard,

1987b). Thus, the more alternative content special-

ized mechanisms an architecture contains, the more

easily domain-general mechanisms can be applied to

the problem spaces they create without being para-

lyzed by combinatorial explosion. Although domain-

general mechanisms may be weak in isolation, they

can valuably broaden the problem solving range of

an architecture if they are embedded in a matrix

of adaptive specializations that can act as a guid-

ance system (Rozin, 1976). For example, humans

have powerful specialized social inference mecha-

nisms that reﬂect the contentful structure of human

metaculture, allowing humans to evaluate and in-

terpret others’ behaviors. This provides a founda-

tion for the human-speciﬁc ability to imitate others

(Galef, 1988; Meltzoff, 1988), greatly increasing the

range of situations to which they can respond appro-

priately.

Therefore, what is special about the human mind

is not that it gave up "instinct" in order to be-

come ﬂexible, but that it proliferated "instincts"-that

is, content-speciﬁc problem-solving specializations-

which allowed an expanding role for psycho- logi-

cal mechanisms that are (relatively) more function-

general. These are presently lumped into categories

with unilluminating labels such as "the capacity for

culture," "intelligence," "learning," and "rationality."

It is time for the social sciences to turn from a nearly

exclusive focus on these embedded, more function-

general mechanisms to a wider view that includes

the crucial, and largely neglected, superstructure of

evolved functional specializations. Equally, we need

to explore how the two classes of mechanisms are

interwoven so that their combined interactive prod-

uct is the zoologically unique yet evolutionarily pat-

terned breadth of functional behaviors.

EVOLUTIONARY PSYCHOLOGY AND THE GEN-

ERATION OF CULTURE

The Pluralistic Analysis of Human Culture

and Mental Organization Malinowski main-

tained that cultural facts are partly to be ex-

plained in psychological terms.

This view has often been met with skepti-

cism or even scorn, as if it were an easily

exposed naive fallacy. What I ﬁnd fallacious

are the arguments usually leveled against

this view. What I ﬁnd naive is the belief that

human mental abilities make culture possi-

ble and yet do not in any way determine its

content and organization.

-DAN SPERBER ( 1985, p. 73)

A large and rapidly growing body of research from

a diversity of disciplines has shown that the content-

independent psychology that provides the founda-

tion for the Standard Social Science Model is an im-

possible psychology. It could not have evolved; it re-

quires an incoherent developmental biology; it can-

not account for the observed problem-solving abil-

ities of humans or the functional dimension of hu-

man behavior; it cannot explain the recurrent pat-

terns and characteristic contents of human mental

life and behavior; it has repeatedly been empirically

falsiﬁed; and it cannot even explain how humans

learn their culture or their language. With the failure

of Standard Model psychology, and the emergence of

a domain-speciﬁc psychology, the remaining logic of

the Standard Social Science Model also collapses.

In this chapter, we have limited ourselves to ana-

lyzing some of the defects of the Standard Social Sci-

ence Model, concentrating on the untenability of the

psychology that forms the foundation for its theory

of culture. Along the way, we have touched on only

a handful of the changes that an evolutionary psy-

chological approach would introduce into the theo-

retical foundations of the social sciences. These and

the remarks that follow should not, however, be mis-

taken for a substantive discussion of , what a new

theory of culture that was based on modern biology,

psychology, and 1 anthropology would look like. Still

less should they be mistaken for a presentation of the

mutually consistent conceptual framework (what we

have been calling the Integrated Causal Model) that

emerges when the various biological, behavioral,

and social science ﬁelds are even partially integrated

and reconciled. Because our argument has been nar-

rowly focused on psychology, we have been unable

to review or discuss the many critical contributions

that have been made to this embryonic synthesis

from evolutionary biology, anthropology, neurobiol-

ogy, sociology, and many other ﬁelds. These must

be taken up elsewhere. In particular, readers should

be aware that the ideas underlying the Integrated

Causal Model are not original with us: They are

the collaborative product of hundreds of individual

scholars working in a diverse array of ﬁelds over the

last several decades! Indeed, the collaborative di-

mension of this new frame- work is key. The eclipse

of the Standard Model and the gradual emergence

of its replacement has resulted from researchers ex-

ploring the natural causal connections that integrate

separate ﬁelds (see, e.g., Barkow, 1989, on the im-

portance of making psychology consistent with biol-

ogy and anthropology consistent with psychology).

The research program we and others are advocating

is one of integration and consistency, not of psycho-

logical or biological reductionism. (See Atran, 1990;

Daly & Wilson, 1988; and Symons, 1979, for exam-

ples of how such integrative approaches can be ap-

plied to speciﬁc problems.)

What does the rise of domain-speciﬁc psychol-

ogy mean for theories of culture? By themselves,

psychological theories do not and cannot constitute

theories of culture. They only provide the foun-

dations for theories of culture. Humans live and

evolved in interacting networks that exhibit com-

plex population-level dynamics, and so theories and

analyses of population-level processes are necessary

components for any full understanding of human

phenomena. Nevertheless, increasing knowledge

about our evolved psychological architecture places

increasing constraints on admissible theories of cul-

ture. Although our knowledge is still very rudimen-

tary, it is already clear that future theories of cul-

ture will differ signiﬁcantly in a series of ways from

Standard Social Science Model theories. Most fun-

damentally, if each human embodies an evolved psy-

chological architecture that comes richly equipped

with content-imparting mechanisms, then the tradi-

tional concept of culture itself must be completely

rethought.

Culture has been the central concept of the Stan-

dard Social Science Model. According to its tenets,

culture is a unitary entity that expresses itself in

a trinity of aspects. (I) It is conceived as being

some kind of contingently variable informational

substance that is transmitted by one generation to

another within a group: Culture is what is socially

learned. (2) Because the individual mind is consid-

ered to be primarily a social product formed out of

the rudimentary infant mind, all or nearly all adult

mental organization and content is assumed to be

cultural in derivation and sub- stance: Culture is

what is contentful and organized in human mental

life and behavior. (3) Humans everywhere shows

striking patterns of local within-group similarity in

their behavior and thought, accompanied by signif-

icant intergroup differences. The existence of sepa-

rate streams of transmitted informational substance

is held to be the explanation for these group pat

terns: ’Cultures are these sets of similarities, and

inter- group or cross-location differences are called

cultural differences.

In the absence of a content-free psychology, how-

ever, this trinity breaks into separate pieces because

these three sets of phenomena can no longer be

equated. We have already sketched out why the

human mind must be permeated with content and

organization that does not originate in the social

world. This breaks apart any simple equivalence

between the ﬁrst two meanings of "culture." Nev-

ertheless, even for those who admit that the mind

has some content that is not socially supplied, the

distribution of human within-group similarities and

between-group differences remains the most persua-

sive element in the Standard Model analysis. These

salient differences are taken to conﬁrm that the so-

cially learned supplies most of the rich substance

of human life (see "The Standard Social Science

Model", pp. 24-34). Because Standard Model ad-

vocates believe that a constant-our universal evolved

architecture-cannot explain what varies, they can see

no explanation for "cultural differences" other than

differences in transmitted information.

Although this conclusion seems compelling, a sim-

ple thought experiment illustrates why it is un-

founded. Imagine that extraterrestrials replaced

each human being on earth with a state-of-the-art

compact disk jukebox that has thousands of songs in

its repertoire. Each jukebox is identical. Moreover,

each is equipped with a clock, an automated navi-

gational device that measures its latitude and longi-

tude, and a circuit that selects what song it will play

on the basis of its location, the time, and the date.

What our extraterrestrials would observe would

be the same kind of pattern of within- group sim-

ilarities and between-group differences observable

among humans: In Rio, every juke box would be

playing the same song, which would be different

from the song that every juke box was playing in Bei-

jing, and so on, around the world. Each juke box’s

"behavior" would be clearly and complexly patterned

because each had been equipped with the same large

repertoire of songs. Moreover, each jukebox’s be-

havior would change over time, because the song it

plays is a function of the date and time, as well as

of its location. Jukeboxes that were moved from lo-

cation to location would appear to adopt the local

songs, sequences, and "fashions. " Yet the genera-

tion of this distinctive, culture-like pattern involves

no social learning or transmission whatsoever. This

pattern is brought about because, like humans, the

juke boxes ( 1) share a universal, highly organized,

architecture, that (2) is designed to respond to in-

puts from the local situation (e.g., date, time, and

location).

All humans share a universal, highly organized

architecture that is richly endowed with contentful

mechanisms, and these mechanisms are designed

to respond to thousands of inputs from local situ-

ations. As a result, humans in groups can be ex-

pected to express, in response to local conditions, a

variety of organized within- group similarities that

are not caused by social learning or transmission.

Of course, these generated within-group similarities

will simultaneously lead to systematic differences be-

tween groups facing different conditions. To take a

single example, differences in attitudes toward shar-

ing between hunter-gatherer groups may be evoked

by ecological variables (for discussion, see Cosmides

& Tooby, this volume).

Thus, complex shared patterns that differ from

group to group may be evoked by circumstances or

may be produced by differential transmission. For

this reason, the general concept of "culture" in the

Standard Model sense is a conﬂation of evoked cul-

ture and transmitted culture (as well as of metacul-

ture and other components). Given that the mind

contains many mechanisms, we expect that both

transmitted and evoked factors will play complemen-

tary roles in the generation of differentiated local

cultures. The operation of a richly responsive psy-

chology, plus the ability to socially "learn," can jointly

explain far more about "culture" and cultural change

than either can alone. For example, when mem-

bers of a group face new and challenging circum-

stances (drought, war, migration, abundance), this

may activate a common set of functionally organized

domain-speciﬁc mechanisms, evoking anew set of at-

titudes and goals. The newly evoked psychological

states will make certain new ideas appealing, caus-

ing them to spread by transmission, and certain old

ideas unappealing, causing them to be discarded. In

contrast, the Standard Model "do what your parents

did" concept of culture is not a principle that can

explain much about why cultural elements change,

where new ones come from, why they spread, or

why certain complex patterns ( e.g., pastoralist com-

monalities) recur in widely separated cultures. Of

course, many anthropologists implicitly recognize

these points, but they need to make the links be-

tween the cultural processes they study and the

underlying evolved content- organizing psychology

they are assuming explicit. For example, economic

and ecological anthropology, to be coherent, neces-

sarily assume underlying content-specialized psycho-

logical mechanisms that forge relationships between

environmental and economic variables and human

thought and action.

It is especially important for post-Standard Model

researchers to recognize that the environmental fac-

tors that cause contentful mental and behavioral or-

ganization to be expressed are not necessarily the

processes that constructed that organization. In the

case of the jukebox, it would be a mistake to at-

tribute the organized content manifest in the mu-

sic to the environmental stimuli (i.e., the location,

date, and time) that caused one song to be played

rather than another. The stimuli did not compose the

music; they merely caused it to be expressed. Simi-

larly, our psychological architectures come equipped

with evolved contentful organization, which can re-

main latent or become activated depending on cir-

cumstances and which may vary in its expression

according to procedures embodying any degree of

complexity. Because our psycho- logical architecture

is complexly responsive, the Standard Model prac-

tice of equating the variable with the learned is a

simple non sequitur. The claim that some phenom-

ena are "socially constructed" only means that the

social environment provided some of the inputs used

by the psychological mechanisms of the individuals

involved.

In short, observations of patterns of similarities

and differences do not establish that the substance

of human life is created by social learning. In any

speciﬁc case, we need to map our evolved psycho-

logical architecture to know which elements (if any)

are provided by transmission, which by the rest of

the environment, which by the architecture, and

how all these elements causally interact to produce

the phenomenon in question. Admittedly, the juke

box thought experiment is an unrealistically extreme

case in which a complex, functionally organized,

content-sensitive architecture internalizes no trans-

mitted informational input other than an environ-

mental trigger. But this case is simply the mirror

image of the SSSM’s extreme view of the human

mind as a content-free architecture where everything

is provided by the internalization of transmitted in-

put. Our central point is that in any particular do-

main of human activity, the programming that gives

our architecture its ability to contingently respond

to the environment mayor may not be designed to

take transmitted representations as input. If it does,

it may mix in content derived from its own structure

and process the resulting representations in complex

and transformative ways. The trinity of cultural phe-

nomena can no longer be equated with ’one another.

Our complex content-speciﬁc psychological architec-

ture participates in the often distinct processes of

generating mental content, generating local similar-

ities and between-group differences, and generating

what is "transmitted." Indeed, it also participates in

the complex process of internalizing what others are

"transmitting."

Inferential Reconstruction and Cultural Epi-

demiology

The Standard Social Science Model has been very

effective in promulgating the unity of the trinity. The

socially learned, the set of within-group common-

alities and between-group differences, and the con-

tentful organization of human mental and social life

have been so thoroughly conﬂated that it is difﬁcult

to speak about human phenomena without using the

word culture. For this reason, we will use culture

to refer to any mental, behavioral, or material com-

monalities shared across individuals, from those that

are shared across the entire species down to the lim-

iting case of those shared only by a dyad, regardless

of why these commonalities exist. When the causes

of the commonality can be identiﬁed, we will use a

qualiﬁer, such as "evoked."

So things that are cultural in the sense of being

organized, contentful, and shared among individu-

als may be explained in a number of different ways.

Within-group commonalities may have been evoked

by common circumstances impacting universal archi-

tectures. An even larger proportion of organized,

contentful, and shared phenomena may be explained

as the expression of our universal psychological and

physiological architectures in interaction with the re-

current structure of the social or non- social world-

what we earlier called metaculture. Metaculture

includes a huge range of psychological and behav-

ioral phenomena that under Standard Model analy-

ses have been invisible or misclassiﬁed (D. E. Brown,

1991). Because the Standard Model attributed ev-

erything that was contentful and recurrent to some

form of social learning, it misinterpreted phenomena

such as anger upon deliberate injury, grief at a loss,

the belief that others have minds, treating species as

natural kinds, social cognition about reciprocation,

or the search for food when hungry as socially man-

ufactured products.

Nevertheless, after the evoked and the metacul-

tural have been excluded, there still remains a large

residual category of representations or regulatory el-

ements that reap- pear in chains from individual to

individual-"culture" in the classic sense. In giving up

the Standard Social Science Model, we are not aban-

doning the classic concept of culture. Instead, we are

attempting to explain what evolved psychological

mechanisms cause it to exist. That way we can get

a clearer causal understanding of how psycho- log-

ical mechanisms and populational processes shape

its content, and thereby restrict its explanatory role

in social theory to the phenomena that it actually

causes.

This subset of cultural phenomena is restricted to

(1) those representations or regulatory elements that

exist originally in at least one mind that (2) come to

exist in other minds because (3) observation and in-

teraction between the source and the observer cause

inferential mechanisms in the observer to recreate

the representations or regulatory elements in his or

her own psychological architecture. In this case,

the representations and elements inferred are con-

tingent: They could be otherwise and, in other hu-

man minds, they commonly are otherwise. Rather

than calling this class of representations "transmit-

ted" culture, we prefer terms such as reconstructed

culture, adopted culture, or epidemiological culture.

The use of the word "transmission" implies that the

primary causal process is located in the individu-

als from whom the representations are derived. In

contrast, an evolutionary psychological perspective

emphasizes the primacy of the psychological mech-

anisms in the learner that, given observations of

the social world, inferentially reconstruct some of

the representations existing in the minds of the ob-

served. Other people are usually just going about

their business as they are observed, and are not nec-

essarily intentionally "transmitting" anything.

More precisely, an observer (who, for expository

simplicity, we will call the "child") witnesses some ﬁ-

nite sample of behavior by others (e.g., public repre-

sentations, such as utterances or other communica-

tive acts; people going about their affairs; people re-

sponding to the child’s behavior). The task of the

mechanisms in the child is to (1) reconstruct within

themselves on the basis of these observations a set of

representations or regulatory elements that (2) are

similar enough to those present in the humans she

lives among so that (3) the behavior her mechanisms

generate can be adaptively coordinated with other

people and her habitat. Thus, the problem of learn-

ing "culture" lies in deducing the hidden represen-

tations and regulatory elements embedded in oth-

ers’ minds that are responsible for generating their

behavior. To the extent that the child’s mechanisms

make mistakes-and mistakes are endemic-and recon-

struct the wrong underlying representations and reg-

ulatory elements, she will not be able to predict other

people’s behavior, interpret their transactions with

one another in the world, imitate them, communi-

cate with them, cooperate with them, help them, or

even anticipate or avoid their hostile and exploitive

actions.

Why did ancestral hominid foragers evolve mecha-

nisms that allowed them to reconstruct the represen-

tations present in the minds of those around them?

Leaving aside the question of their costs and limita-

tions, the advantage of such mechanisms is straight-

forward. Information about adaptive courses of ac-

tion in local conditions is difﬁcult and costly to ob-

tain by individual experience alone. Those who have

preceded an individual in a habitat and social en-

vironment have built up in their minds a rich store

of useful information. The existence of such infor-

mation in other minds selected for specialized psy-

chological adaptations that were able to use social

observations to reconstruct some of this informa-

tion within one’s own mind (e.g., Boyd & Richerson,

1985; Tooby & DeVore, 1987). By such inferential re-

construction, one individual was able to proﬁt from

deducing what another already knew. When such in-

ferential reconstruction becomes common enough in

a group, and some representations begin to be stably

re-created in sequential chains of individuals across

generations, then the structure of events begins to

warrant being called "cultural. "

As discussed earlier, this task of reconstruction

would be unsolvable if the child did not come

equipped with a rich battery of domain-speciﬁc in-

ferential mechanisms, a faculty of social cognition,

a large set of frames about humans and the world

drawn from the common stock of human metacul-

ture, and other specialized psychological adaptations

designed to solve the problems involved in this task

(see, e.g., Boyer, 1990; Sperber, 1985, 1990; Sperber

& Wilson, 1986; Tooby & Cosmides, 1989a). Con-

sequently, epidemiological culture is also shaped by

the details of our evolved psycho- logical organiza-

tion. Thus, there is no radical discontinuity inher-

ent in the evolution of "culture" that removes hu-

mans into an autonomous realm. Mechanisms de-

signed for such inferential reconstruction evolved

within a pre-existing complex psychological archi-

tecture and depended on this encompassing array

of content-structuring mechanisms to successfully

interpret observations, reconstruct representations,

modify behavior, and so on. Solving these, infer-

ential problems is not computationally trivial, and

other species, with a few possible minor exceptions,

are not equipped to per- form this task to any signif-

icant degree (Galef, 1988).

Moreover, outside of contexts of competition,

knowledge is not usually devalued by being shared.

Consequently, to the substantial extent that individu-

als in a hunter- gatherer group had interests in com-

mon and had already evolved mechanisms for infer-

ential reconstruction, selection would have favored

the evolution of mechanisms that facilitated others’

inferences about one’s own knowledge (as, for ex-

ample, by communicating or teaching). The mutual

sharing of valuable knowledge and discoveries has

a dramatic effect on the usefulness of mechanisms

that attempt to adaptively adjust behavior to match

local conditions (Boyd & Richerson, 1985; Tooby

& DeVore, 1987). Because of combinatorial explo-

sion, knowledge of successful local techniques is pre-

cious and hard to discover, but relatively cheap to

share (once again, ignoring the cost of the psycho-

logical mechanisms that facilitate or perform such

sequential reconstruction). Within limits, this cre-

ates economies of scale: The greater the number of

individuals who participate in the system of knowl-

edge sharing, ( 1) the larger the avail- able pool

of knowledge will be, (2) the more each individ-

ual can derive from the pool, (3) the more advanta-

geous reconstructive adaptations will be, and (4) the

more it would pay to evolve knowledge-dependent

mechanisms that could exploit this set of local repre-

sentations to improvise solutions to local problems.

This collaborative information-driven approach to

the adaptive regulation of behavior can be thought of

as the "cognitive niche" (Tooby & DeVore, 1987). The

mutual beneﬁt of such knowledge sharing led to the

co-evolution of sets of adaptations, such as language,

elaborated communicative emotional displays, and

pedagogy that coordinate specialized processes of in-

ferential reconstruction with the specialized produc-

tion of behavior. designed to facilitate such recon-

struction (e.g., Ekman, 1984; Freyd, 1983; Fridlund

in press; Pinker & Bloom, this volume; Premack, in

prep.; Sperber & Wilson, 1986).

Because reconstructive inferences are often erro-

neous and what others "know" is often of dubious

quality or irrelevant, such inferential processes could

not have evolved without adaptations that assessed

to some degree the value of such reconstructed

knowledge and how it ﬁts in with knowledge derived

from other sources. If a representation is easy to suc-

cessfully reconstruct and is evaluated positively, then

it will tend to spread through inter-individual chains

of inference, becoming widely shared. If it is difﬁcult

to reconstruct or evaluated as not valuable, it will

have only a restricted distribution or will disappear

(Sperber, 1985, 1990; Sperber & Wilson, 1986). This

evaluation process gives sequentially reconstructed

culture its well-known, if partial, parallels to nat-

ural selection acting on genes; that is, the selec-

tive retention and accumulation of favored variants

over time (e.g., Barkow, 1989; Boyd & Richer- son,

1985; Campbell, 1965, 1975; Cavalli Sforza & Feld-

man, 1981; Dawkins, 1976; Durham, 1991; Lums-

den & Wilson, 1981). Moreover, these psychological

mechanisms endow sequentially reconstructed cul-

ture with its epidemiological character, as a dynam-

ically changing distribution of elements among indi-

viduals living in populations over time. As Sperber

says, "Cultural phenomena are ecological patterns of

psychological phenomena. They do not pertain to

an autonomous level of reality, as antireductionists

would have it, nor do they merely belong to psychol-

ogy as reductionists would have it" ( 1985, p. 76).

The more widely shared an element is, the more

people are inclined to call it "cultural," but there

is no natural dividing point along a continuum of

something shared between two individuals to some-

thing shared through inferential reconstruction by

the entire human species (Sperber, 1985, 1990). The

Standard Model practice of framing "cultures" as sets

of representations homogeneously shared by nearly

all members of discrete and bounded "groups" does

not capture the richness of the ecological distribu-

tion of these psychological elements, which cross-cut

each other in a bewildering variety of fractal pat-

terns. Language boundaries do not correspond to

subsistence practice boundaries, which do not corre-

spond to political boundaries or to the distribution

of rituals (for discussion, see Campbell & LeVine,

1972). Within groups, representations occur with

all kinds of different frequencies, from beliefs passed

across generations by unique dyads, such as shaman-

istic knowledge or mother-daughter advice, to be-

liefs shared by most or all members of the group.

The belief that sequentially reconstructed repre-

sentations exist primarily in bounded cells called

"cultures" derives primarily from the distribution of

language boundaries, which do happen to be dis-

tributed more or less in this fashion. As Pinker and

Bloom (this volume) point out, communication pro-

tocols can be arbitrary, but must be shared between

sender and receiver to be functional. The beneﬁt

of learning an arbitrary linguistic element is propor-

tional to how widely it is distributed among a set

of interacting individuals. Therefore, well-designed

language acquisition mechanisms distributed among

a local set of individuals will tend to converge on a

relatively homogeneous set of elements: It is use-

ful for all local individuals to know the same lo-

cal language. Although there are other reasons

why reconstructed elements may show sharp coor-

dinated boundaries (e.g., ethnocentrism, common

inheritance, sharp habitat boundaries, geographical

barriers, and so on), most classes of representations

or regulatory elements dynamically distribute them-

selves according to very different patterns, and it

is probably more accurate to think of humanity as

a single interacting population tied together by se-

quences of reconstructive inference than as a collec-

tion of discrete groups with separate bounded "cul-

tures."

Finally, the reconstruction of regulatory elements

and representations in a psycho- logical architecture

should not be thought of as a homogeneous process.

Given that our minds have a large set of domain-

speciﬁc mechanisms, it seems likely that different

mechanisms would be selected to exploit social ob-

servations in different ways and have quite distinct

procedures for acquiring, interpreting, and using in-

formation derived from the social world. Certainly,

the language acquisition device appears to have its

own special properties (e.g., Pinker, 1989), and

many other domains appear to follow their own spe-

cial rules (Carey & Gelman, 1991). It seems unlikely

in the extreme that the different modules underly-

ing mate preferences (Symons, 1979; Buss, in prep.),

food preferences (e.g., Galef, 1990), display rules for

emotional expression (Ekman, 1982), fears (Cook et

al., 1986) and so on, process social observations ac-

cording to a single unitary process. Moreover, to the

extent that there may exist a large, potentially in-

teracting store of representations in the mind (see,

e.g., Fodor 1983), nothing in the psychological ar-

chitecture necessarily segregates off representations

derived through "epidemiological culture" from rep-

resentations and regulatory elements derived from

other sources.

This brief sketch suggests a few of the features fu-

ture theories of culture may incorporate, once the

Standard Model concept of learning is discarded.

These are organized by two themes. First, what is

presently attributed to "culture" will come to be plu-

ralistically explained as metaculture, evoked culture,

epidemiological culture, and individual mental con-

tents that are internally generated and not derived

through inferential reconstruction (see table below).

Second, with the fall of content-independent learn-

ing, the socially constructed wall that separates psy-

chology and anthropology (as well as other ﬁelds)

will disappear. The heterogeneous mechanisms com-

prising our evolved psychological architecture partic-

ipate inextricably in all cultural and social phenom-

ena and, because they are content-specialized, they

impart some contentful patterning to them. Indeed,

models of psychological mechanisms, such as social

exchange, maternal attachment, sexual attraction,

sexual jealousy, the categorization of living kinds,

and so on, are the building blocks out of which fu-

ture theories of culture will, in part, be built (Sper-

ber, 1990; Tooby & Cosmides, 1989a). By no means

do we deny or minimize the existence of emergent

phenomena, such as institutions, or the fact that

population-level processes alter the epidemiological

distribution of cultural contents over time. The point

is simply that cultural and social phenomena can

never be fully divorced from the structure of the hu-

man psychological architecture or understood with-

out reference to its design.

The Twilight of Learning as a Social Science Expla-

nation

Advocates of the Standard Social Science Model

have believed for nearly a century that they have

a solid explanation for how the social world in-

serts organization into the psychology of the devel-

oping individual. They maintain that structure en-

ters from the social (and physical) world by the pro-

cess of "Iearning"-individuals "learn" their language,

they "learn " their culture, they "learn " to walk, and

so on. All problems- whether they are long-enduring

adaptive problems or evolutionarily unprecedented

problems-are solved by "learning." In the intellectual

communities dominated by the SSSM, learning has

been thought to be a powerful explanation for how

certain things come about, an explanation that is

taken to refer to a well-understood and well- spec-

iﬁed general process that someone (i.e., the psycho-

logical community) has documented. For this rea-

son, "learning," and such common companion con-

cepts as "culture," "rationality," and "intelligence," is

frequently invoked as an alternative explanation to

so-called "biological" explanations (e.g., sexual jeal-

ousy did not evolve, it is learned from culture; one

doesn’t need to explain how humans engage in social

exchange: They simply used their "reason" or "intel-

ligence").

Of course, as most cognitive scientists know (and

all should), "learning"-like "culture," "rationality,"

and "intelligence"-is not an explanation for anything,

but is rather a phenomenon that itself requires ex-

planation (Cosmides & Tooby, 1987; Tooby & Cos-

mides, 1990b). In fact, the concept of "learning" has,

for the social sciences, served the same function that

the concept of "protoplasm" did for so long in bi-

ology. For decades, biologists could see that living

things were very different from nonliving things, in

that a host of very useful things happened inside of

living things that did not occur inside of the nonliv-

ing (growth, the manufacture of complex chemicals,

the assembly of useful structures, tissue differentia-

tion, energy production, and so on). They had no

idea what causal sequences brought these useful re-

sults about. They reiﬁed this unknown functional-

ity, imagining it to be a real substance, and called it

"protoplasm," believing it to be the stuff that life was

made of. It was a name given to a mystery, which

was then used as an explanation for the functional

results that remained in genuine need of explana-

tion. Of course, the concept of protoplasm eventu-

ally disappeared when molecular biologists began to

determine the actual causal sequences by which the

functional business of life was transacted. "Proto-

plasm " turned out to be a heterogeneous collection

of incredibly intricate functionally organized struc-

tures and processes-a set of evolved adaptations, in

the form of microscopic molecular machinery such

as mitochondria, chloroplasts, the Krebs cycle, DNA

transcription, RNA translation, and so on.

Similarly, human minds do a host of singularly

useful things, by which they coordinate themselves

with things in the world: They develop skill in

the local community’s language; upon exposure to

events they change behavior in impressively func-

tional ways; they reconstruct in themselves knowl-

edge derived from others; they adopt the practices of

others around them; and so on. Psychologists did not

know what causal sequences brought these useful re-

sults about. They reiﬁed this unknown functional-

ity, imagining it to be a unitary process, and called

it "leaming." "Leaming" is a name given to the un-

known agent imagined to cause a large and hetero-

geneous set of functional outcomes. This name was

(and is) then used as an explanation for results that

remained in genuine need of explanation. We expect

that the concept of learning will eventually disap-

pear as cognitive psychologists and other researchers

make progress in determining the actual causal se-

quences by which the functional business of the mind

is transacted. Under closer inspection, "learning" is

turning out to be a diverse set of processes caused

by a series of incredibly intricate, functionally orga-

nized cognitive adaptations, implemented in neuro-

biological machinery (see, e.g., Carey & Gel- man,

1991; Gallistel, 1990; Pinker 1989, 1991; Real, 1991

). With slight qualiﬁcations about the exact contexts

of usage, similar things could be said for "culture,"

"intelligence," and "rationality." The replacement of

the concept of protoplasm with a real understand-

ing of the vast, hidden, underlying worlds of molec-

ular causality has trans- formed our understanding

of the world in completely unexpected ways, and we

can only anticipate that the same will happen when

"learning" is replaced with knowledge.

Evolutionary Social Psychology

Chapter

Full-text available

Jul 2024

Mapping the Science against the Technology

Conference Paper

Nov 2023

Simon Reay Atkinson

Evolution is both incremental and by step change. Science and Technology follow different entangled harmonic time constants that combine and separate about conjoined surge / stagnate, and invention / innovation logistic curves. These time constants are used to understand how the Science at different discovery levels may interact with Technology, at different readiness levels-and vice versa. This, the first of three papers, examines the adoption of Technology Readiness Level (TRL) as a means of assessing against the science. This gives rise to the adoption of a Scientific Discovery Level (SDL) for subjectively situating TRL. The SDL-TRL Instrument identified allows for objective assessments to be made for situating science and technology. To arrive at the SDL-TRL instrument, historical analysis of previous scientific ages; associated technologies; and human generations is undertaken. This leads into analysis of future scientific ages, with ethical and moral implications for humanity. The next paper will examine uncertainty graph modelling for aiding scientific and technology assessments. The final paper will examine means for translating and operationalising science and technologies into successful capabilities.

A Textbook About Evolution, for the General Reader?

Article

Full-text available

Nov 2012

A review of the book The tangled bank: an introduction to evolution by Carl Zimmer (2010).

Evolutionary Approaches to Stereotyping and Prejudice

Chapter

Oct 2016

The Cambridge Handbook of the Psychology of Prejudice aims to answer the questions: why is prejudice so persistent? How does it affect people exposed to it? And what can we do about it? Providing a comprehensive examination of prejudice from its evolutionary beginnings and environmental influences through to its manifestations and consequences, this handbook is an essential resource for scholars and students who are passionate about understanding prejudice, social change, collective action, and prejudice reduction. Featuring cutting-edge research from top scholars in the field, the chapters provide an overview of psychological models of prejudice; investigate prejudice in specific domains such as race, religion, gender, and appearance; and develop explicit, evidence-based strategies for disrupting the processes that produce and maintain prejudice. This handbook challenges researchers and readers to move beyond their comfort zone, and sets the agenda for future avenues of research, policy, and intervention.

Massive modularity? The relationship between context-relevance, information encapsulation and functional specialization

Article

Full-text available

Jan 2010

George Varvatsoulias

In this article, I discuss the debate between domain-specificity and content-generality in regard to the human mind. My main objective is to argue that the human mind can both be understood as a content-dependent machinery, as well as a general-purpose system that encapsulates information and manifests it accordingly. In evolutionary psychology a strong case is made for the mind"s domain-specific architecture but assumptions of domain-general importance could also be taken into account. In evolutionary terms, it is argued that the mind is composed of content-dependent and specialized functions that have evolved in order to help humans deal with adaptive problems. The mind"s evolved functionality of domain-specific systems has helped humans not only to survive but also to be creative in their relationships to others and the environment. However, advocates of behaviourist models and learning theories have proposed that the mind is a content-independent device which encapsulates information of a general nature. In this paper, domain specificity will be explained in terms of functional specialization, and content-generality in terms of informational encapsulation. My main aim will be to argue that both assumptions can be viewed in context-relevance to each other.

Cross-societal variation in norm enforcement systems

Article

Full-text available

Jan 2024

Across human societies, people are sometimes willing to punish norm violators. Such punishment can take the form of revenge from victims, seemingly altruistic intervention from third parties, or legitimized sanctioning from institutional representatives. Although prior work has documented cross-cultural regularities in norm enforcement, substantial variation exists in the prevalence and forms of punishment across societies. Such cross-societal variation may arise from universal psychological mechanisms responding to different socio-ecological conditions, or from cultural evolutionary processes, resulting in different norm enforcement systems. To date, empirical evidence from comparative studies across diverse societies has remained disconnected, owing to a lack of interdisciplinary integration and a prevalent tendency of empirical studies to focus on different underpinnings of variation in norm enforcement. To provide a more complete view of the shared and unique aspects of punishment across societies, we review prior research in anthropology, economics and psychology, and take a first step towards integrating the plethora of socio-ecological and cultural factors proposed to explain cross-societal variation in norm enforcement. We conclude by discussing how future cross-societal research can use diverse methodologies to illuminate key questions on the domain-specificity of punishment, the diversity of tactics supporting social norms, and their role in processes of norm change. This article is part of the theme issue ‘Social norm change: drivers and consequences’.

Theorizing the Anthropology of Belief: Magic, Conspiracies, and Misinformation

Book

Jan 2024

This book explores both scientific and humanistic theoretical traditions in anthropology through the lens of ontology. The first part of the book examines different methods for generating valid anthropological knowledge and proposes a shift in current consensus. Drawing on Western scholars of antiquity and the medieval period and moving away from 20th-century theorists, it argues that we must first make ontological assumptions about the kinds of things that can exist (or not) before we can then develop epistemologies that study those kinds of things. The book goes on to apply the ontology-first theory to a set of case studies in modern day conspiracy theories, misinformation, and magical thinking. It asserts that we need to move away from unneeded metaphysical assumptions of conspiracy theories being misinformation and argues that reconstructing particular historical events can be a fruitful zone for application of quantitative methods to humanistic questions. Theorizing the Anthropology of Belief is an excellent supplementary suitable for upper-level undergraduate and graduate courses in anthropological theory.

Prospects in the field of learning and individual differences: Examining the past to forecast the future using bibliometrics

Article

Full-text available

Dec 2023
LEARN INDIVID DIFFER

Over two centuries, research has delved into individual differences in learning across educational and professional contexts. This commentary conducts a bibliometric analysis of 6556 articles, identifying key research keywords, topics and themes, and their historical evolution. The findings revealed a longstanding emphasis on educational psychology, particularly motivation and achievement, rather than cross-curricular competencies and learner’s well-being and socio-economic background. Notably, self-regulated learning (SRL) emerged as an overarching research subject in terms of motivation and achievement, but, surprisingly, not (meta)cognition. Prospects for the field build on cross-disciplinary research, theoretical refinement, and methodological advances. Further, the field is expected to maintain academic rigor, address diversity among learners, foster global collaboration, and focus on underprivileged populations.

The Adaptive Value of Women’s Orgasm

Chapter

Jul 2022

The interface of sexual behavior and evolutionary psychology is a rapidly growing domain, rich in psychological theories and data as well as controversies and applications. With nearly eighty chapters by leading researchers from around the world, and combining theoretical and empirical perspectives, The Cambridge Handbook of Evolutionary Perspectives on Sexual Psychology is the most comprehensive and up-to-date reference work in the field. Providing a broad yet in-depth overview of the various evolutionary principles that influence all types of sexual behaviors, the handbook takes an inclusive approach that draws on a number of disciplines and covers nonhuman and human psychology. It is an essential resource for both established researchers and students in psychology, biology, anthropology, medicine, and criminology, among other fields. Volume 3: Female Sexual Adaptations addresses theory and research focused on sexual adaptations in human females.

Is evolutionary psychology a scientific revolution? A bibliometric response to Buss (2020)

Preprint

Full-text available

Nov 2023

Andrea Zagaria

In recent times, David Buss (2020) has claimed that Evolutionary Psychology (EP) represents a genuine "scientific revolution," drawing parallels with the Kuhnian concept. This paper aims to empirically examine Buss's assertion by conducting a bibliometric analysis of the prevalence of the evolutionary approach in Psychology. EP (broadly defined) is juxtaposed with its classical counterpart, the Standard Social Science Model (SSSM). According to Kuhn's framework, a discipline undergoing a scientific revolution evolves into the dominant school of thought, eclipsing all other theories. Contrary to Buss's contention, the analysis uncovers that the SSSM enjoys significantly greater prominence than EP and is advancing at a swifter pace. The data also indicates a “cultural evolutionary” approach is somewhat lacking (i.e. cross-fertilization between EP and SSSM is not evident). Despite sharing the aspiration for an evolutionary revolution in psychology, it is contended that, for this purpose, a prudent approach involves recognizing the current status of affairs, envisioning realistic change, and building a more methodologically diverse research community.

The genetical evolution of social behaviour. I

Article

Aug 1964

William D. Hamilton

A genetical mathematical model is described which allows for interactions between relatives on one another's fitness. Making use of Wright's Coefficient of Relationship as the measure of the proportion of replica genes in a relative, a quantity is found which incorporates the maximizing property of Darwinian fitness. This quantity is named “inclusive fitness”. Species following the model should tend to evolve behaviour such that each organism appears to be attempting to maximize its inclusive fitness. This implies a limited restraint on selfish competitive behaviour and possibility of limited self-sacrifices. Special cases of the model are used to show (a) that selection in the social situations newly covered tends to be slower than classical selection, (b) how in populations of rather non-dispersive organisms the model may apply to genes affecting dispersion, and (c) how it may apply approximately to competition between relatives, for example, within sibships. Some artificialities of the model are discussed.

The Psychological Foundations of Culture

Recommended publications

Psychologische Grundlagen der Gesundheitswissenschaften

Comment on Trendler’s (2019) “Conjoint measurement undone”

McCauley, Robert N., and Thomas Lawson (eds.). 2002. Bringing ritual to mind. Psychological foundati...

Media and Religion: Foundations of an Emerging Field