Content uploaded by Craig Joseph
Author content
All content in this area was uploaded by Craig Joseph on Aug 17, 2015
Content may be subject to copyright.
Content uploaded by Craig Joseph
Author content
All content in this area was uploaded by Craig Joseph on May 17, 2015
Content may be subject to copyright.
367
UNCORRECTED PROOF
1 Introduction
Morality is one of the few topics in academe endowed with its own protective spell.
A biologist is not blinded by her biological nature to the workings of biology. An
economist is not confused by his own economic activity when he tries to understand
the workings of markets.
1
But students of morality are often biased by their own
moral commitments. Morality is so contested and so passionately valued that it is
often diffi cult to set aside one’s humanity and study morality in a clinically detached
way. One problem is that the psychological study of morality, like psychology itself
(Redding, 2001), has been dominated by politically liberal researchers (who include
us). The lack of moral and political diversity among researchers has led to an inap-
propriate narrowing of the moral domain to issues of harm/care and fairness/reci-
procity/justice (Haidt and Graham, 2007). Morality in most cultures (and for social
conservatives in Western cultures) is in fact much broader, including issues of in-
group/loyalty, authority/respect, and purity/sanctity (Haidt and Graham, 2007, and
in press).
This chapter is about how morality might be partially innate, by which we sim-
ply mean organized, to some extent, in advance of experience (Marcus, 2004). We
begin by arguing for a broader conception of morality and suggesting that most of the
discussion of innateness to date has not been about morality per se; it has been about
whether the psychology of harm and fairness is innate. Once we have made our case
that morality involves fi ve domains, not two, we turn our attention to the ways in
which this diverse collection of motives and concepts might be innate. We consider
19
jonathan haidt and craig joseph
The Moral Mind
How Five Sets of Innate Intuitions
Guide the Development of Many
Culture-Specifi c Virtues, and
Perhaps Even Modules
We thank Peter Carruthers, Jesse Graham, Angeline Lillard, Shige Oishi, and Dan Sperber for helpful
comments on the fi rst draft of this chapter.
1. Biologists and economists may be blindly loyal to academic theories, but we suggest that these biases
are themselves often manifestations of moral commitments (e.g., the polemics of Steven Jay Gould).
367
Carruthers Chap19.indd 367Carruthers Chap19.indd 367 8/30/2007 12:08:38 PM8/30/2007 12:08:38 PM
368 Culture, Motivation, and Morality
UNCORRECTED PROOF
fi ve hypotheses about the origins of moral knowledge and value, and we endorse one
of them (a form of fl exible and generative modularity) as being the best candidate.
Next, we develop this version of modular morality by describing how the innately
specifi ed “fi rst draft” of the moral mind gets modifi ed during development.
Specifi cally, we link our view of moral innateness with virtue theory, an ancient
approach that is consistent with the insights of many modern perspectives. In doing
so, we are extending our exploration of the possibilities of virtue theory, which we
began in an earlier article (Haidt and Joseph, 2004). We are not proposing that virtue
ethics is the best normative moral theory. We speak only descriptively, and we note
that there is a growing rapport between philosophical writings on virtue and emo-
tions, empirical research on moral functioning, and cognitive science, a rapport that
suggests that virtue theory may yield deep insights into the architecture of human
social and moral cognition.
In the fi nal section, we discuss the importance of narrativity in moral function-
ing. In some respects, this is another corrective to what we see as an overemphasis
on deductive and calculative conceptions of value and rationality among both phi-
losophers and psychologists. We attempt to show, in this last section, that a narrative
approach to morality fi ts well with the nativist “fi ve foundations” view we developed
in the fi rst part of the chapter, and also helps to explain how the intuitive, evolved
foundations of morality are elaborated by cultural activity into the complex, diverse
moral functioning that mature human beings display.
2 Morality Is Many Things
Soon after human beings began to write, they began to write about morality. Many
of the earliest moral texts are largely lists of laws and prohibitions (e.g., the Code of
Hammurabi; the older parts of the Old Testament). But as the Axial Age progressed
(800 BCE-200 BCE), many cultures East and West began to develop a more sophis-
ticated psychology of the virtues. We fi nd explicit discussions of virtues, often in the
context of stories about role models who exemplifi ed them (e.g., Homer and Aesop
in Greece; the Mahabharata in India). An important feature of this approach is that
moral education is accomplished by shaping emotions and intuitions, rather than by
dictating explicit rationales or principles. The wisdom of Confucius and of Buddha,
for example, comes down to us as lists of aphorisms and metaphors that produce
fl ashes of intuitive understanding.
A second feature of these virtue-based approaches is that they emphasize prac-
tice and habit rather than propositional knowledge and reasoning. Buddha urged
his disciples to follow the Eightfold Noble Path—a set of daily practices—to reach
moral and psychological perfection. Aristotle and Confucius both compared the
development of virtue to the slow practice needed to develop what we now call
“virtuosity” on a musical instrument (Aristotle, 1941; Hansen, 1991).
For the ancients there were many virtues, covering most aspects of human activ-
ity. Virtues were excellences that people were expected to cultivate in themselves,
depending on their social roles and stations in life. Two of the greatest thinkers in
ancient Greek philosophy—Plato and Aristotle—conducted much of their inquiries
Carruthers Chap19.indd 368Carruthers Chap19.indd 368 8/30/2007 12:08:38 PM8/30/2007 12:08:38 PM
The Moral Mind 369
UNCORRECTED PROOF
into ethics by examining the concept of virtue and the individual virtues, although
they had very different notions of what virtues were, what grounded them, and how
they were acquired.
2.1 Quandary Ethics and the Great Narrowing
The idea that morality is a set of virtues to be cultivated through practice remained
the dominant approach throughout the world until at least the Middle Ages.
St. Thomas Aquinas followed Aristotle in ethics as in other things, and even Islamic
thinkers, such as Miskawayh and al-Ghazali, borrowed from Aristotle in constructing
their theories of morality. Even up to the middle of the twentieth century, infl uential
philosophers and psychologists (Dewey, 1922; Hartshorne and May, 1928) contin-
ued to assume the essential validity of virtue theory and to base empirical research
programs on the assumption that virtues were psychologically real and served to
organize much of moral life.
But Western philosophers’ ideas about morality began to change in the eigh-
teenth century. For the most part, virtue- and religion-based moralities are character-
ized by specifi c, substantive beliefs and commitments, “thick” ideas about human
nature and society. With the Enlightenment, those assumptions came under increas-
ing scrutiny, and philosophers began to search for groundings for moral judgment
that did not depend upon specifi c metaphysical beliefs or group identities. What
MacIntyre (1981) has called “the Enlightenment project” was the attempt to ground
morality in highly abstract, even logical, truths and to disengage it (especially) from
religious belief. Two types of alternatives emerged that are of continuing relevance
today: formalist theories and consequentialist theories. Formalist theories of ethics,
of which Kant’s is the best-known example, defi ne moral judgments by reference
to their logical form—for example, as maxims or prescriptive judgments—rather
than by their content. The moral status of an action is judged by reference to the
kind of norm that underlies it. “Formalist” theories, in the sense we are using the
term here, would also include most varieties of contractualist theory, such as those
of John Rawls (1971) and Thomas Scanlon (1998), as well as of Locke, Hobbes,
and Rousseau. Like strictly formalist theories, contractualism attempts to ground
(or explain) moral judgments by positing hypothetical, contract-like relationships
between agents. Though they are more attentive to the realities of human nature
and of social and political arrangements, they still attempt to ground morality in
formal (in this case contractual) relations (in this case between individuals).
In contrast, consequentialist theories, including especially utilitarianism, attempt
to explain and ground moral judgments in premoral assessments of the consequences
of actions; the morally right thing to do is defi ned, fundamentally, as the thing that
will have the best consequences (however that very important term is understood).
Despite their differences—and they are great—both formalist and consequen-
tialist approaches to morality seek to detach moral judgment as much as possible
from the messy world of social practices and specifi c behaviors. Formalism replaces
substantive moral judgment with a logical rationality, whereas consequentialism
replaces it with a calculative rationality. Both approaches privilege parsimony:
Moral decisions should be made with respect to a foundational principle, such as
Carruthers Chap19.indd 369Carruthers Chap19.indd 369 8/30/2007 12:08:38 PM8/30/2007 12:08:38 PM
370 Culture, Motivation, and Morality
UNCORRECTED PROOF
the categorical imperative or the maximization of utility. Both insist that moral deci-
sions should be governed by reason and logic, not emotion and intuition. And both
devalue the particular in favor of the abstract.
The commonalities between these two approaches to ethics have led to a mod-
ern consensus about the scope of ethical inquiry: Morality is about resolving dilem-
mas involving the competing interests of people. The philosopher Edmund Pincoffs
(1986) calls this modern approach “quandary ethics,” and he laments the loss of
the older philosophical interest in virtue. Whereas the Greeks focused on character
and asked what kind of person we should each become, modern ethics focuses on
actions, trying to determine which ones we should do.
Nevertheless, quandary ethics has continued to fl ourish in philosophy and in psy-
chology, where it has guided the operationalization of morality. Lawrence Kohlberg’s
(1969) pioneering method was the longitudinal study of how children resolve moral
dilemmas: Should Heinz steal a drug to save his dying wife? Kohlberg’s conclusion
was that children get progressively better at quandary ethics until they reach the
highest stage, stage 5, at which all decisions are made by reference to the univer-
sally applicable, self-constructed, and nonconsequentialist principle of justice. Carol
Gilligan (1982) challenged Kohlberg’s conclusions by using a different dilemma: She
interviewed women facing the quandary of an unwanted pregnancy, and she offered
a competing highest principle: care. Social psychologists have also operationalized
morality as quandary, putting research subjects into diffi cult situations where they
must make choices that will help or harm a stranger (e.g., the “good Samaritan”
study, Darley and Batson, 1973; empathy-altrusim research, Batson et al., 1983; obe-
dience studies, Milgram, 1963). Baron (1993) has declared that consequentialism is
the normatively correct understanding of morality, and much of the research done
in connection with his approach involves presenting subjects with tradeoffs between
decision alternatives, each of which has costs and benefi ts. And when moral philoso-
phers conduct experiments, as they are beginning to do, they experiment primarily
on quandaries such as trolley and lifeboat problems that pit utilitarian and deonto-
logical concerns against each other (Greene et al., 2001; Petrinovich, O’Neill, and
Jorgensen, 1993).
Even when research methods have not used quandaries per se, they have
adopted the implicit boundary condition of quandary ethics: Moral issues are those
that pertain to the rights and welfare of individuals. Morality is about helping and
hurting people. Elliot Turiel, a former student of Kohlberg and a major fi gure in
moral psychology, codifi ed this individual-centered view of morality in his infl uen-
tial defi nition of the moral domain as
prescriptive judgments of justice, rights, and welfare pertaining to how people
ought to relate to each other. Moral prescriptions are not relative to the social con-
text, nor are they defi ned by it. Correspondingly, children’s moral judgments are
not derived directly from social institutional systems but from features inherent to
social relationships—including experiences involving harm to persons, violations of
rights, and confl icts of competing claims. (Turiel, 1983, p. 3)
Turiel’s delimitation of the moral domain seems obviously valid to many people
in modern Western cultures. However, for people in more traditional cultures, the
Carruthers Chap19.indd 370Carruthers Chap19.indd 370 8/30/2007 12:08:38 PM8/30/2007 12:08:38 PM
The Moral Mind 371
UNCORRECTED PROOF
defi nition does not capture all that they see as falling within the moral domain.
In other words, Turiel’s defi nition (we are asserting) is inadequate as an inductive
generalization. It is a stipulative defi nition which does not match the empirical
facts. When the moral domain is defi ned as “justice, rights, and welfare,” then the
psychology that emerges cannot be a true psychology of morality; it can only be
a psychology of judgments about justice, rights, and welfare. And when the domain
of morality is narrowed in this way, overly parsimonious theories of moral psychol-
ogy fl ourish. For example, morality can be explained evolutionarily as the extension
of kin altruism plus reciprocal altruism out to larger groups than those in which we
evolved. And morality can be explained developmentally as the progressive exten-
sion of the child’s understanding that harming others (which includes treating them
unfairly, unreciprocally) is bad.
But what if there is more to morality than harm, rights, and justice? What if
these concerns are part of a bigger and more complicated human capacity that can’t
be explained so parsimoniously? Might theories about the origins and development
of morality have been formulated prematurely?
2.2 The Rebirth of Breadth
One of the distinctions that has been most important in the study of morality, but
also most problematic, is that between “moral” and “conventional” judgments. Turiel
(and cognitive developmental theorists generally) distinguish the two domains of
social judgment on the basis of the presence of issues of “justice, rights, and welfare.”
Moral rules are those related to justice, rights, and harm/welfare (e.g., don’t hit, cheat,
or steal), and they can’t be changed by consensus because doing so would create new
classes of victims. In contrast, all those other rules children encounter (e.g., don’t call
adults by their fi rst names, do place your hand over your heart while saying the Pledge
of Allegiance) are matters of tradition, effi ciency, or social coordination that could just
as well be different if people in power, or if people in general, chose to change them.
In Western societies in which people accept a version of contractualism as the
basis for society, this distinction makes sense. But in most cultures the social order
is a moral order, and rules about clothing, gender roles, food, and forms of address
are profoundly moral issues (Abu Lughod, 1986; Meigs, 1984; Parish, 1994; Shweder,
Mahapatra, and Miller, 1987; Hampshire, 1982). In many cultures the social order
is a sacred order as well. Even a cursory look at foundational religious texts reveals
that while the gods do seem to care about whether we help or hurt each other,
they care about many other things besides. It would be a gross misunderstanding of
ancient Judaism, for example, to describe the Ten Commandments as a mixture of
moral rules (about not stealing, killing, or lying) and social conventions (about the
Sabbath, and prescribed ways of speaking and worshipping). Kelly and Stich (this
volume), in fact, argue that the domain theory propounded by Turiel and others
is simply false. They question the very categories of “moral” and “conventional”
as psychologically distinct domains, and they point to their own research showing
that even for some matters of harm, rights, and justice (e.g., fl ogging a disobedi-
ent sailor), Western adults judge transgressions to be somewhat authority-dependent
and historically contingent (D. Kelly et al., forthcoming).
Carruthers Chap19.indd 371Carruthers Chap19.indd 371 8/30/2007 12:08:38 PM8/30/2007 12:08:38 PM
372 Culture, Motivation, and Morality
UNCORRECTED PROOF
As cultural psychologists, we share Kelly and Stich’s concerns. We approach
moral functioning as an example of the general proposition that culture and mind
“make each other up,” to use Shweder’s (1990b) phrase. In contrast to Kohlberg (for
example), we think it is important to begin the explanation of moral functioning
by observing the individual and cultural facts about moral functioning, not with a
stipulative defi nition of the moral domain inherited from moral philosophers. This
approach is more “bottom-up,” not just empirically but also conceptually. We take
as given (at least at the beginning of analysis) that what people think are their moral
concepts are, in fact, moral concepts—rather than dividing them into “moral” and
“conventional” concepts at the outset.
All human societies generate and enforce norms (Brown, 1991). Sripada and
Stich (2007) have provided a useful defi nition of “norm”: “a rule or principle that
specifi es actions which are required, permissible or forbidden independently of any
legal or social institution.” To summarize their discussion, they characterize norms
as (a) rules or principles (b) with independent normativity and which generate (c) in-
trinsic motivation and (d) punitive attitudes toward violators. Quite simply, people
expect others to act in certain ways and not in others, and they care about whether
or not others are following these norms. The fi rst step in mapping the moral domain
of any culture, we believe, should therefore be to list and count the norms that get
the most attention. What norms and norm violations do people gossip about? What
norms are broken and punished in myths and folk tales? When people reject or criti-
cize other members of their community, or when they express shock at the practices
of another community, which norms are involved? (See Cosmides and Tooby, this
volume, for a similar approach to defi ning the moral domain.)
Such quantitative ethnography is diffi cult, but several research projects have
attempted to draw maps experimentally. Using their knowledge of the local norms
in the state of Orissa in India, Shweder, Mahapatra, and Miller (1987) created a list
of 39 actions, some of which directly caused harm, and others of which involved
matters of food, clothing, forms of address, and other traditions that would count
as social conventions on Turiel’s defi nition. Shweder’s Indian subjects—adults and
children—gave responses that revealed a moral domain very different from that of
his comparison sample of adults and children in Chicago. The American respon-
dents saw harm and rights violations in many of the actions (e.g., a husband beats
his wife for disobedience), and moralized them accordingly. The Indian respon-
dents, in contrast, revealed a broader moral world in which issues of respect and
hierarchy (e.g., a wife’s obedience to her husband) and spiritual purity/sanctity (e.g.,
not eating spiritually polluting foods at proscribed times) seemed to be at least as
important as issues of harm, rights, and justice. Haidt, Koller, and Dias (1993) later
showed that the cultural difference was not due to “hidden harms,” as Turiel et al.
(1987) had charged. Using a new set of harmless norm violations (including using a
fl ag to clean a toilet, and having sex with a chicken carcass), Haidt et al. found that
only an elite American college population limited the moral domain to matters of
harm, rights, and justice. For other groups, particularly for low socioeconomic status
groups in Brazil and the United States, actions that were disrespectful or disgusting
were said to be morally wrong (universally wrong and unchangeable) even when
respondents specifi cally stated that nobody was harmed by the action.
Carruthers Chap19.indd 372Carruthers Chap19.indd 372 8/30/2007 12:08:38 PM8/30/2007 12:08:38 PM
The Moral Mind 373
UNCORRECTED PROOF
Shweder (1990a; Shweder et al., 1997) later offered a useful systematization of
the breadth and variation of the moral domain. From cluster analyses of the moral
discourse provided by his Indian respondents, and from his own reading of the
anthropological literature, Shweder proposed that moral discourse around the world
generally draws on one or more of three “ethics”: autonomy, community, and divin-
ity. Each ethic is a set of interrelated moral claims that function to protect a different
entity. The “ethic of autonomy” functions to protect individuals, using concepts
such as harm and suffering, rights and justice, freedom and autonomy. This is the
moral domain as Turiel defi nes it.
In most cultures, however, people believe that there are things worth protecting
besides individuals. The “ethic of community” functions to protect groups, insti-
tutions, and other collective entities, using concepts such as duty, respect, honor,
loyalty, and tradition. The ethic of divinity functions to protect and glorify God,
particularly as God is manifested within each person. This ethic involves moral
concepts such as purity, piety, chastity, and other forms of self-restraint that help
people live in a more divine, less carnal way. When empirical comparisons of moral
discourse are made between more and less Westernized groups (Haidt et al., 1993;
Jensen, 1998), or between more or less politically and religiously conservative groups
in the United States (Haidt and Graham, 2007; in press; Haidt and Hersh, 2001;
Jensen, 1997), it is generally found that well-educated, secular, liberal Westerners
largely limit their moral discourse to the ethic of autonomy, whereas other groups
make use of a wider set of concepts, drawing heavily on the ethic of community and
often (though not as pervasively) on the ethic of divinity.
This broader conception of morality raises two challenges for innateness theo-
rists. First, they must explain how knowledge of or responses to this full set of moral
issues—not just harm, rights, and justice—is innate. Second, they must reconcile
their story about innateness with the obvious variation of moral rules and practices,
and of the moral domain itself, across cultures.
3 Five Ways Morality Could Be Innate
Given the theme of this volume, our goal here is to explore whether a broad and
heterogeneous set of moral concepts and motives—beyond harm, rights, and
justice—might refl ect the existence of some kind of innately given mental content.
We had better, then, be clear about what we mean by “innate.” The word has been
used in so many ways by philosophers, biologists, and ethologists (Wimsatt, 1999,
lists 13 distinct meanings) that some scholars have despaired of fi nding the concept
useful at all (e.g., Griffi ths, 2002). But we fi nd a simple and congenial approach in
the writings of Gary Marcus (2004), who studies the developmental pathways by
which genes guide the construction of brains. Marcus uses the metaphor that genes
create the fi rst draft of the brain, and experience later edits it: “Nature bestows upon
the newborn a considerably complex brain, but one that is best seen as prewired—
fl exible and subject to change—rather than hardwired, fi xed, and immutable” (p.
12). Marcus further explains that the editing—the changes in the brain as it learns
and grows—is itself governed by genetic processes. Genes are not just templates for
Carruthers Chap19.indd 373Carruthers Chap19.indd 373 8/30/2007 12:08:38 PM8/30/2007 12:08:38 PM
374 Culture, Motivation, and Morality
UNCORRECTED PROOF
making proteins, as was thought decades ago; rather, a part of each gene is devoted
to regulatory processes—switching the gene on and off in response to various chemi-
cal signals. Marcus (2004, p. 40) explains that “ ‘built-in’ does not mean unmallea-
ble; it means organized in advance of experience.” (Samuels, 2004, and this volume,
considers many meanings of innateness and reaches a similar conclusion.) We adopt
Marcus’s view of innateness, and in this section and the next one we try to explain
the ways that human morality may be “organized in advance of experience.” In sec-
tions 5 and 6 we try to explain how cultural and personal experience revises the fi rst
(universal) draft during childhood development.
We now describe fi ve ways that morality could be innate. We begin with two
theories of moral development—constructivism and connectionism—that are essen-
tially “blank slate” theories in that they posit only innate general learning processes,
not innate moral content (e.g., ideas, knowledge). We suggest that both theories
are partially correct as descriptions of the editing process, and that both can be
improved by positing at least some content that is organized in advance of experi-
ence. We then describe three approaches that do posit domain-specifi c innate moral
content. We believe that all fi ve of these approaches are useful, particularly when
the fi rst draft and editing processes are distinguished. In section 4 we propose our
own hypotheses as to what is inscribed in the fi rst draft of moral judgment, and in
sections 5 and 6 we describe how we think it is edited.
3.1 Piagetian Constructivism
Jean Piaget (1965/1932) got down on his knees and played marbles to study the pro-
cess by which children come to understand rules. In contrast to his contemporaries
who stressed the child’s passive internalization of the morals of the father (Freud) or
society (Durkheim), Piaget showed that children are active participants in their own
development. Development in any domain was, for Piaget, the product of the child’s
continuous interaction with the environment, as a result of which the child con-
tinually constructed and reconstructed a progressively more adequate understand-
ing. (Piagetian theory is therefore often referred to as “constructivism.”) Lawrence
Kohlberg worked out the sequence of progressively more adequate understandings
of morality in his famous six stages, and he credited the process of “role-taking” with
being the driving force of moral development.
For Piaget and Kohlberg, there was just one word written in the fi rst draft of
the moral mind: empathy (or perhaps a compound word: “perspective-taking”).
There were also some words written in other (nonmoral) chapters of the fi rst draft:
like and dislike. As long as children liked some things (such as pleasure, candy, or
friendship) and disliked others (such as pain, frustration, or rejection), then as they
became increasingly good at taking the perspective of others during the concrete
operational stage, they could feel for themselves (empathetically) that actions that
hurt others were bad, while actions that made others happy were good. In this way
children come to understand the values of different kinds of rules, and to appreciate
that rules and social practices have to be justifi ed by reference to something else.
For Kohlberg, that something else was tradition, authority, and society for “conven-
tional” moral reasoners, but it was justice for the most advanced moral reasoners.
Carruthers Chap19.indd 374Carruthers Chap19.indd 374 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
The Moral Mind 375
UNCORRECTED PROOF
The knowledge of justice was not innate; it was the crowning achievement of the
editing process, and the editing process was constructivism driven by the experience
of role-taking.
This approach is elegant in explaining how so much can be derived from so
little innate knowledge, and it is reasonable if you believe that the moral domain
is restricted to matters of harm, rights, and justice. However, if you believe that
concepts such as obedience, respect, honor, chastity, temperance, and sacrilege are
truly moral concepts that need to be explained (rather than overcome by the child
on the way to moral autonomy and perfect justice reasoning), then it is not clear
how these can be derived from empathy and role-taking. Why not posit that the fi rst
draft of the moral mind has several words written into it beyond empathy, and that
Piaget and Kohlberg are correct that constructivism is a part of the editing process?
3.2 Connectionism
A second approach that focuses on the editing process is the connectionist paradigm
proposed by Paul Churchland (1996, 1998). Churchland starts from the manifest
fact that in addition to the physical environment, human beings are born into and
live in a sociomoral world that is extremely dense and complex. The problem for
such creatures is to learn to navigate this environment successfully by developing
adequate representations of it and pairing those representations with appropriate
behavioral responses. There is no special faculty for accomplishing this task; it is
accomplished in the same way that people learn to represent and live in their physi-
cal environment: through the gradual tuning up of expertise by a mind that is pro-
duced by a brain that is a neural network.
Churchland’s connectionist account of moral functioning is essentially an
account of moral learning. For Churchland, moral development does not mean,
as it did for Kohlberg, the gradual formulation of abstract, universal moral prin-
ciples; rather, it is “a matter of slowly generating a hierarchy of moral prototypes,
presumably from a substantial number of relevant examples of the moral kinds at
issue” (Churchland, 1996, p. 102). Churchland’s account of moral cognition and
competence is closely congruent with one modern view of cognition, and it is also
congruent with some of the most ancient understandings of moral development,
particularly Aristotle’s (Casebeer, 2003). Churchland’s approach is essentially a
translation of the terms of virtue ethics into connectionist language. It treats moral
competence as a set of skills cultivated gradually, by practice, and helped along by
adult emphasis on moral stories, fables, and role models.
Churchland’s approach is eloquent on how the editing process occurs, but it is
silent on the contents of the fi rst draft. In fact, it invites the inference that there is
no fi rst draft, other than an innate interest in people and social events. We see this as
an easily correctable fl aw, for it implies equipotentiality in moral learning: Children
could just as easily learn to navigate and value any artifi cial moral world that adults
created for them, such as those of communes and kibbutzim that tried, unsuccess-
fully, to overcome people’s preferences for sharing material goods with their close
kin, and their aversion to mating with the people they were raised with. Just as chil-
dren enter the world with some initial settings in the food domain (a liking for sweet,
Carruthers Chap19.indd 375Carruthers Chap19.indd 375 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
376 Culture, Motivation, and Morality
UNCORRECTED PROOF
a dislike of bitter) which are then extended by cultural learning, so it seems likely
that children enter the world with some initial settings in the social domain (a liking
for fairness, a dislike of harm) which are then extended by cultural learning. It does
little violence to Churchland’s theory to propose that the initial state of the system,
for example, the initial weights of particular synaptic connections, is “organized in
advance of experience,” and then edited by experience in the way that Churchland
describes. (See A. Clark’s [2000] critique of Churchland for a similar point.)
3.3 Relational Models
A third approach to innateness is Alan Fiske’s (1991, 1992) theory of “relational mod-
els.” Fiske’s theory is elegant and parsimonious; it was designed to explain cross-
cultural similarity and diversity in a wide range of cultural domains. Fiske (2004,
p. 3) gives the following capsule summary:
Relational models theory is simple: People relate to each other in just four ways.
Interaction can be structured with respect to (1) what people have in common, (2)
ordered differences, (3) additive imbalances, or (4) ratios. When people focus on
what they have in common, they are using a model we call Communal Sharing.
When people construct some aspect of an interaction in terms of ordered differ-
ences, the model is Authority Ranking. When people attend to additive imbalances,
they are framing the interaction in terms of the Equality Matching model. When
they coordinate their actions according to proportions or rates, the model is Market
Pricing. . . In short, four innate, open-ended relational structures, completed by
congruent, socially transmitted complements, structure most social action, thought,
and motivation. That’s the theory.
Fiske suggests that the fi rst three models are innate, and are clearly found
in other primates (Haslam, 1997). But Market Pricing—social relations based on
ratios—appears to be uniquely human, emerging later in both phylogenetic and
ontogenetic development than Communal Sharing, Authority Ranking, and
Equality Matching. The case for the innateness of Market Pricing is therefore much
weaker, and Fiske suggests that it might be in the process of becoming an innate psy-
chological mechanism for social relationships. Fiske therefore is very clear that the
fi rst draft of the social mind contains three primitive and inescapable social-perceptual
tendencies: to see people as groups in which all are the same, as rank-orderable on
one or more dimensions, or as entities whose relative position must be kept equal
despite periodic shifts out of balance. These innate models are used in all cultures to
structure the social world, although cultures often choose to use different combina-
tions of models to govern any given relationship. For example, the division of house-
hold labor may rely on Authority Ranking in some cultures (e.g., the husband is the
authority and dictates who does what); or on Equality Matching (both spouses take
turns at each job); or on Communal Sharing (everyone pitches in without keeping
track of who does what); or on Market Pricing (jobs are assigned values proportional
to their diffi culty, and children, or hired helpers, are paid to perform them).
Most of Fiske’s relational models theory is about the editing process: how this
fi rst draft, with three (or four) simple and open cognitive frames gets fi lled in and
Carruthers Chap19.indd 376Carruthers Chap19.indd 376 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
The Moral Mind 377
UNCORRECTED PROOF
tuned up during childhood development. We have no criticism of Fiske’s theory,
except that we think a bit more is given in the fi rst draft, as we explain below.
3.4 Massive Modularity
The fourth and most widely discussed approach to innateness is the concept of
modularity. One can distinguish between minimalist and maximalist theories of
modularity. On the minimalist view (Fodor, 1983), a very small number of mental
functions— primarily having to do with sense perception and language—are modu-
lar, in the sense that they are innate, fast, informationally encapsulated, functionally
specialized computational mechanisms. Fodor offered a stringent defi nition of what
it takes to be a module and then claimed that there are very few modules in the mind,
and none for handling higher-order tasks such as reasoning or moral judgment.
On the maximalist view proposed by evolutionary psychologists (e.g., Barkow,
Cosmides, and Tooby, 1992, and this volume; Buss, 2004; Pinker, 1997), the vast
majority of the mind is composed of modules—hundreds or thousands of them—
designed by natural selection to solve specifi c problems that were recurrent in the
ancestral environment. According to Tooby, Cosmides, and Barrett (2005), many of
these modules play an important role in our moral lives. They argue that the study
of valuation, even more than other kinds of cognition, reveals just how crucial it is
to posit innate mental content, not just innate learning processes. Children are born
with a preference (value) for sweetness and against bitterness; any parent knows that
the preference for candy over broccoli is not learned by “socialization” and cannot
be undone by role models, threats, or rewards. Tooby et al. suggest that the same
thing is true for valuation in all domains:
The proprietary content introduced by the architecture constitutes a form of knowl-
edge: the architecture must know (in some sense) that living children are better
than dead children, social approval is better than disapproval, salt and sweet are
better than acrid or putrefying, sex with your mother or father is to be avoided,
helping siblings is (within certain tradeoffs) better than helping fungi, your mate
copulating with your sexual rival is worse than his or her fi delity, spiders on your
cheek are worse than in the garden, understanding is better than confusion, skill
mastery is better than inept performance, and so on. (2005, p. 317)
In this passage Tooby et al. gather many kinds of valuation together under the rubric
of what they elsewhere call “motivational principles,” but not all of them are rel-
evant for our purpose, which is to think through the ways in which specifi cally
moral judgment might have an innate foundation. Salt, sweetness, and spiders, for
example, while clearly the objects of tastes and preferences, seem different in kind
from preferences connected with understanding, sexual fi delity, and helping, if for
no other reason than that the latter seem to have more conceptual and less percep-
tual content. Tooby et al. are interested in developing an inventory and a science of
motivation in general, rather than a theory of morality, and while the latter is cer-
tainly related to the former (and may, perhaps, turn out to be just a special case of
it), the moral domain is distinctive enough that their very useful account will need
a little modifi cation.
Carruthers Chap19.indd 377Carruthers Chap19.indd 377 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
378 Culture, Motivation, and Morality
UNCORRECTED PROOF
We agree with Tooby et al. that valuation—for social behavior as for food—is
impossible to explain if one refuses to entertain the notion that there is innate struc-
ture and content built into the mind. As they put it:
there must be an irreducible core set of initial, evolved, architecture-derived
content-specifi c valuation assignment procedures, or the system could not get
started. The debate cannot sensibly be over the necessary existence of this core set.
The real debate is over how large the core set must be, and what the proper compu-
tational description of these valuation procedures and their associated motivational
circuitry is. (2005, p. 317)
Massive modularity is a controversial notion. Jerry Fodor, the original author of
the concept of mental modules, has said that “the massive modularity thesis pretty
clearly isn’t true” (Fodor, 2000, p. 23), and a number of other thinkers have followed
his lead for diverse reasons (see, e.g., Buller, 2005; Buller and Hardcastle, 2000). We
see two principal diffi culties in applying the massive modularity thesis to morality.
The fi rst is one of Fodor’s main concerns, known as the “fl exibility problem.”
Higher- order human cognition—and certainly moral cognition—is quite fl exible.
People and societies are quite good at invoking moral concepts that suit their pur-
poses, or twisting those that don’t into more amenable shapes. Reactions that are
often said to be based on modularized knowledge, such as fear of spiders, seem to
have a more low-level, stimulus-response quality to them.
2
The second problem is the encapsulation problem: While many moral judg-
ments meet most of Fodor’s criteria for modularity—including domain specifi city
and speed—it is implausible to think that moral judgments are as informationally
encapsulated as the sorts of phenomena usually used to illustrate modularity at
the perceptual level. For example, the Müller-Lyer illusion is unaffected by one’s
knowledge of the true lengths of the lines, but moral judgments are easily affected
by learning new facts about the situation, or by experimental manipulations of
mood or other factors that seem extraneous to the operation of a moral module (e.g.,
Valdesolo and DeSteno, 2006). We are sympathetic to the possibility of substantial
domain-specifi c knowledge in the fi rst draft of the moral mind, but we would like a
version of modularity that can solve these two problems.
3.5 “Teeming” Modularity
Several theorists (e.g., Carruthers
3
) have sought a middle way between completely
nonmodular conceptions of the mind and massively modular theories. These think-
ers speak of “moderately massive modularity” or “modularity to some interesting
2. Modules can be combined to create systems that generate fl exible behavior, as is said to be the case
with the “language organ” (Pinker, 1997). But because we do not believe there is a single morality organ
(see Greene and Haidt, 2002), we search for moral modules at a simpler level, as multiple sources of
intuition. Hauser (2006) claims that there is a moral organ, but we believe he has described just a harm-
processing organ.
3. Available at http://www.philosophy.umd.edu/Faculty/pcarruthers/Moderate-modularity.htm.
Carruthers Chap19.indd 378Carruthers Chap19.indd 378 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
The Moral Mind 379
UNCORRECTED PROOF
degree.” They doubt that there are no conceptual modules, but they are also skeptical
that the mind is a Swiss army knife crammed with tools that were fully designed long
ago. One of the most important of these moderate modularists is the anthropologist
Dan Sperber (1994, 2005). As an anthropologist, Sperber’s goal was to explain both
the diversity and the stability of culture. Massive modularity with Fodorean modules
is hard (though not impossible) to reconcile with the cultural diversity of concepts
and behavioral patterns, and with the sometimes rapid pace of cultural change. But
on the other hand, the nearly blank slate models assumed by many anthropologists
cannot explain either the deep and surprising similarities between cultures (e.g., in
gods, ghosts, and witches; Boyer, 2001), or the degree of cross-generational stability
that most cultures achieve (Sperber and Hirschfeld, 2004).
Sperber’s solution is a version of massive modularity, but his modules are
decidedly un-Fodorean: They are highly variable (some meet all of Fodor’s cri-
teria, some meet only a few); they are often nested within each other (just as the
digestive system is a biological module that contains many submodules); and,
most important, most of Sperber’s modules are not innate; they are generated
during development by a smaller set of “learning modules” which are innate tem-
plates or “learning instincts” (Sperber, 2005, p. 57, citing Marler, 1991). Some of
these innate modules have specifi c perceptual content built in; for example, a
fruit-learning module will “know” that fruit is sweet, and will subsequently gener-
ate only fruit recognition submodules (e.g., one for apples, one for bananas) for
objects in the environment that meet those prespecifi ed criteria. Other learning
modules may be more purely conceptual; for example, if there is an innate learn-
ing module for fairness, it generates a host of culture-specifi c unfairness detection
modules, such as a “cutting-in-line detector” in cultures where people queue up,
but not in cultures where they don’t; an “unequal division of food” detector in
cultures where children expect to get exactly equal portions as their siblings, but
not in cultures where portions are determined by age. Because Sperber envisions
a core set of innate modules generating a great diversity of other modules, he uses
the evocative term “teeming modularity.”
At this point, any reader who is not already a modularity theorist is likely to
think that we have joined Sperber in a jump off a cliff into a land where everything
and everybody is named “module.” Let us explain why we are intrigued by Sperber’s
ideas. Our goal is to understand the fi rst draft and the revision processes that create
the moral mind. Our empirical research is on moral intuition and moral dumb-
founding (for Haidt), and on culture and virtue (for Joseph). We have both found
that moral judgment is not well described by the domain-general application of
rules and principles to specifi c cases, as though moral judgment were a product of
moral reasoning in the Kohlbergian sense. Rather, when people are interviewed
about taboo violations (such as consensual sibling incest, or harmless cannibalism),
they answer very quickly, and their answers show what appears to be a kind of Müller-
Lyer-like encapsulation: People can sometimes be pushed in cross-examination to
say “I don’t know why, I can’t explain it, I just know it’s wrong” (Haidt, 2001; Haidt
and Bjorklund, in press). We have argued (Haidt and Joseph, 2004) that the adult
mind is full of moral intuitions, which are like little bits of input-output program-
ming connecting the perception of a pattern in the social world (often a virtue or a
Carruthers Chap19.indd 379Carruthers Chap19.indd 379 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
380 Culture, Motivation, and Morality
UNCORRECTED PROOF
vice) to an evaluation and, in many cases, a specifi c moral emotion (e.g., anger,
contempt, admiration). When people think, gossip, and argue about moral issues,
the playing fi eld is not affectively fl at and open to any kind of reason; it is more like
a minefi eld or a pinball machine where fl ash after fl ash of affectively laden intuition
bounces around one’s attention and pushes one toward specifi c conclusions. (See
Damasio, 1994, for a similar idea, and for descriptions of what happens to moral
thinking when these fl ashes are removed.) These intuitions are not Fodor modules,
but they are modular “to some interesting degree” (Sperber, 1994). They are fast,
domain-specifi c bits of mental structure that strongly infl uence moral judgment
(Haidt, 2001). Where do all these intuitions come from?
Perhaps they are all innate, and people simply learn what events, in their culture,
count as acts of harm or unfairness (e.g., cutting in line). But Sperber’s approach allows
us to explain certain acquired moral tastes in much the way that other kinds of acquired
tastes and fears are explained. People are innately attracted to fruits and to meat, but
there are special learning mechanisms that can generate a new and enduring disgust
toward specifi c foods, particularly meats. For example, in 1805, when the Lewis and
Clark expedition survived months of starvation in the Bitterroot Mountains of Idaho
and fi rst made contact with the Nez Perce tribe along the Columbia River, the men
gorged themselves on salmon and on a root vegetable, both of which were new to them.
Many of them got sick that night, probably from the barely digestible root vegetable,
but because of innate one-trial learning mechanisms that associate nausea preferen-
tially with meat (meats are much more likely to contain bacterial contaminants than
are vegetables), they developed a disgust toward salmon. The disgust was so strong that
in subsequent days the men purchased dogs from the locals to eat because that was the
only meat available to them other than salmon (Burns, 1997). Was this an example of
domain-general learning? Fodor would have to say yes, but Sperber would say no, it was
the genesis of a new module from an innate learning module. We agree with Sperber;
we see this as a new intuition (a gut feeling) generated by an innate learning process
that can radically alter the value of things on the basis of experience, but only within
limits related to evolutionary adaptation. The new intuition was partially encapsulated:
If Lewis and Clark had convinced their crew that the root vegetable was the real culprit,
the men would still have felt disgust toward salmon.
Moral development shows some of these same features. Children gradually
come to recognize a large set of input patterns to which they then react quickly,
automatically, and emotionally. For example, Americans in recent decades have
become fi nely attuned to the issue of sexual abuse of children, so much so that they
are horrifi ed by social patterns that are quite normal in other parts of the world, such
as having children sleep in the same bed as an opposite sex parent through middle
childhood (Shweder, Balle-Jensen, and Goldstein, 1995), or kissing the genitals of
infant boys as an expression of affection, as is done in some Middle Eastern coun-
tries. Explaining to Americans that these practices are not thought by participants
to have anything to do with sexuality is not going to eliminate the disgust and con-
demnation—or the charge of child abuse. Is there an innate sexual abuse detector?
Probably not. But as we will explain below, we think there is something innate—
something “organized in advance of experience”—that makes sexual activity, and
the protection of children, evolutionarily prepared domains for moral concern.
Carruthers Chap19.indd 380Carruthers Chap19.indd 380 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
The Moral Mind 381
UNCORRECTED PROOF
Other examples would be the speed, ease, and passion with which the American
“Religious Right” sees sin, temptation, and sacrilege, or the American political Left
sees racism, oppression, and victimization. Whether or not these moral reactions are
seen as manifestations of acquired (teeming) modules or just as subcultural exper-
tise, they are examples of moral intuitions: bits of mental structure that connect the
perception of specifi c patterns in the social world to evaluations and emotions that
are not fully controllable or revisable by the person who experiences them.
4 The First Draft of Morality: The Five Foundations of
Intuitive Ethics
We have long been searching for the foundations of intuitive ethics—the psycho-
logical primitives that are the building blocks from which cultures create moralities
that are unique yet constrained in their variations. Recently we (Haidt and Joseph,
2004) examined a number of theories (including Shweder’s and Fiske’s) about the
breadth of human morality and about its precursors in other primates (e.g., de Waal,
1996). We tried to identify the full range of phenomena across cultures that would
need to be explained by any adequate theory of human morality. We identifi ed fi ve
sets of concerns, each linked to an adaptive challenge and to one or more moral
emotions, as the best candidates for the psychological foundations of human moral-
ity. The fi ve foundations we identifi ed are harm/care, fairness/reciprocity, in-group/
loyalty, authority/respect, and purity/sanctity. Each of these fi ve is a good candidate
for a Sperber-style learning module. However, readers who do not like modularity
theories can think of each one as an evolutionary preparedness (Seligman, 1971) to
link certain patterns of social appraisal to specifi c emotional and motivational reac-
tions. All we insist upon is that the moral mind is partially structured in advance
of experience so that fi ve (or more) classes of social concerns are likely to become
moralized during development. Social issues that cannot be related to one of the
foundations are much harder to teach, or to inspire people to care about.
Table 19.1 gives our theory in a concise form. The fi rst row lists fi ve long-standing
adaptive challenges that highly social mammals such as our ancestors faced for mil-
lions of years, creating conditions that favored the reproductive success of individuals
who could solve the problems more effectively. For each challenge, effective adapta-
tion meant being able to detect certain patterns in the social world and respond to
them with an altered motivational profi le. Sperber (1994) refers to the set of objects
that a module was “designed” to detect as the proper domain for that module. He
contrasts the proper domain with the actual domain, which is the set of all objects
that in fact trigger the module. Even when an animal lives in the environment for
which it is adapted, the actual domain never perfectly matches the proper domain;
the module produces false positives and false negatives. For a species such as ours,
which has experienced rapid environmental change, and which deliberately manip-
ulates its physical and social environment for a variety of reasons, we can be quite
certain that the actual domain diverges from the proper domain to a substantial
degree: Our moral intuitions are sure to be engaged—or suppressed—in ways that
do not contribute to each individual’s Darwinian success.
Carruthers Chap19.indd 381Carruthers Chap19.indd 381 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
382 Culture, Motivation, and Morality
UNCORRECTED PROOF
table 19.1 The Five Foundations of Intuitive Ethics
Harm/Care Fairness/Reciprocity In-group/Loyalty Authority/Respect Purity/Sanctity
Adaptive challenge Protect and care Reap benefi ts of dyadic Reap benefi ts of group Negotiate hierarchy, Avoid microbes and
for young, vulnerable, cooperation with cooperation defer selectively parasites
or injured kin non-kin
Proper domain Suffering, distress, Cheating, cooperation, Threat or challenge Signs of dominance Waste products,
(adaptive triggers) or threat to one’s kin deception to group and submission diseased people
Actual domain Baby seals, cartoon Marital fi delity, broken Sports teams one Bosses, respected Taboo ideas
(examples of characters vending machines roots for professionals (communism,
modern triggers) racism)
Characteristic Compassion anger, gratitude, guilt Group pride, Respect, fear Disgust
emotions belongingness;
rage at traitors
Relevant virtues Caring, kindness, fairness, justice, Loyalty, patriotism, Obedience, deference Temperance, chastity,
(and vices) (cruelty) honesty, self-sacrifi ce (disobedience, piety, cleanliness
trustworthiness (treason, cowardice) uppitiness) (lust, intemperance)
(dishonesty)
Carruthers Chap19.indd 382Carruthers Chap19.indd 382 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
The Moral Mind 383
UNCORRECTED PROOF
The way to read the table is to read down each column. For example, the Harm/
Care foundation can be understood by beginning with the fact that mammals by
defi nition face the need to care for vulnerable offspring, and nothing could be more
central to evolutionary success than keeping these offspring alive. It is therefore
implausible that mammals learn entirely through domain-general learning mecha-
nisms how to recognize suffering or distress in their offspring. Rather, many mammals
have innate harm detection modules that were shaped by evolution to be responsive
to the proper domain of signs of suffering in their own offspring. In actual practice
this module (or set of modules) is responsive to many things besides the suffering and
distress of one’s own children. For humans, suffering by or harm to almost any child-
like entity is part of the actual domain of this module. (A poster showing baby seals
being clubbed to death by large men is a deliberately contrived superstimulus for
this module.) These modules generally have as one of their outputs the emotion of
compassion: The individual is motivated to act so as to relieve suffering or otherwise
protect the child. We do not know whether there is a single harm module that has
both innate and learned triggers (as Hauser, 2006, suggests), or whether the teem-
ing modularity account, in which the human mind is innately prepared to generate
a host of specifi c harm-related modules, is correct. However, if all people have an
emotional sensitivity to harm, particularly harm to the weak or vulnerable, and if
people have language, then they are likely to develop a vocabulary for talking about
their emotional reactions. They are likely to have virtue and vice words with which to
praise and condemn people, and to instruct their children. Such virtue talk can then
feed back to fi ne-tune the bounds and applications of the modules: Cultures can
become expert in perceiving certain kinds of harm (e.g., sexual abuse or witchcraft).
We tell a similar story for each of the other four columns. The Fairness/
Reciprocity foundation, for example, is just an elaboration of the story told by Trivers
(1971) about how a suite of emotions may have evolved that helps social organisms
reap the gains of reciprocal altruism with non-kin or distant kin. Along with the
evolved individual-level attributes of heightened interest in and emotional reactiv-
ity to signs of cheating and cooperation comes a suite of cultural products, such
as virtue and vice words related to fairness, religious injunctions about reciprocity,
cultural constructs such as rights, and social institutions related to justice.
The next column, for the In-group/Loyalty foundation, organizes phenomena
related to the well-studied human tendency to aggregate into tribes, gangs, and teams
that compete with other tribes, gangs, and teams. Tajfel et al.’s (1971) minimal group
experiments demonstrate that people will form such groups on the basis of even
trivial similarities; groups based on shared blood, religion, or language are vastly
more powerful. Confl icts over territory or attacks from other groups seem to call
particularly keenly upon virtues related to this foundation, such as loyalty, heroism,
and self-sacrifi ce for the common good. When these in-group virtues are prominent,
the group is correspondingly likely to be highly vigilant about and punitive toward
traitors, profi teers, and slackers. The destructive potential of this module is on daily
display around the world, including episodes of genocide and ethnic cleansing. The
moralization of unity during wartime is also evident in the title of a recent book by an
American archconservative, outraged at dissent during the Iraq war: Treason: Liberal
Treachery From the Cold War to the War on Terrorism (Coulter, 2003).
Carruthers Chap19.indd 383Carruthers Chap19.indd 383 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
384 Culture, Motivation, and Morality
UNCORRECTED PROOF
The fourth column is about the psychological and social concomitants of life in
dominance hierarchies. Many primates live in such hierarchies, and the common
display patterns of dominance and submission across species and across cultures
strongly suggest that something in the human mind was organized in advance of
experience, making it easy for humans to develop a suite of emotions and behaviors
related to authority and power. (See Boehm, 1999, for a discussion of how egalitarian
societies arise despite the human predisposition for hierarchical living.) However,
as Fiske points out repeatedly, Authority Ranking is a two-way street: Subordinates
must show respect and deference, but superiors must then protect them from exter-
nal threats and maintain order within the group. This pro-social side of authority
seems to go unrecognized in many contemporary psychological accounts of hier-
archy which, as Shweder et al. (1997) point out, see all forms of inequality as forms
of oppression. In societies that value authority, however, norms and related virtues
govern the behavior of superiors (e.g., impartiality, magnanimity, fatherliness) and
subordinates (e.g., respect, deference).
And fi nally, the fi fth foundation, Purity, is unique in that it is the only one for
which the original adaptive challenge was not social, but nutritive. The omnivo-
rous food strategy of human beings, combined with our relatively large group sizes
(compared to other primates; Dunbar, 1993) means that we have long been exposed
to very high levels of threat from bacteria and parasites, which spread by physical
contact. Humans (but no other animals) therefore developed a suite of cognitive
and emotional adaptations related to disgust that make us wary but fl exible about
the kinds of things we eat, and about the contact histories of the things we eat (Rozin
and Fallon, 1987). This food evaluation and rejection system was well adapted for
social evaluation and rejection, and most (if not all) human societies use some of the
vocabulary and logic of physical disgust in their moral life (Haidt et al., 1997; Rozin,
Haidt, and McCauley, 2000). In some societies the ability to track contagion and
value purity seems to contribute to ideas about sacredness—about keeping religious
objects set apart from pollutants and profane objects, and about overcoming carnal
desires and treating the body as a temple (see Eliade, 1957/1959, on sacredness). This
foundation therefore often generates virtues such as chastity and temperance, and
vices such as lust and intemperance.
Purity is often deeply moralized, not only as a concern about the self but also in
the form of beliefs and feelings about groups and the world as a whole. This is one
source of what might be called the “dark side” of purity intuitions, and indeed a con-
cern (or obsession) with purity is often associated with horrifi c violence and oppres-
sion, particularly when it pairs up with intuitions from the In-group foundation—for
example, the Holocaust, ethnic “cleansing,” and the Jim Crow laws in the American
South that kept African American bodies and body processes separated from those
of whites.
We believe these fi ve sets of issues, sensitivities, and social-perceptual skills are
the best candidates for being the foundations of intuitive ethics for several reasons.
First, in the ways that cultures deal with these fi ve adaptive challenges we fi nd a
surprising degree of similarity—for example, in the logic of initiation rites that cre-
ate a strong in-group; in the ways that hierarchy and submission are marked; and
in the purity and pollution rules that so often regulate biological processes such as
Carruthers Chap19.indd 384Carruthers Chap19.indd 384 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
The Moral Mind 385
UNCORRECTED PROOF
menstruation, birth, and defecation. Second, four of our proposed fi ve foundations
(all but purity) appear to involve psychological “building blocks” that are present in
other primates (de Waal, 1996, although reciprocity is still debated; Hauser, 2006),
giving us further confi dence that something about these foundations is “specifi ed in
advance of experience.” Third, our fi ve foundations fi t perfectly with Shweder’s three
ethics (the Harm and Fairness foundations give rise to the discourse of the ethic of
autonomy; the In-group and Authority foundations support the ethic of community;
and the Purity foundation supports the ethic of divinity). Fourth, three of our foun-
dations are coincident with Fiske’s fi rst three relational models (Fairness = Equality
Matching; In-group = Communal Sharing; Authority = Authority Ranking). To the
extent that our fi ve foundations don’t match Fiske’s four models, the discrepancy
is due to the fact that Harm and Purity are not primarily modes of interpersonal
relationship. We include them because they are important and probably innate
sources of human moral valuation; we do not include Market Pricing because we
do not think it is so clearly innate. We could easily be wrong about excluding Market
Pricing. We do not claim that there are only fi ve foundations. There are probably
many more, but we believe the fi ve we have identifi ed are the most important ones
for explaining human morality and moral diversity.
5 The Editing Process: Developing Virtues
The fi ve foundations are, to propose an analogy, the innate “taste buds” of the moral
sense. The human tongue has fi ve kinds of receptors, each of which translates a
chemical pattern in a substance into an affective experience that is positive (for sweet,
salt, and glutamate) or negative (for bitter and, beyond a certain level of intensity,
for sour). These taste buds tell us something about how our ancestors lived: They ate
fruit and meat, and had a variety of perceptual tools in their tongues (and noses and
eyes) that meshed with conceptual tools in their brains to help guide them to fruit
and meat. Similarly, the fi ve foundations suggest some things about how our ances-
tors lived: They were ultrasocial creatures (Richerson and Boyd, 1998), fi nely tuned
for (1) rearing children and helping kin, (2) selectively cooperating with non-kin
while remaining vigilant for cheaters, (3) forming strong in-groups for the purpose of
cross-group competition, (4) organizing themselves hierarchically, and (5) attending
to each other’s physical states, and altering interactions and contacts accordingly.
The taste buds on the tongue gather perceptual information (about sugars, acids,
etc.), whereas the taste buds of the moral sense respond to more abstract, conceptual
patterns (such as cheating, disrespect, or treason). Nonetheless, in both cases, the
output is an affectively valenced experience (like, dislike) that guides subsequent
decisions about whether to approach or avoid the object/agent in question.
Of course there is much more to moral judgment (and to food selection) than
the operation of fi ve “taste buds.” Mature moral functioning does not consist only,
or even primarily, of simple affective or intuitive reactions to social stimuli. Disgust
felt toward dog feces, or even toward an act of homosexual intercourse, is not in
itself a moral judgment. Moral development is also characterized by the acquisi-
tion and use of a wide variety of moral concepts. Some of these are categories of
Carruthers Chap19.indd 385Carruthers Chap19.indd 385 8/30/2007 12:08:39 PM8/30/2007 12:08:39 PM
386 Culture, Motivation, and Morality
UNCORRECTED PROOF
actions—lies, betrayals, favors, and so on. Others are categories of persons, or more
specifi cally, categories of characteristics of persons, including positively valenced
traits
4
such as kindness, loyalty, and trustworthiness, and negatively valenced ones
such as cruelty, dishonesty, and cowardice.
These traits—virtues and vices—are beginning to reemerge in empirical moral
psychology after a long period of exile, occasioned in part by critiques by Lawrence
Kohlberg and other theorists. We have discussed the role that virtues can play in
a comprehensive theory of moral functioning and moral development (Haidt and
Joseph, 2004). We repeat our main points here, as a prelude to a discussion of one
aspect of virtue theory and of our theory in particular: narrative.
First, what is a virtue? There are many views, but most virtue theorists would
agree at least that virtues are characteristics of a person that are morally praiseworthy.
Virtues are therefore traits, as long as one doesn’t think of traits as global tendencies
to act in a particular way (e.g., honest, brave) across widely varying circumstances.
Rather, we think of traits as John Dewey did: as dynamic patternings of perception,
emotion, judgment, and action (Dewey, 1922; see also Churchland, 1998). Virtues
are social skills. To possess a virtue is to have extended and refi ned one’s abilities
to perceive morally relevant information so that one is fully responsive to the local
sociomoral context. To be kind, for example, is to have a perceptual sensitivity to
certain features of situations, including those having to do with the well-being of
others, and for one’s motivations to be appropriately shaped and affected. To be cou-
rageous is to have a different kind of sensitivity; to be patient, still another.
5
One of the crucial tenets of virtue theory is that the virtues are acquired induc-
tively, through exposure to—sometimes with efforts to copy—many examples of the
virtue in practice. Each of these examples contains information about a number of
aspects of the situation, including the motivations of the protagonists, their state of
being (suffering, disabled, hostile, rich, etc.), the categorization of the situation, and
the evaluation of the outcome offered by more experienced others. Only over time
will the moral learner recognize what information is important to notice and retain,
and what can be safely disregarded.
Philosophers and cognitive scientists have recently been arguing, with respect
both to morality and to cognition more generally, that this kind of learning cannot
4. The word “trait” is fraught with signifi cance in psychology; in particular, it is the focus of a heated
debate between personality and social psychologists. Some psychologists have placed traits at the center
of the study of personality, while others, for various reasons, are skeptical or dismissive of the very concept
(Mischel 1968; Ross and Nisbett, 1991). This debate has penetrated philosophical and psychological dis-
cussions of morality, with “situationists” such as John Doris (1998) and Gilbert Harman (1999) constru-
ing virtues as traits and then dismissing their existence, and virtue theorists (Sreenivasan, 2002; Merritt,
2000) defending versions of virtue theory against the situationist critique. We believe the virtue theorists
are right; virtues, as we construe them, are highly situation-specifi c skills or capacities rather than broad
behavioral dispositions. This way of seeing virtues obviates the basic charge of the situationists, and is
consistent with Mischel’s original and ongoing critique of trait theories (Cervone and Shoda, 1999), and
with recent work in cognitive neuroscience (e.g., Casebeer, 2003; Churchland, 1998).
5. For a classic exposition of the construal of virtues as sensitivities or perceptual capacities, see McDowell
(1979).
Carruthers Chap19.indd 386Carruthers Chap19.indd 386 8/30/2007 12:08:40 PM8/30/2007 12:08:40 PM
The Moral Mind 387
UNCORRECTED PROOF
be replaced with top-down learning, such as the acceptance of a rule or principle
and the deduction of specifi c responses from it. Interestingly, this aspect of virtue
theory shows Aristotle to have been a forerunner of the current application of the
connectionist approach to morality that we described above (see May, Friedman,
and Clark, 1996). In this model, the mind, like the brain itself, is a network that
gets tuned up gradually by experience. With training, the mind does a progressively
better job of recognizing important patterns of input and of responding with the
appropriate pattern of output.
For those who emphasize the importance of virtues in moral functioning, then,
moral maturity is a matter of achieving a comprehensive attunement to the world, a
set of highly sophisticated sensitivities embodied in the individual virtues. Reasoning
and deliberation play important roles in this conception as well; part of being a vir-
tuous person is being able to reason in the right way about diffi cult or problematic
situations. But virtue theory is nevertheless a departure from theories of morality that
see deliberation as the basic moral psychological activity.
Virtue theory posits a particular kind of organization of moral competence,
one in which perception, motivation, action, and reasoning correspond to demands
placed on the person by features of situations. Naturally, the objectivity of these
demands, and the moral relevance of features of situations, are to some degree dic-
tated by the culture, by the moral concepts, social structures, and narratives that are
current in the immediate social context. But this does not mean that the content or
structure of a virtue is completely culturally relative. As Aristotle pointed out, and as
current virtue ethicists have elaborated (Nussbaum, 1993), what it means for a per-
sonality characteristic to be a virtue, and not simply a behavioral regularity, is largely
that it consists in functioning well in a specifi c “sphere of existence.” And what
Aristotle and Nussbaum mean by “sphere of existence” is similar to what evolution-
ary biologists would recognize as persistent adaptive challenges and other types of
environmental constraint. Virtues are therefore quite at home in a scientifi c theory
of moral functioning based on evolutionary psychology and cultural psychology.
The intersection of virtue theory, cognitive science, and the empirical study
of morality is just beginning to be explored in earnest, and many questions remain
open. We have discussed connectionist accounts of cognition and their extension to
social perception and social cognition, because connectionist theorists (Casebeer,
Churchland) have taken a notable interest in social and moral cognition. But it is
quite possible that some other account of cognition is as well suited to describing
and explaining morality.
Our commitment to virtue theory in particular, however, is more fi rm. In our view,
it draws together what is true and useful in the fi ve approaches that we sketched out in
section 3. The child is indeed an active participant in development; moral knowledge
and skills are not just “downloaded” into the child’s mind, as blank slate socialization
theories would have it. Piaget and Kohlberg are correct that there is a substantial ele-
ment of self-construction in moral development. However, what is being learned is best
described as the skills of social perception and reaction discussed by connectionists and
virtue theorists. Most of these skills are about how to interact with other people—how
to fi ll in the three (or four) innately given models for social relationships described
by Fiske. However, some of this knowledge is not about relationships per se; there is
Carruthers Chap19.indd 387Carruthers Chap19.indd 387 8/30/2007 12:08:40 PM8/30/2007 12:08:40 PM
388 Culture, Motivation, and Morality
UNCORRECTED PROOF
also much else that is innate, particularly when we look at the origins of valuation, as
described by Tooby et al. (2005). In addition to being “organized in advance of experi-
ence” for Fiske’s fi rst three models (which involve in-groups, authority, and reciproc-
ity), the mind is also innately prepared to perceive and care about harm from a very
early age (Zahn-Waxler et al., 1979), and also about disgust, purity, and pollution (from
a later age, perhaps not fully until the age of eight; Rozin, Haidt, and McCauley, 2000).
Turiel may have been correct to focus on harm, and the child’s ability to understand
and dislike suffering, as the most important intuition of early moral development. We
believe he was wrong, however, to suggest that children derive all of their other moral
concepts by self-constructing them on this single foundation.
To summarize: The characteristic developmental trajectory in the moral domain
is a movement from crude, global judgments, articulated using a small number
of innate moral intuitions, to highly sophisticated and differentiated perceptions,
beliefs, emotional responses, and judgments. This is consistent with Sperber’s notion
of “teeming modularity”: Domain-specifi c, modulelike intuitive mini-programs give
rise, in the mature moral agent, to an expansive and fl exible set of moral modules
that are more powerful and subtle than the innate modules that compose the fi ve
foundations of intuitive ethics that we have been discussing. Sperber’s approach
suggests that virtues are not themselves innate, but rather are acquired through a
generative process in which the domain-specifi c capacities of the modules that com-
pose the fi ve foundations are multiplied, expanded, and refi ned.
6 The Editing Process: Learning Narratives
Constructivists such as Piaget and Kohlberg called our attention to the ways that
children actively create ever more nuanced understandings of moral issues. Kohlberg
studied the most explicit, discursive, deliberative aspects of moral functioning. Stage
growth was measured through the analysis of verbal reasoning used to justify responses
to hypothetical moral dilemmas. In the cognitive-developmental tradition, moral
thinking was seen as akin to logical thinking; Piaget (1965/1932, p. 398) said explicitly
that “Logic is the morality of thought just as morality is the logic of action.”
It is true that children refl ect on moral questions, particularly when in discussion
with others, and it is attractive to posit a domain-general work space where moral
thinking (as well as other kinds of thinking) is carried out. But must all conscious
verbal moral thinking be logical thinking? Do children really think about moral prin-
ciples and the ways that they do or do not apply to a given situation? There is another
kind of thinking, a different kind of rationality, that seems to play a crucial role in
moral thinking and development. Jerome Bruner (1986) distinguished between the
narrative mode of cognition and the paradigmatic or logico-scientifi c mode:
There are two modes of cognitive functioning, two modes of thought, each pro-
viding distinctive ways of ordering experience, of constructing reality. . . . A good
story and a well-formed argument are different natural kinds. Both can be used
as a means for convincing another. Yet what they convince of is fundamentally
different: arguments convince one of their truth, stories of their lifelikeness. The
Carruthers Chap19.indd 388Carruthers Chap19.indd 388 8/30/2007 12:08:40 PM8/30/2007 12:08:40 PM
The Moral Mind 389
UNCORRECTED PROOF
one verifi es by eventual appeal to procedures for establishing formal and empirical
proof. The other establishes not truth but verisimilitude. (Bruner, 1986, p. 11)
Bruner observes that we know a great deal more about the paradigmatic mode of
thought because cognitive psychology has concentrated its attention on it, whereas
the narrative mode has been comparatively ignored. More important for the pur-
poses of this chapter, he also points out that each mode of thought relies upon its own
“prostheses”—aids to thinking provided by a culture. For the paradigmatic mode,
prostheses include logic, mathematics, and the sciences; for the narrative mode, the
most common prosthetic devices are texts. Texts, among their many other functions,
serve to store up cultural meanings and, through both their content and their struc-
ture, they help to guide the thinking of individuals. We think that moral thinking,
argument, and refl ection (outside of philosophy departments, at least) is much better
described as a kind of narrative thinking than as a kind of paradigmatic thinking.
There are many different kinds of narratives, of varying levels of complexity, and
as a result narrativity (see Carrithers, 1991, for a discussion of this concept) shapes
moral functioning and moral development at several levels of organization. Some of
the most powerful moral narratives are the simplest. For Westerners, parables such
as those found in Jesus’ teaching in the New Testament are familiar examples, and
other religions have similar tools; for example, the hadith, or sayings and doings of
the Prophet Muhammad, in Islam. Shweder and Much (1991) found that narratives
are commonly invoked in Hindu cultures as a mode of moral argument. In their
interviews with Indian informants on moral dilemmas, they often found that ques-
tions about the rightness or wrongness of a particular act elicited a response begin-
ning, “Let me tell you a story about . . .” (for example, stealing). Revealingly, when
Shweder sent interview texts of this kind to Kohlberg for scoring, Kohlberg reported
that much of the interview material was uncodable in his system.
More recently, the sociologist Christian Smith (2003, p. 64) has observed that
we are “animals who make stories but also animals who are made by our stories.”
Smith describes a variety of high-order, often unconscious, narratives that orga-
nize identity and moral judgment at both the individual and the group levels. For
example, he notes that Americans and “militant Muslims” interpret the 9/11 attacks
in the light of very different metanarratives: Americans see things through what
Smith calls “the American Experiment” narrative, in which Americans fl ed the
oppression of the Old World and ever since have been a shining beacon of liberty
and hope, while the “Militant Islamic Resurgence” narrative gives a radically dif-
ferent view in which America has long been a bully and a hypocrite. There are
other narratives, each of which Smith spells out almost like a recipe. Among them
are the “Capitalist Prosperity” narrative, the “Progressive Socialism” narrative, the
“Expressive Romantic” narrative, and the “Scientifi c Enlightenment” narrative.
Smith is especially helpful in making explicit the narratives that motivate
and guide American sociologists and other academics. For example, the “Liberal
Progress” narrative tells the following story:
Once upon a time, the vast majority of human persons suffered in societies and social
institutions that were unjust, unhealthy, repressive, and oppressive. These traditional
societies were reprehensible because of their deep-rooted inequality, exploitation,
Carruthers Chap19.indd 389Carruthers Chap19.indd 389 8/30/2007 12:08:40 PM8/30/2007 12:08:40 PM
390 Culture, Motivation, and Morality
UNCORRECTED PROOF
and irrational traditionalism. . . . But the noble human aspiration for autonomy,
equality, and prosperity struggled mightily against the forces of misery and oppres-
sion, and eventually succeeded in establishing modern, liberal, democratic, capi-
talist welfare societies. [However] there is much work to be done to dismantle the
powerful vestiges of inequality, exploitation, and repression. This struggle . . . is the
one mission truly worth dedicating one’s life to achieving. (Smith, 2003, p. 82.)
This narrative draws heavily on the Harm and Fairness foundations to tell a story of
triumph. It explicitly rejects the Authority foundation as a source of value, portraying
authority and its attendant valuation of tradition as the root cause of evil in the world.
In contrast, the “Community Lost” narrative is more politically conservative; it
relies primarily on the In-group and Authority foundations to tell a story of decline
and decay:
Once upon a time, folk lived together in local, face to face communities where we
knew and took care of each other . . . life was securely woven in homespun fabrics
of organic, integrated culture, faith, and tradition. We truly knew who we were and
felt deeply for our land, our kin, our customs. But then a dreadful thing happened:
Folk community was overrun by the barbarisms of modern industry, urbanization,
rationality, science, fragmentation, anonymity. . . . Faith began to erode, social trust
[to] dissipate, folk customs [to] vanish. . . . All that remains today are tattered vestiges
of a world we have lost. The task of those who see clearly now is to memorialize and
celebrate folk community, mourn its ruin, and resist and denounce the depravities of
modern, scientifi c rationalism that would kill the Human Spirit. (Smith, 2003, p. 84)
Neither narrative is correct in any objective sense. Both are ways that sociolo-
gists have tried to make sense of history. As though employing two different cuisines,
liberal and conservative, academics artfully combine and recombine a few favored
elements. Without the innately given fi ve foundations, there could be no emotion-
ally compelling moral narratives. But without narrative, our moral concepts would
be disjointed and hard to integrate into coherent action plans.
We have included this extended discussion of narrative in this chapter on moral
innateness for two main reasons. First, narrative is a major cultural tool for the modi-
fi cation and socialization of the fundamental intuitions that are at the core of this
chapter. The telling of stories is an indispensable part of moral education in every
culture, and even adult moral discourse frequently reverts to appeals to narratives
as a means of claiming authority. Second, as the reference to Bruner makes plain,
narrative thinking itself is innate and a fundamental aspect of our cognitive archi-
tecture—at least, it is as fundamental as the “propositional” mode of thought. It
seems plausible that human morality and the human capacity for narrativity have co-
evolved, mutually reinforcing one another in our recent phylogenetic development.
7 Conclusion
De Waal (1996) suggests that a building block of human morality visible in chimpan-
zees is the desire for peace and harmony within the group. Celebrations break out
when long-simmering power struggles are resolved. We think this desire is related
Carruthers Chap19.indd 390Carruthers Chap19.indd 390 8/30/2007 12:08:40 PM8/30/2007 12:08:40 PM
The Moral Mind 391
UNCORRECTED PROOF
to the In-group foundation: Group-living creatures prefer (have an innate tendency
to value) harmony within the cooperative groups upon which they depend both for
material sustenance and for intergroup defense. We fi nd this desire in ourselves: We
are a part of the community of morality researchers that has long been divided on
the question of moral innateness. This makes us uncomfortable, for we truly like
and value the many members of our community, and we have tried, in this chapter,
to show how all are right about something, all have something to contribute. We
propose that some degree of harmony and synergy can be restored if most morality
researchers are willing to endorse this statement: The fi rst draft of the moral mind
has diverse moral content that was specifi ed in advance of experience, but this innately
given content gets revised and greatly extended during the course of development as
children actively construct their moral knowledge within a cultural context that uses
narrative to shape and guide the development of specifi c virtues. Is anyone ready to
celebrate with us—or to propose an alternative consensus statement?
Carruthers Chap19.indd 391Carruthers Chap19.indd 391 8/30/2007 12:08:40 PM8/30/2007 12:08:40 PM
UNCORRECTED PROOF
Carruthers Chap19.indd 392Carruthers Chap19.indd 392 8/30/2007 12:08:40 PM8/30/2007 12:08:40 PM
