GROUP SELECTION AND THE LIMITS TO ALTRUISM
Paul H. Rubin
Department of Economics and School of Law
Emory University
Atlanta, GA 30322-2240
Voice: 404-727-6365
Fax: 404-727-4639
Email: prubin@emory.edu
Keywords: evolution, altruism, morality, utilitarianism, Marxism, Rawls, fairness.
JEL D634
Synopsis:
Several evolutionary mechanisms have been identified in the literature that would
generate altruism in humans. The most powerful (except for kin selection) and most
controversial is group selection, as recently analyzed by Sober and D.S. Wilson. I do not
take a stand on the issue of the existence of group selection. Instead, I examine the level
of human altruism that could exist if group selection were an engine of human evolution.
For the Sober and Wilson mechanism to work, groups practicing altruism must grow
faster than other groups. I call altruistic behavior that would lead to faster growth
“efficient altruism.” This often consists of cooperation in a prisoner’s dilemma.
Altruistic acts such as helping a temporarily hungry or injured person would qualify as
efficient altruism. Efficient altruism would also require monitoring recipients to avoid
shirking. Utilitarianism would be an ethical system consistent with efficient altruism, but
Marxism or the Rawlsian system would not. Discussions of efficient altruism also help
understand intuitions about fairness. We perceive those behaviors as “fair” that are
consistent with efficient altruism. It is important to understand that, even if humans are
selected to be altruistic, the forms of altruism that might exist must be carefully
considered and circumscribed.
Introduction
A major debate in biology and in particular in the study of human evolution is the
extent and significance of group selection. Since the work of Williams (1966) it has
generally been thought that any form of group selection was impossible because free
riding within the group would eliminate the possibility. More recently, Sober and D.S.
Wilson (1998) have reignited the debate. They do not propose the old “good of the
species” form of group selection as advocated by Wynne-Edwards (1962) and discredited
by Williams (1966). Rather, they argue that in some circumstances groups containing
more altruists1 would grow faster than groups with fewer altruists, even though the
number of altruists within each type of group would decrease over time. If the groups
remain isolated, then the standard result would obtain, and the number of altruists in each
group and thus in the population would ultimately go to zero. However, if the groups
periodically split and reform, then the number of altruists in the population could
increase. They also argue that this mechanism would have been particularly important in
human evolution, and present some data indicating that human behavior is consistent
with their argument.2
I do not want to address the issue of group selection here. Rather, I am going to
assume for the purpose of argument that there is group selection of exactly the sort that
Sober and Wilson describe. I ask the question, What are the implications of group
selection for the analysis of human altruism? This is consistent with Alexander’s (1987,
p. 192) argument that “morality tends to mimic the effects of group selection.” Also, it
is well known that Darwin (1871) based his views of evolved morality on group
selection, although not of the sort discussed by Sober and Wilson.
The rationale for this examination is as follows: Analysts have identified several
individually based mechanisms for altruism. Some are: reciprocal altruism; kin selection;
the willingness of some to punish others for shirking, or for not punishing those who
refuse to punish shirkers (second degree punishment; Axelrod, 1997; Boyd and
Richerson, 1992; Hirshleifer and Martinez-Coll, 1988); assortment by degrees of
altruism, so that more altruistic persons interact with others with the same behaviors; and
asymmetric interaction, such as Maynard Smith’s (1982) “bourgeois” strategy, in which
the first player to arrive wins the resource.3 Hirshleifer (1999) has a particularly nice
GROUP SELECTION AND ALTRUISM
PAGE 2
discussion of alternative mechanisms for evolution of cooperation, or altruism. But while
all of these mechanisms are plausible and all would generate some altruism, none are as
powerful at generating altruism among non-related individuals as is complete group
selection. Therefore, if we can determine the types of altruism that would be favored by
group selection, we can understand the limits of altruism. If some suggested form of
altruism is not compatible with group selection, then it is unlikely that humans have
evolved a preference for this behavior. Of course, the converse may not be true: if some
mechanism is consistent with group selection, this does not prove that this mechanism
exists unless we are confident that some form of group selection actually occurred or
unless we can identify some mechanism consistent with individual selection to generate
the behavior. Nonetheless, consideration of group selection will give us an “upper
bound” on the level and form of evolved human altruism.
The maintained assumption is that any “tastes” for altruism would have evolved
and we living humans would have those tastes that would have led to increased fitness of
our ancestors in the environment of evolutionary adaptatedness (EEA). We would
perceive these tastes as a sense of “right” or morality, but this feeling has evolved. (This
point is discussed in many places; see for example Ruse and E.O. Wilson, 1986.) Then
humans today can be no more altruistic than it would have paid to be if there was group
selection in the EEA because group selection sets the limits for altruism. This general
approach to political tastes is explained and amplified in Rubin (2000a).
The other point to note is that most agree that the EEA was an environment of
intense group competition. Predation by hominids on hominids or by humans on humans
was a very important factor in the EEA. Many analysts believe that this selection
pressure was responsible for the tremendous increase in intelligence and brain size as our
ancestors evolved from an ape like creature to become humans. Thus, in evolutionary
times, group size was important. That is, having more individuals in the group up to the
limit that could be supported by the existing technology at a given time was a positive
externality for each member. This is a consequence of the “balance of power” argument
that the purpose of sociality was defense against predators – primarily other hominids
(Alexander, 1987).
GROUP SELECTION AND ALTRUISM
PAGE 3
Then there are circumstances in which transfers to some individuals would have
been in the group interest, and presumably in those circumstances the utility or wellbeing of the recipients would have become an element in the utility functions of the
donors. The main point is that in situations where a low cost transfer could save the life
of the recipient and enable him to continue to participate in the group defense, or enable
her to reproduce other defenders, then transfers would have been desired. There are
probably two situations that would have been relevant in the EEA, and moreover which
are still relevant today (because we have tastes evolved in the EEA.) These are
temporary income shortfalls, which could lead to starvation, and illness or injury, which
could lead to death. Both are discussed below.
There is another mechanism, in addition to direct competition, that would have
favored more efficient preferences. Mobility between groups was often possible in the
EEA (as it is today.) Then groups with preferences for more efficient policies would
have been richer or more successful (or fit), and in many cases individuals would have
voluntarily chosen such groups. This raises the issue of cultural rather than biological
selection of preferences. The most useful models of such selection processes allow genes
and culture to interact; see Boyd and Richerson (1985). In their model, there is
evolutionary interaction between genes and culture. For example, if cultural norms in
some group favor some form of altruism, then individuals who most easily adapt to this
form of altruism will be more successful in that group. Cultural and genetic evolution
can reinforce each other. But in this paper I focus on the genetic part of the process.
The analysis here is positive, not normative. I ask what sorts of tastes humans
might have for altruism. In designing optimal policies (that is, in undertaking normative
analysis) it is possible to choose other policies that may not agree with these tastes.
However, decision-makers should be aware of the sorts of constraints under which they
operate. If policies are adopted that conflict with these evolved tastes, then there will be
some discontent with these policies, and they will be more expensive to implement.
There may also be pressure to change the policies. I will from time to time point out
situations where current policies agree with or conflict with evolved tastes.
GROUP SELECTION AND ALTRUISM
PAGE 4
Efficient Altruism
The analysis of group selection leads to the notion of efficient altruism. In the
Sober and Wilson analysis, groups with altruists grow faster, and this explains the
survival of altruism in the population. But only some types of altruism would generate
this differential growth rate. Group selection is not an indiscriminate mechanism for
generating everything that is called or appears to be altruism, or for indiscriminate
“generosity.” Rather, some behaviors that some have called altruistic might actually lead
groups that practiced these behaviors (or that had many individuals practicing these
behaviors) to grow more slowly. For example, Binmore (1994, p. 259) argues that in a
“first best” world (a world with costless enforcement of social contracts) there would be
much more generosity than we actually observe. But the argument here is that natural
selection could not have generated preferences for such behavior in humans. I call
altruism that does lead to increased growth of groups with more such altruists (or that
would have done so in the EEA) “efficient altruism.”
The general example of such efficient altruism would be cooperation in some
variant of a prisoner’s dilemma game. The prisoner’s dilemma is the generic cooperatedon’t cooperate game, where cooperation is efficient for the cooperators jointly in that it
leads to greater payoffs for both participants, but non-cooperation is the best strategy for
each player. Then some group in which more people played “cooperate” would have
greater income and wealth than some group where fewer individuals cooperated, and the
population in the more cooperative would grow more quickly. E.O. Wilson (1998, p.
252) relates morality to cooperation in a prisoner’s dilemma; see generally Binmore
(1994). Sober and Wilson (1998, Chapter 5) show that most or all societies enforce
“social norms.” One important function of such norms is to induce cooperation and
reciprocation – that is, to force a cooperative solution to a prisoner’s dilemma like game.
Cooperation: Examples
One example of efficient altruism would be cooperation in the purchase or
creation of public goods. In the EEA, many things that we think of as public goods
(roads, public buildings) would not have been meaningful. But law and order –
enforcement of rules – would have been a useful goal. Much of the discussion in Sober
GROUP SELECTION AND ALTRUISM
PAGE 5
and Wilson and others of rule enforcement (punishment for rule violation) and second
level rule enforcement (punishment for not punishing rule violations) may be construed
as contribution to a public good. Thus, one form of altruism that may be favored in
human populations is providing institutions or mechanisms for enforcement of rules.
Rules and rule enforcement would lead to larger incomes or increased survival in groups,
and to exactly the sort of differential growth that is required in the Sober and Wilson
model. Rule enforcement in universal among humans (Brown, p. 138).
Cooperation in productive activities would also be an efficient form of altruism.
An example is group hunting. Indeed, Ridley (1996) indicates that big game was the first
public good. Cooperation would be to do one’s share even if the activity is dangerous;
shirking or non-cooperation would be to free ride and not do one’s assigned task. Then if
all cooperate the hunt will be more successful, so this is exactly a prisoner’s dilemma.
Groups with hunters who were more cooperative (or, more generally, workers who were
more cooperative) would grow faster than others, and so this form or altruism could
survive in the EEA in a Sober and Wilson process.
Similarly, if group conflict was an issue in the EEA – and most think that such
conflict has been an ongoing part of human and even prehuman existence – then
contributions to defense or offense would also have been a public good. The main form
of this public good would probably have been rewards to individuals, largely young
males, who participated in defense or in predation, and there seems to be much evidence
that such rewards are in fact forthcoming. Thus, while individuals may have participated
in defensive or offensive “military” activities for private motives, the reason that private
motives would have been available is because there were public rewards to such
participation. Cooperation by young males in offensive or defensive activities (or in their
contemporary surrogates, such as gangs or sports) is a form of altruism that would have
survived a group selection process.
Another form of altruism – one more consistent with the normal usage of the term
– would have been food sharing. Food sharing as a form of insurance could again have
been a form of efficient altruism. If one person is low on food today (perhaps because a
hunt was unsuccessful) he may die. If another shares his food, this will have desirable
effects on the group, since it will lead to a larger group size which is useful for protection
GROUP SELECTION AND ALTRUISM
PAGE 6
(Alexander, 1987). The prisoner’s dilemma occurs because sharing in such
circumstances is efficient for both parties, but the party who first receives the transfer has
an incentive to shirk and not reciprocate. Reciprocal altruism (Trivers, 1971) can make
this policy privately as well as socially desirable, but in either case it would be an
efficient form or altruism. Binmore (1994) points out that reciprocal altruism is another
name for the notion in game theory that cooperative solutions can be reached in repeated
games – the so-called “folk theorem.” Moreover, some of the rules that the group might
efficiently enforce would be rules mandating sharing and reciprocation. A formal analysis
of this problem is in Binmore (1994, pp. 212-226). We would expect humans to agree to
share food, and food sharing is universal among humans: Brown, 1991, p. 138.
But it is important to note that such food sharing would be limited and would
occur only in well-defined circumstances. Sharing creates risks of shirking of two sorts.
One is not sharing when one is obliged to (the EEA equivalent of tax evasion). The other
is shirking in food seeking or production in order to free ride on those who do produce
and share (the EEA equivalent of welfare fraud). Kelly (1995, Chapter 5) discusses these
issues and shows that sharing, shirking, and detection of shirking all exist among huntergatherers; see also Boehm (1999a). Then food sharing would not be efficient if the
recipient were lacking in food because of shirking. In the EEA, societies were relatively
small, and monitoring to reduce shirking would have been possible. Rubin (1982) stresses
the difficulty in monitoring shirking under current conditions as compared with
conditions in the EEA. Cosmides and Tooby (1992) point out that this issue still informs
debate on welfare and other contemporary sharing systems. Proponents of such programs
stress the random nature of poverty and the role of luck4; opponents stress shirking. Both
sides are trying to capitalize on evolved tastes for or against redistribution in particular
circumstances. The difficulty is in trying to design a system of transfers in a world where
monitoring of recipients to detect shirking is difficult. McGuire (1992) points out that
there are mental mechanisms for calculating the appropriate amount of moralistic
aggression associated with differing forms and degrees of shirking.
In addition to food sharing, assistance to the ill, hurt or injured who would be
expected to recover would have been an efficient form of altruism. This is because
someone hurt or injured could, upon recovery, again become a useful member of the
GROUP SELECTION AND ALTRUISM
PAGE 7
band. Injury was probably much more common in the EEA than today; injury has been
falling steadily in historic times, and is lower now than even in our own recent past.
Methods of production in the EEA, such as hunting, would have been particularly prone
to injury. Even today many view medical care as a “necessity” and are opposed to
treating medical care as simply another expenditure item; this may be because such a
taste for treating medical care as different could have evolved in the EEA. Moreover, the
nature of care in the EEA was probably such that shirking in the form of consumption of
additional care would not have been an alternative. This may be why people today think
in terms of the amount of medical care “needed” without realizing that the amount of care
consumed can be increased if costs are reduced.
None of these forms of altruism would have supported providing food or other
resources to a non-productive individual who was not expected to become productive, or
to a shirker who was expected to continue to shirk. That is, altruism might have
supported transfers to a temporarily unlucky individual or to an individual who was
temporarily incapacitated through illness or injury. But it would not have supported
transfers to someone who would be expected to remain unproductive. (Kin selection
mechanisms might have induced family members to provide some support, for example,
to the elderly, and there is evidence that children support aged and unproductive parents
in hunter gatherer societies, as long as such support does not lead to excessive
deprivation for children or grandchildren.) Continual transfers to non-productive
individuals would have been a drain on a band and would have reduced the ability to
compete with other bands in the EEA. Boehm (1999a; 1999b) indicates that among
hunter-gatherers older individuals who cannot walk are often killed or abandoned.
Westermarck (1912) has a lengthy discussion of this issue in Volume 1, Chapter 17, and
discusses the killing of certain sick persons as well. Binmore (1994, p. 259) asks why it
is that people seem to feel less empathy towards schizophrenic homeless persons. These
attitudes and behaviors are consistent with the argument here.
There is substantial evidence that in hunter-gatherer societies, there was what has
been called an “egalitarian ethic.” (Boehm, 1993, 1999b; Knauft, 1991; Wiessner, 1996.)
This might seem to imply that we have inherited from our hunter-gatherer ancestors a
desire for income transfers and equality of outcome. But careful reading of Boehm
GROUP SELECTION AND ALTRUISM
PAGE 8
(especially 1999b) indicates that what was actually involved was a desire not to be
dominated; this is consistent with Knauft’s analysis, although Knauft does not state this
as directly as does Boehm. That is, individuals who expected to be the lower ranked
members of any hierarchy banded together to prevent potential higher ranked members
from actually forming a hierarchy. Food sharing (especially of meat) was virtually
universal in such societies, but care is taken not to use such sharing to provide excess
status or power to successful hunters (Wiessner, 1996). There is no evidence that
potentially dominant members wanted to eliminate hierarchies. Rather, individuals
wanted to avoid being dominated. Humans still have such preferences. For example,
Buchanan and Tullock (1965) provide a theory of constitutions based on this desire.
Bentham, Yes; Marx and Rawls, No
Three standard philosophical systems that discuss the optimal moral stance for
society are utilitarianism (Bentham, 1781/1988), Marxism (Marx, 1888) or, more
generally socialism, and the Rawlsian system (Rawls, 1971). Utilitarianism is roughly
consistent with the form of altruism that would occur under group selection; the Rawls
“difference principle” is not, nor is a communist or socialist system.
My analysis here is obviously sketchy and uninformed by academic discussions
of moral theory. However, Posner (1999) indicates that such discussions are not very
fruitful anyway. E.O. Wilson (1998) argues that moral philosophy is not generally based
on empirical biological knowledge. Alexander (1987, p. 145-165) discusses the
relationship between moral philosophy and biology as viewed by moral philosophers. He
concludes that in general moral philosophers have not paid adequate attention to biology,
and have not taken account of biological knowledge. Earlier, Westermarck (1932) had
reached a similar conclusion regarding an earlier generation of moral philosophers.
There is also a body of literature by biological informed writers applying
biological knowledge to moral discourse. Westermarck (1912, 1932) was an early
scholar working in this vein, although his writing is impaired by the limited knowledge of
biology available when he was writing; for example, he commonly writes of things
“useful for the species.” Westermarck does sometimes refer to natural selection (as in the
famous discussion of incest avoidance based on youthful propinquity: Westermarck,
GROUP SELECTION AND ALTRUISM
PAGE 9
1912, Volume II, Chapter 40) but in general his method of analysis is comparative rather
than evolutionary. This may be because the analytical tools needed to derive specific
explanations (e.g., theories of kin selection and of reciprocal altruism) were not available.
I believe that Alexander (1987) is the best example of this literature, and my writing here
and elsewhere is deeply informed by his analysis. Binmore (1994) is also an analysis of
similar issues, based on both evolutionary theory and game theory. Boehm (1999b) has a
discussion of the evolved human desire to morally justify particular actions. I will refer
to this literature from time to time. However, I do not believe that the basic point of this
article has been made within this literature.
Bentham
Something approximating utilitarianism might well be consistent with efficient
altruism. Utilitarianism argues for maximizing some function of individual utilities.
Utility functions are related to (if not the same as) fitness (Hirshleifer, 1977): we get
pleasure from those things that led to increased reproductive success in the EEA and pain
from those things that hindered our ancestors’reproductive success (Johnson, 1999).
Thus, maximization of the sum of individual utilities would be the equivalent of
maximizing group fitness in the EEA, and would have exactly been the result of efficient
altruism in a group selection process. This may be why utilitarianism has been a long
lived and successful moral theory – it is a theory that is consistent with our evolved moral
preferences.
Economists use an essentially utilitarian theory in welfare economics. But there
is a deep professional belief in the impossibility of “interpersonal comparisons of utility.”
That is, economists are unwilling to compare utilities across persons. The argument
made here will at least partially solve the problem of interpersonal comparison. That is, ,
we can in some circumstances make interpersonal comparisons of utility based on
efficient altruism. In particular, in those circumstances discussed above, where the
potential recipient of altruism can have his life saved by a relatively small transfer, then
humans are often willing to make the transfer, and are at least implicitly making an
interpersonal comparison.
GROUP SELECTION AND ALTRUISM
PAGE 10
Posner (1999) indicates that most academic criticisms of utilitarianism proceed by
showing that the logical implications of utilitarianism are undesirable if the argument is
carried to its logical extremes. But the argument discussed here is that utilitarianism is
essentially the result of fitness maximizing preferences within the relevant group. On this
reading, any implications of utilitarianism that conflict with fitness maximization for the
group (for example, the sadist who gains great utility from torture) are illegitimate
extensions of the theory, and should be ignored. Utilitarianism would perhaps be
modified to maximization of utility in terms that would have been consistent with fitness
maximization for members of the group in the EEA. Of course, we now extend our
altruistic preferences beyond the level of the group, so that some fitness maximizing
behaviors – for example, massacre of males of neighboring tribes – would no longer be
considered consistent with utilitarianism. J.Q. Wilson (1993) and Binmore (1994) among
others discuss tendencies to expand the set of humans included within the group. Indeed,
Westermarck (1932) makes a similar point.
Alexander (1987) discusses utilitarianism. He indicates (p. 88) that “The only
time that utilitarianism … is predicted by evolutionary theory is when the interests of the
group … and the individual coincide… ” Alexander’s argument differs from that made
here in two respects. First, he is discussing the extent to which an individual will behave
in a utilitarian manner; I am concerned with utilitarianism as a social philosophy.
Second, Alexander is hostile to group selection arguments (although in some of his
discussion “generalized reciprocity” acts almost like group selection), and I am
proceeding here as if such mechanisms were feasible. If they are not and if there is no
individual selection mechanism that will generate some version of utilitarianism, then it
may be a difficult philosophy to justify in biological terms.
Rawls
Consider Rawls. Rawls argues that humans behind a “veil of ignorance” would
choose an income distribution system based on the “difference principle.”5 This says that
any inequality in income distribution can be justified only if it benefits the worst off
individuals in society. By arguing that this is the rule individuals would choose behind
the veil, Rawls is arguing that people are exceedingly risk averse. A huge amount has
GROUP SELECTION AND ALTRUISM
PAGE 11
been written about Rawls’analysis, and I do not plan a thorough analysis. But note that
this principle is not consistent with efficient altruism as defined above. A group adopting
this principle would not have increased its fitness.
Consider: Group A does not adopt Rawls’principle. All active adult males have a
nutritional level of 2500 calories per day except for one individual who obtains 2000 and
is somewhat malnourished (but perhaps not in danger of immediate death). Group B
begins with the same distribution but adopts the difference principle and redistributes
accordingly. The efficiency costs of redistribution (deadweight losses due to “taxation”
in the form of forced transfers and shirking in he expectation of receiving the benefits of
redistribution) are so high that once the redistributive process has run its course total
income falls. As a result, at equilibrium, everyone in the group has a nutritional level of
2100 calories per day. According to Rawls, Group B is morally superior to Group A
because the poorest person in B has 2100 calories and the poorest in A has only 2000.
But even though Rawls and his disciples may prefer this outcome, it is not the outcome
that would be selected in the EEA. Rather, members of Group A would translate their
extra nutrition into increased strength or more hand axes and likely massacre the males
(at least) in Group B. (Or, in a more benign scenario, all members of Group B except for
the original poorest person would migrate to Group A.)
In other words, the costs of Rawls’policy is so large that it would not meet the
Sober and Wilson criterion for being the outcome of a group selection process, and so it
is unlikely that humans have been selected to desire Rawls’preferred outcome. That is,
members of Group B would not be our ancestors and their set of tastes would not have
survived. We are descended from people or proto-humans with tastes more like those of
Group A.6 Moreover, tastes are exactly relevant for analyzing Rawls’analysis, since he
asks what people in the original position would choose – an exact question of tastes.
Alexander (1987) has a discussion of Rawls. However, he confines his analysis
to a discussion of the “veil of ignorance,” arguing that this may be an appropriate
construction in a society in which individuals are less able to foresee the positions of their
descendents. He does not discuss in detail the hypothesized outcome of the process.
Wilson (1998, p. 249) does indicate that Rawls’ “offered no evidence that justice-asfairness is consistent with human nature.”
GROUP SELECTION AND ALTRUISM
PAGE 12
Marx
A socialist system divorces returns from inputs. Such systems are inefficient and
lead to reductions in output. As such, in a highly competitive environment such as the
EEA, they could not survive. Indeed, extreme socialist regimes apparently cannot
survive even in our relatively more relaxed environment. Moreover, tastes for socialism
would not have evolved since those with such tastes would not have done well. Bailey
(1992) has shown that efficient definitions of property rights are virtually universal in
hunter-gatherers. Brown (1991, p. 139-140) indicates that private property and
inheritance are universal among humans. Boehm (1997, p. S117) argues that socialism is
a form of altruism but one that is “so very difficult to execute.” The difficulty is
associated with its inefficiency, and more generally with evolved preferences that are
inconsistent with this social arrangement. Elsewhere Boehm (1999b) argues that Marx
failed by ignoring the power of the human desire for dominance. (Other evolutionary
based criticisms of Marx are presented in Rubin, 2000b. This argues that Marxism is an
error associated with confusion of types of hierarchies. It is also pointed out that
Marxism is a form of group selection, with class, or, in neo-Marxism, race, class and
gender as the group characteristics.)
Fairness
Humans have a preference for “fairness.” Trivers (1971) indicates that we feel
“moralistic aggression” if we feel that we have been treated unfairly; this is part of the
mechanism for policing reciprocal altruism. Westermarck (1912) discusses a similar
point, which he calls “moral indignation.” Binmore (1994) argues that our notions of
fairness are an equilibrium selection device – a method of selecting a particular Pareto
improving equilibrium when many are available. But the concept of fairness is not well
defined. Lakoff (1996, pp. 60-61) has an interesting categorization of various views of
fairness. He suggests that there are at least 10 ways in which the term can be used:
1)
2)
3)
4)
Equality of distribution (one child, one cookie)
Equality of opportunity (one person, one raffle ticket)
Procedural distribution (playing by the rules determines what you get)
Rights-based fairness (you get what you have a right to)
GROUP SELECTION AND ALTRUISM
PAGE 13
5)
6)
7)
8)
9)
Need-based fairness (the more you need, the more you have a right to)
Scalar distribution (the more you work, the more you get)
Contractual distribution (you get what you agree to)
Equal distribution of responsibility (we share the burden equally)
Scalar distribution of responsibilities (the greater your abilities, the greater
your responsibilities)
10) Equal distribution of power (one person, one vote)
Lakoff does not provide any basis for this list of meanings or intuitions. (Indeed,
Carroll, 1999, has recently provided a general critique of Lakoff’s work indicating that
the work is not grounded in evolutionary theory, and would be improved if it were.) But
it is easy to see that some of Lakoff’s categories are consistent with the evolutionary
explanation for altruism that is proposed here. That is, under proper circumstances many
of Lakoff’s categories could have been associated with a form of efficient altruism. (Of
course, under different circumstances, many of these categories could be mutually
inconsistent, which is why economists generally avoid the use of the concept of fairness.)
Moreover, the actions associated with many of these concepts are universal among
humans (Brown, 1991.) I will briefly consider each.
Equality of distribution: This would be consistent with food sharing, which, as
discussed above, is sometimes an efficient form of altruism. Some food sharing,
although not equality of distribution, is universal among humans (Brown, 1991, p. 138.)
Full equality, however, would not be an efficient form of altruism because it would lead
to excessive shirking.
Equality of opportunity: This would probably have been the norm in the EEA. It
would also have served to reduce the power of dominant individuals by reducing their
ability to pass on their success to their children. Among mobile hunter-gatherers and
their predecessors, hereditary wealth would have been minimal (since the mobile nature
of the society meant that wealth in general was minimal), so there would have been a
tendency towards de facto equality of opportunity. Today the term is used loosely: no
one (in America at least) advocates true equality of opportunity, since this would mean
removing all children from their parents at birth and raising them communally.
Otherwise, differential parental abilities in child raising will effect children. Moreover,
children can of course inherit wealth in our society.
GROUP SELECTION AND ALTRUISM
PAGE 14
Procedural distribution: At least to some extent, playing by the rules would have
been important. Brown (1991, p. 138) indicates that rules are universal among humans
and that sanctions for violations are also universal. If the rules are themselves aimed at
forcing cooperation in prisoner’s dilemmas and at enforcing reciprocity as Sober and
Wilson indicate, then tying income distribution to obeying rules would be efficient.
Rights-based fairness: If rights are defined efficiently, then this form of fairness
would also be efficient. Private property rights are of course an efficient way to define
rights (e.g., Pipes, 1999). Brown (1991, p. 139-140) does indicate that ownership of
property and inheritance of property are universal among humans. Bailey (1992) shows
that among hunter-gatherers, property rights are generally defined efficiently. To the
extent that rights based fairness is a form of property rights based fairness, then we can
understand this notion and associated preferences.
Need-based fairness: This is consistent with an insurance model of sharing.
Those who have a shortfall in a period receive transfers. This will keep them alive as
potential fighters when needed. This form of behavior may also be part of a process of
reciprocal altruism. Of course, for this form of altruism to be efficient, some monitoring
to reduce shirking is needed. Moreover, “need” would be carefully circumscribed,
perhaps including only being at risk of or close to immediate death.
Scalar distribution: This is an efficient principle of distribution because it
provides incentives for work. In many hunter-gatherer societies, the hunter who kills the
game has rights to distribute it, subject to various social norms. If this were the pattern in
the EEA (and chimpanzees also behave in this way, suggesting that it is an evolutionary
old behavior in the hominid line) then it might be an evolved principle of distribution and
a form of efficient altruism.
Contractual distribution: This would be equivalent to reciprocal altruism, an
agreement to return a gift, and would be an efficient method of solving intertemporal
prisoner’s dilemmas. A norm of reciprocity is universal among humans (Brown, 1991, p.
139). Of course, today we have third party (court) enforcement of contracts.
Equal distribution of responsibility: This principle would have served to constrain
dominant individuals in society, and to create more equality.
GROUP SELECTION AND ALTRUISM
PAGE 15
Scalar distribution of responsibilities: This seems to have been an important
principle in hunter-gatherers. Better, more skilled hunters do catch more game even
though they are obligated share much of their catch with others. (Kelly, 1995, p. 177179.) Because of this enforced sharing, the willingness of skilled hunters to catch more
game is itself somewhat of a puzzle. Kelly provides some hypotheses to explain the
behavior, but ultimately the reason seems to be a mystery. Nonetheless, such behavior
would be an efficient form of altruism and could be consistent with a group selection
model. Of course, today, in a market economy rewards are relatively closely related to
effort.
Equal distribution of power: As Boehm (1993, 1999b) and Knauft (1991) have
argued, in small hunter-gatherer bands, which were probably the most important human
and hominid population structure in the EEA and for most of the existence of humans and
our immediate ancestors, there is something like a “reverse dominance hierarchy.” This
is interpreted to mean that no individual is dominant and decisions are jointly made.
Brown (1991, p. 138) indicates that no humans have perfect democracy and none have
complete autocracy, so “they always have a de facto oligarchy.” This principle would be
associated with reducing the power of dominant individuals, and with Boehm’s “reverse
dominance hierarchy.” While many human societies have been centrally run for the
benefit of dominant individuals (Betzig, 1986) there seems to be a preference for
relatively democratic societies when possible.
Thus we see that at least five of Lakoff’s principles are associated with efficiency:
procedural distribution; rights based fairness; scalar distribution; contractual distribution;
and scalar distribution of responsibilities. Two principles are associated with insurance:
equality of distribution and need based fairness. The other three principles, equality of
opportunity, equal distribution of responsibility, and equal distribution of power, have the
effect of reducing the power of dominants. Our intuitions about fairness seem to be
consistent with evolution of tastes in the directions discussed in this paper. Of course, as
mentioned above, it is easy to construct examples where these principles conflict with
each other. For some examples, equality of opportunity can lead to inequality of
distribution, and either procedural distribution or rights based fairness can be inconsistent
with need-based fairness.
GROUP SELECTION AND ALTRUISM
PAGE 16
This list and the discussion may seem trivial. But this is because the notions are
so intuitive that it is not easy to see how there could exist alternatives. Consider some
potential notions of fairness or morality that no one would endorse: Take from the poor
and give to the rich. Cheaters who cheat fellow group members deserve as much as they
can get away with. Promising to return a favor and then reneging is moral.
Compensation should be inversely related to effort. Good hunters should stay at home
and poor hunters should have the responsibility for bringing in game. Each person
should do that at which he is worst (has a comparative disadvantage). The least
competent person, perhaps a child, should make decisions for the group. Dictatorship is a
moral form of government. No good deed should go unpunished.7
Obviously, these are silly principles, and clearly not intuitively “fair” in any
sense. But that is the point: they do not seem fair or desirable, and they would have been
associated with group and individual death in the EEA. If any of our putative ancestors
had beliefs such as these, they did not survive to be our ancestors. We have those
intuitive notions of fairness that led to survival, and we still view these behaviors as fair.
J.Q. Wilson (1993) also categorizes various measures of fairness: fairness as
equity; fairness as reciprocity; and fairness as impartiality. This categorization is less
nuanced than that of Lakoff, but consistent with it: reciprocity is the basis of Lakoff’s
third, fourth, and seventh principle; equity is consistent with the first, second, and
perhaps eighth point. Lakoff does not explicitly mention impartiality, but it is implicit in
his analysis. But, unlike Lakoff, Wilson does explicitly relate his notions of fairness to
evolved preferences, and in particular to their value in a social species. Indeed, Wilson
generally relates his notions of a moral sense to evolution, in a way consistent with the
arguments here.
Summary and Implications for Further Research
I assume for the sake of argument that group selection was involved in human
evolution, including evolution of altruistic preferences. Nonetheless, there would still be
limits to the amount of altruism that could have evolved. Altruism in contemporary
group selection models is associated with faster growth of altruistic groups. This can
occur if such altruism leads to cooperation in some prisoner’s dilemma. I call altruism
GROUP SELECTION AND ALTRUISM
PAGE 17
that would lead to faster growth “efficient altruism.” A common form of efficient
altruism would be saving the sick or injured, or providing food to a hungry or starving
person who is down on his luck. But such altruism must also be associated with
monitoring the recipient to avoid free riding, for societies that allowed excessive shirking
and free riding would not have grown as fast as others and in the highly competitive EEA
would not have survived. Utilitarianism as a moral system is approximately consistent
with efficient altruism; Marxism or the Rawlsian systems are not. The notions of
efficient altruism and of evolved political preferences also help explain human intuitions
about fairness.
It would be useful to explore the limits of efficient altruism. It would also be
useful to understand the relation of efficient altruism to other ethical systems, in addition
to those sketched out here. It might also be worth working out the implications of
utilitarianism if we understand that utility functions have evolved. From this basis, it
might also be possible to derive a more evolutionarily based system of morality. While
we might decide to adopt a moral system that is not consistent with our evolved
preferences, it is still useful to understand these preferences. For example, if we choose
policies that are too inconsistent with these preferences, then costs of enforcing these
policies might be very high.
Acknowledgments: The author would like to thank Frank Salter and an anonymous
referee for helpful comments, while accepting blame for all errors.
GROUP SELECTION AND ALTRUISM
PAGE 18
References cited
Alexander, Richard D. 1987. The biology of moral systems. Aldine de Gruyter, New
York.
Axelrod, Robert. 1997. The complexity of cooperation: agent-based models of
competition and collaboration. Princeton University Press, Princeton, Chapter 3:
Promoting Norms.
Bailey, Martin J. 1992. Approximate optimality of aboriginal property rights. Journal of
Law and Economics. 35:183-198.
Barkow, Jerome H., Leda Cosmides and John Tooby. 1992. editors, The adapted mind.
Oxford University Press, New York.
Bentham, Jeremy (1781/1988), The principles of morals and legislation. Prometheus
Books, Amherst, NY.
Betzig, Laura L. 1986. Despotism and differential reproduction: a Darwinian view of
history. Hawthorne. Aldine de Gruyter, New York.
Binmore, Ken. 1994. Game theory and the social contract II: just playing. MIT Press,
Cambridge, MA.
Boehm, Christopher. 1993. Egalitarian behavior and reverse dominance hierarchy.
Current Anthropology. 14:227-254 (including comments and reply).
Boehm, Christopher. 1997. Impact of the human egalitarian syndrome on darwinian
selection mechanics. American Naturalist, 150 (Supplement):S100-S121.
Boehm Christopher. 1999a. The natural selection of altruistic traits. Human Nature.
10:205-253.
Boehm, Christopher. 1999b. Hierarchy in the forest: The evolution of egalitarian
behavior. Harvard University Press, Cambridge.
Boyd, Robert and Peter J. Richerson. 1985. Culture and the evolutionary process.
University of Chicago Press, Chicago.
Boyd, Robert and Peter J. Richerson. 1992. Punishment allows the evolution of
cooperation (or anything else) in sizeable groups. Ethology and Sociobiology.
13:171-195.
Brown, Donald E. 1991. Human universals. McGraw Hill, New York.
Buchanan, James and Gordon Tullock. 1965. The calculus of consent. University of
Michigan, Ann Arbor.
Carroll, Joseph. 1999. The deep structure of literary representations. Evolution and
Human Behavior. 20:159-173.
Cosmides, Leda and John Tooby. 1992. Cognitive adaptations for social exchange, in
Jerome H. Barkow, Leda Cosmides and John Tooby (1992), editors, The adapted
mind. Oxford University Press, New York.
GROUP SELECTION AND ALTRUISM
PAGE 19
Darwin, Charles. 1871/1981. The descent of man and selection in relation to sex.
Princeton University Press, Princeton.
Hirshleifer, Jack. 1977. Economics from a biological viewpoint. Journal of Law and
Economics. 20:1-52.
Hirshleifer, Jack. 1999. There are many evolutionary pathways to cooperation. Journal of
Bioeconomics. 1:73-93.
Hirshleifer, Jack and J. Martinez-Coll. 1988. What strategies can support the evolutionary
emergence of cooperation? Journal of Conflict Resolution, 32: 367-398.
Johnson, Victor S. 1999. Why we feel: the science of human emotions. Perseus Books,
Reading, MA.
Kelly, Robert L. 1995. The foraging spectrum: diversity in hunter-gatherer lifeways.
Smithsonian Institution Press, Washington, D.C.
Knauft, Bruce. 1991. Violence and sociality in human evolution. Current Anthropology.
32:391-428 (including comments and reply).
Lakoff, George. 1996. Moral politics: what conservatives know that liberals don’t.
University of Chicago Press, Chicago.
Marx, Karl. 1888. Manifesto of the Communist party (web edition: Project Guttenberg).
Maynard Smith, John. 1982. Evolution and the theory of games. Cambridge University
Press, New York.
McGuire, Michael. 1992. Moralistic aggression, processing mechanisms, and the brain. in
Roger D. Masters and Margaret Gruter, eds., The sense of justice: biological
foundations of law. Sage Publications, Newbury Park.
Pipes, Richard. 1999. Property and freedom. Knopf, New York.
Posner, Richard. 1999. The problematics of legal and moral theory. Belknap Press of
Harvard University Press, Cambridge, MA.
Rawls, John. 1971. A theory of justice. Belknap Press of Harvard University Press,
Cambridge, MA.
Rubin, Paul H. 1982. Evolved ethics and efficient ethics. Journal of Economic Behavior
and Organization. 3:161-74.
Rubin, Paul H. 2000a. The state of nature and the evolution of political preferences.
American Law and Economics Review. (In press).
Rubin, Paul H. 2000b. Hierarchy. Human Nature (In press).
Ruse, Michael and Edward O. Wilson. 1986. Moral philosophy as applied science.
Philosophy. 61:173-192.
Sober, Elliott and David Sloan Wilson. 1998. Unto others: the evolution and psychology
of unselfish behavior. Harvard University Press, Cambridge, MA.
Trivers, Robert L. 1971. The evolution of reciprocal altruism. Quarterly Review of
Biology. 46:35-57.
GROUP SELECTION AND ALTRUISM
PAGE 20
Westermarck, Edward. 1912. The origin and development of the moral idea, (2 Volumes,
Second Edition). Macmillan and Co. London.
Westermarck, Edward. 1932. Ethical relativity. Harcourt Brace and Co. New York.
Wiessner, Polly 1996. Leveling the hunter: constraints on the status quest in foraging
societies, in Polly Wiessner and Wulf Schiefenh¬ vel., eds., Food and the status
quest: an interdisciplinary perspective. Berghan Books Providence, RI.
Williams, George C. 1966). Adaptation and natural selection: a critique of some current
evolutionary thought. Princeton University Press, Princeton.
Wilson, Edward O. 1998. Consilience: the unity of knowledge. Knopf, New York.
Wilson, James Q. 1993. The moral sense. Free Press, New York.
Wynne-Edwards, Vero Copner. 1962. Animal dispersion in relation to social behaviour.
Hafner Publishing Company, New York.
GROUP SELECTION AND ALTRUISM
PAGE 21
Endnotes
1
Hirshleifer (1999) indicates that “altruism” is a measure of attitude, and what is relevant is behavior –
what he calls “helping behavior.” Here, for consistency with the majority of the literature, I will follow the
standard usage, although Hirshleifer’s terminology is more appropriate.
2
Boehm (1999b) also proposes a form of group selection, based on mechanisms different from those
suggested by Sober and Wilson. However, the arguments in this paper would apply to Boehm’s
mechanism as well.
3
Sober and Wilson define group selection rather broadly, so that any individuals interacting, however
briefly, may constitute a “group.” By their definition any of these mechanisms would be a form of group
selection. I do not need to address this issue here.
4
Binmore, 1994, p. 259: “It is true that a substantial minority of people on the street are mentally ill, but
the majority are just plain unlucky.” No source is given for this assertion. Binmore appears to be an
advocate of increased generosity towards street people.
5
I do not discuss the mechanism of the “veil of ignorance” itself. For an elaborate discussion, see Binmore
(1994). However, the results of deliberations behind the veil must be hypothetical. This is because we are
never behind such a veil. It is also true that different people will intuit different outcomes of such a process
– for example, Rawls and myself. Therefore, the notion of the veil is difficult to operationalize.
6
Interestingly, Sober and Wilson (1998, p. 237) discuss Rawls and the difference principle as an example
of a moral rule, but do not consider the consistency of this rule with their view of evolved altruism.
7
This last is a stock joke phrase; the joke is in the violation of intuitions about ethics.