锋利的利他主义的另一端

 原著 Holly Arrow, 韩顺生 译
 (原文: Holly Arrow, The Sharp End of Altruism. Science 26 October 2007:Vol. 318. no. 5850, pp. 581 – 582)
这两种社会你更喜欢哪一个: 一个社会由自私而宽容的自由交换者组成, 另一个由相互帮助的斗士组成. 如果你认为利他主义和宽容更有价值, 那么它们都不完美. 在Choi 和 Bowles用计算机模拟的世界里, 宽容且利他的社会是少见的 (1). 相反, 利他主义总是和对外敌对相互促进, 而战争是这个过程的动力也是结果. 对于 同胞, 利他主义者不怕牺牲, 是 侠客; 而外族人所遇到的是利他主义行为锋利的另一端.
从进化的角度看, 利他主义是难以理解的. 有人用血缘的利他主义 (2) 和互利的利他主义 (3) 来解释这个迷. 就是说去帮助有血缘关系的人或者会给回报的人. 两者都不能解释在不能得到回报时, 人们还会以很大的代价帮助素不相识的人. 战争英雄是一个例子, 用他或他的亲属可以得到间接利益来解释这种极度的利他主义是很难的.越来越多的研究工作试图用对个体和族群共同的生存选择来解释利他主义.
在这样多层次的模拟过程中 (4), 进化的结果决定于个体竞争与族群竞争的相对影响. 对个体的选择促使利他主义的行为从基因库中消失. 对族群的选择产生相反的作用, 即相对由更多自私人组成的族群, 由更多利他主义者组成的族群更利于生存. 对于大多数物种, 对个体的选择更重要. 然而强大的族群选择力对人类更适用. 有证据表明族群间的暴力杀死了我们人类不少的祖先 (5), 这使人怀疑战争是强而有力的族群选择动力. 由于人们根据是否属于同一族群在战争中相互残杀, 所以战争是很强的候选因素.
在 Choi 和 Bowles 的模拟中, 二十个人群一起生存数千代. 每一个成员有二个基因, 每一个基因有二个类型. 一个基因要么是宽容 (T), 要么是 地域性 (P). 另一个基因要么是利他的 (A) 要么不是 (N). 除了少数突变, 后代都继承父母的特征. 利他主义者以自己的代价帮助自己族群的成员. 非利他主义者不会这样作. 宽容的成员与外族人交换而获利. 地域性的成员却不会. 在一个族群中, 高比例的地域性成员会限制所有成员的交换机会.
在所有可能的四种特征组合中, TN是最有利的, 这些自我本位的商人从外族人渔利 并接受利他主义者的捐献. 牺牲最大的组合是PA. 这些慈善的斗士捐献给同袍, 并冒险保护平民, 去给同族的后代征服新的领地. 个体选择有利于TN组合. 而战争的顺利则有利于由更多PA成员组成的族群.其他两个特征的组合是PN 恶霸, 他们即自私又仇视, TA慈善家, 他们与其他人交换, 也捐献给别人.
在每一代, 这些族群随即配对, 下面的结果决定于宽容型成员与斗士型成员的比例.如果两个族群都主要由宽容型成员组成, 宽容型成员从交换中获益.  如果在任何一个族群中宽容型成员不占大多数, 那么和平交换的可能就急剧下降. 代而替之的是无效的对峙或是战争. 如果两个族群含有相当数目的斗士, 结果就是对峙. 力量越不平衡, 战争的可能就越大. 战争可能一方得胜也可能 平分秋色. 一部分斗士会因战争而死亡. 一方胜利后, 另一方的平民也会被杀. 战后, 胜方大量生育, 后代迁徙到征服的土地上.
这些进化的族群在两种情况下达到稳定: 或者当自私的商人 (TN) 占绝大多数, 或者当慈善的斗士 (PA) 占绝大多数. 少数的恶霸 (PN) 和更少的慈善家 (TA) 可以共存于中. 交换的社会是和平的, 而对峙与战争在斗士的社会更常见, 即使不太经常的战争, (10% – 20% 的遭遇) 也可以维持相当高度的地方性利他主义. 间断性战争对英雄主义的影响 (6) 同样表明战争作为一个强大的选择力不需要持续进行.
Choi 和 Bowles 的模拟结果--利他主义与地方主义的结合—是与行为研究的结果相一致的. 比如, 在两难选择的实验中, 如果游戏中存在多个小组, 自私的选择就会大大减少 (7). 对外族的存在的意识可以促进利他主义的行为, 这种现象更多见于女性 (8). 这也与战争对作为斗士的男人有更大的选择压力相一致. 有意思的是, 女人的利他主义尽管相对不受族群间敌对的影响, 也仍然是很高的. 利他主义的进化途径在两性中是不同的.
这个进化途径的所有因素中应该包括文化的进化. 在另一个研究中, Bowles 等显示社会规范能通过促进同一而支持利他主义 (9). 在目前的模拟中, 斗士占多数的族群实施贸易壁垒. 但是这个是预先设置的. 自由贸易能不能胜过独立主义的地区保护? 惩罚懦夫的规范是否和战争与利他主义共同进化?
这个模拟的发现启示战争的遗产是宽容与利他主义间的不相容. 然而, 跨文化的研究使人乐观. 在一个研究中, 来自十五个社会的人玩捐献的游戏 (10). 平均慷慨的程度与一个社会的市场交换总量, 经济合作相关. 通过在模拟中加入可变的规范, 可以进一步探索宽容的利他主义的社会的潜在活力.
对相互交织的人类的要给与, 要交换, 要攻击外族人的冲动的理解. 能帮助我们提倡有利于社会的行为, 同时把利他主义的锋利的另一端放入鞘内.
参考文献:
1.    J.-K. Choi, S. Bowles, Science 318, 636 (2007).
2.    W. D. Hamilton, J. Theor. Biol. 7, 1 (1964).
3.    R. L. Trivers, Q. Rev. Biol. 46, 35 (1971).
4.    E. Sober, D. S. Wilson, Unto Others: The Evolution and Psychology of Unselfish Behavior (Harvard Univ. Press, Cambridge, MA, 1998).
5.    L. H. Keeley, War Before Civilization; The Myth of the Peaceful Savage (Oxford Univ. Press, New York, 1997).
6.    H. Arrow, O. Smirnov, J. Orbell, D. Kennett, in Conflict in Organizational Teams: New Directions in Theory and Practice, L. Thompson, K. Behfar, Eds. (Northwestern Univ. Press, Evanston, IL, 2007), pp. 113-142.
7.    G. Bornstein, I. Erev, Int. J. Conflict Manag. 5, 271 (1994).
8.    M. Van Vugt, D. De Cremer, D. P. Janssen, Psychol. Sci. 18, 19 (2007).
9.    S. Bowles, J.-K. Choi, A. Hopfensitz, J. Theor. Biol. 223, 135 (2003).
10.    J. Henrich et al., Am. Econ. Rev. 91, 73 (2001).
11.    I thank J. Orbell, K. Henry, and B. H. Rogers for helpful comments, and O. Smirnov and D. Kennett for earlier discussions on altruism and war.



原文: The Sharp End of Altruism
                 Holly Arrow



Which would you prefer: a society of selfish but tolerant freetraders, or a warrior society in which people help one another but are hostile to outsiders? If you value both altruism and tolerance, neither seems ideal. Societies of tolerant altruists, however, are exceedingly rare in the simulation presented by Choi and Bowles on page 636 of this issue (1). Instead, altruism flourishes only in the company of outgroup hostility (parochialism), with war as both the engine of this coevolutionary process and its legacy. For a compatriot, the  altruist who risks his life is a shining knight, whereas the outsider encounters the sharp end of this altruism.

From an evolutionary perspective, altruism--acts that benefit others at a personal cost--is puzzling. Some influential theories that address this puzzle are kin altruism (2), the tendency to help blood relations; and reciprocal altruism (3), the tendency to help people who are likely to return the favor. Neither explains generosity to non-kin when costs are high and reciprocation unlikely. Heroism in warfare is an example. Explaining such extravagant altruism via indirect benefits to altruists and their kin has proved difficult. A growing body of work seeks instead to explain altruism with models that include selection on both individuals and groups.

In such "multilevel" models (4), the evolutionary outcome depends on the relative impact of competing pushes and pulls at individual and group levels. Individual selection pushes counterproductive behaviors like altruism out of the gene pool. Group selection exerts a contrary pull, favoring groups with many altruists over groups of more selfish folk. In most species, individual selection wins out. For humans living in small groups, however, a strong group selection pull is plausible. Evidence that intergroup violence killed a nontrivial proportion of our ancestors (5) has fueled interest in war as a force for robust group selection. War is a strong candidate because people kill each other based on group membership.

In Choi and Bowles' simulation, 20 small groups of agents interact over thousands of generations. Agents have two genes, each with two alleles. They are either tolerant (T) or (P) and either altruistic (A) or not (N). Offspring inherit their parents'traits, with occasional random mutations. Altruists help fellow group members at a personal cost; non-altruists do not. Tolerant agents have lucrative exchanges with outsiders; parochial agents do not. A high proportion of parochials in groups restricts trading opportunities for all.

Among the four possible combinations of traits, TN is the most profitable. These self-interested traders profit both from contact with outsiders and from the donations made by altruists. The most costly combination is PA. These generous warriors make donations and also risk their lives to protect noncombatants and conquer new territory for the group's offspring. Individual selection favors the T and N alleles over the P and A alleles. Victory in war favors groups with more PA types over those with fewer. The other two trait combinations are PN bullies, who are both hostile and selfish, and TA philanthropists, who both trade and donate to others.

In each generation, groups are randomly paired. What happens next depends on the proportions of tolerant types and warriors in the paired groups. If two highly tolerant groups are paired, tolerant members reap the benefits of trade. If the proportion of tolerant types drops below a strong majority in either group, however, the likelihood of peaceful trade plummets. Instead, the groups have either an unproductive standoff or a war. If both groups have the same numbers of warriors, a standoff results. War becomes increasingly likely the greater the imbalance of power, and wars end in a victory or a draw. Some proportion of warriors are killed regardless of outcome. In a victory, however, many civilians on the losing side are also killed, and offspring from a postwar baby boom among the victors migrate into the conquered territory.

The societies that evolve are stable in two conditions: when either selfish traders (TN) or generous warriors (PA) are the dominant type. A few PN bullies and even fewer TA philanthropists can coexist within trader or warrior regimes. The trading regime is peaceful. Standoffs and wars are more common in the warrior regime, but even infrequent war--10 to 20% of encounters--can maintain high levels of parochial altruism. Similar findings for the impact of intermittent war on the evolution of heroism (6) suggest that war need not be "constant" to act as a powerful selective force.

The convergence of altruism and parochialism in Choi and Bowles' simulation is consistent with links between the two found in behavioral studies. Selfish choices in social dilemma experiments, for example, diminish markedly when the game is embedded in an intergroup context (7). The boost in altruism caused by awareness of an outgroup is also more marked among women than men (8), consistent with war exerting stronger selective pressure on males as warriors. Interestingly, altruism levels for women, although relatively unaffected by intergroup hostility, were still high. It appears that the relative importance of alternative evolutionary pathways to altruism may differ for men and women.

A full accounting of such pathways must include cultural evolution. In other work, Bowles and colleagues show how norms can support altruism by promoting conformity (9). In the current simulation, warrior-rich groups enforce a trading ban. However, this norm is predetermined. An obvious extension would be to allow norms to evolve. Can pro-trade norms outcompete more isolationist parochial norms? Do norms that punish cowards naturally coevolve with war and altruism?

The simulation findings suggest that one legacy of war is an inherent tension between tolerance and altruism. Cross-cultural studies, however, provide grounds for optimism. In one study, people from 15 small-scale societies played a donation game (10). Average generosity correlated with the amount of market exchange and economic cooperation typical in the society. By adding mutable norms to the simulation, the potential viability of societies of tolerant altruists could be further explored.

A better understanding of how our impulses to give, to trade, and to attack outsiders are intertwined should help in the quest to promote pro-social behavior while keeping the sharp end of altruism sheathed.