EvoWiki is now a project of the
We are moving all content to RationalWiki.
See the for details! .
Altruism may reasonably be defined as acting to benefit others at one's expense. This sometimes takes spectacular forms, like the altruism of cells in a multicellular organism. Nearly all of them will die when the organism dies -- or before. A blue whale contains 10^17 cells, and a giant sequoia tree contains 10^19 cells, but on average, only two of them survive. And many body parts consist of dead cells, like
- Wood (yes, a tree is mostly dead, and its living part is a thin layer, the cambium)
- Land-vertebrate epidermis (outer skin layer)
- Vertebrate hair, nails, claws, feathers, etc.
Some body parts are alive, but nevertheless die before the organism, like some flower parts and the leaves of deciduous trees. Also, growth and development often involve the dying of cells that are out-of-place, like the cells between our (embryonic) fingers (see: apoptosis).
Altruism is not quite as spectacular at the organism level, but it nevertheless happens in abundance. Parental care is an obvious example, and it sometimes involves the parent dying, as in plants that "go to seed" and octopuses. Looking further, social animals are often altruistic to each other, and social-insect colonies usually include individuals that do not reproduce -- often the great majority of individuals.
The occurrence of altruism is a seeming paradox for Charles Darwin's mechanism of natural selection, because its natural consequence is a sort of universal selfishness. It is for good reason that Richard Dawkins refers to the "selfish gene". However, there are three ways in which altruism can appear:
The theory of reciprocal altruism holds that psychological mechanisms for providing benefits to non-relatives can evolve insofar as the benefit is reciprocated at some point in the future. This is essentially "I'll scratch your back if you scratch mine", and is known from several species of social animals.
The problems with cooperation between unrelated individuals are similar to the Prisoner's Dilemma (PD) in game theory, originally formulated by John Kagel and Alvin Roth in response to John Nash's equilibrium. As the story goes, two men, Fred and Ed, have committed a serious crime. However, the District Attorney is unable to make a conviction because an essential piece of evidence is deemed inadmissible. Yet she continues to hold them, booking them on lesser charges and keeping them apart, so as they cannot possibly communicate. Finally, upset, she goes to each one separately and proposes the same deal: "Come clean with me; confess, and if your partner refuses, I will convince the jury he is the brains behind the operation and you a reformed man, giving him a long ten year sentence and you just one. If you refuse, and he agrees, the opposite is going to be true; he will get off with a slap on the wrist, and you will be sent down the river. If neither of you confess, you will be charged on this lesser crime, but are assured two years, but if you both do, I will be forced to let justice take it course, and sentence you both to five."
Knowing that both parties are devoted to the best outcomes, we can write the following matrix:
Ed confesses, Fred doesnt -- 1 year, 10 years
Ed confesses, Fred confesses -- 5 years each
Fred confesses, Ed doesnt -- 1 year, 10 years
Fred doesnt, Ed doesnt -- 2 year each
Each will then reason: "If he does not confess, then I will be out in 1 year if I confess and 2 years if I stay silent. However, if he does confess, then I will be in prison for 5 years if I also confess and 10 years if I stay silent. Either way, it makes perfect sense for me to confess." The District Attorney, pleased by the results, collects the confessions and promptly sends them away for five years; three years longer than had they simply both kept their mouths shut. In sum, defecting leaves them worse off than if they had cooperated.
The parallels with reciprocal altruism are obvious. Both parties to the interaction could gain from cooperation, but both are tempted to take advantage of the other's generocity without following up with reciprocation. In a one-shot PD, as above, it also makes sense for both to defect. But the real world is not a one-shot PD. In an iterated prisoner's dilemma (IPD), when the game is repeated a number of times with each actor not knowing when it will end, certain long term strategies come into play. After testing hundreds of different versions, solicited from biologists, psychologists, economists and many others around the globe, Robert Axelrod and William D. Hamilton had a single winner, tit for tat.
The winning strategy, tit-for-tat, is amazingly simple. Unlike some of the other strategies solicitied and received by Axelrod and Hamilton which often involved hundreds of lines of FORTRAN (computing language) code, tit-for-tat had four, spelling out two simple rules: 1) cooperate on the first move, and 2) continue cooperating if the other actor if cooperating, defect if the other defects. This is essentially a way to punish those who would take advantage of cooperaters, yet retain the benifits for honest parties. It says, essentially, "if you dont scratch my back, I won't scratch yours."
Examples of this sort of behavior abound. Vampire bats will starve if they do not eat for more than two or three days, but a bat that has eaten will vomit up its meal for one that has not -- provided that the hungry bat had been willing to do the same when it was well-fed. Attempts to detect cheating -- and to cheat -- are speculated by some to be a major driving force in the evolution of intelligence.
This was first speculated on by Charles Darwin, and worked out mathematically by William D. Hamilton. In this mechanism, altruism perpetuates itself because it helps owners of copies of the altruist's genes, thus helping to increase inclusive fitness. Parental care is an obvious example, because one's offspring will transmit one's genes further in time. Body cells share genes with the reproducing cells (gametes, spore cells), and nonreproducing "worker" insects help raise their reproducer siblings, which they share genes with. Social groups tend to be somewhat inbred, so altruism toward other members of a group can be favored by kin selection.
It is possible that in early human societies women chose men to mate with and men chose women to mate with who had altruistic personalities. This would be because the offspring would be more likely to survive if the parents were willing to sacrifice themselves for the survival of the children. Therefore, the genes that promote altruism are more likely to be passed on because the humans with those genes are more likely to find mates.
Hamilton had worked out mathematically that a diploid individual shares 1/2 its genes with its siblings and 1/8 with its first cousins. This pattern translates into "Hamilton's Rule": c > rb, where c is the cost to the actor, r is the degree of relatedness, and b is the benefit to the recipient. Insofar as, for example, the benifits to a .5 kin are more than twice the cost to the actor, selection will favor the altruistic behavior. If the benifit to a .25 kin is more than four times the cost, or the benifit to a .125 kin is eight times the cost, the action will similarly be selected for.
It is not surprising that the kin people are most eager to help are siblings, and that this tendency decreases linearly with relatedness (Burnstein et al, 1994). Also illustrative are patterns of inheritance. Smith et al (1987) looked at the bequests of 1000 randomly chosen Canadian individuals (552 men and 448 women), working under the kin selection assumptions that 1) people will leave more of their estates to genetic kin (including spouses, who they assume will distribute resources to children), 2) they will leave more of their estates to closer genetic kin than more distant ones, and 3) they will leave more of their estate to offspring than full siblings. Although the kinship correlations of each are the same (.5 in both cases), the rationale is that children will be of a more viable reproductive age. All three predictions were confirmed. In total, 92.3% of estates were given to genetic kin, with 46% going to those sharing 50% of their genes (children, full siblings), 8% to those sharing 25% (nephews, nieces, grandchildren), and less than 1% to those sharing 12.5% or less (cousins and others).
Hymenopterans (ants, bees, wasps) have an interesting wrinkle: they are haplodiploid, with males being haploid and females being diploid. Concomitantly, female hymenopterans are more closely related to their sisters than to their offspring (2/3 for having the same father vs. 1/2). This circumstance predisposes worker females to care for their queens' offspring, and may explain why nonreproducing workers have emerged several times in Hymenoptera and much less often elsewhere in the animal kingdom (termites, naked mole rats).
A third way altruism might evolve is via friendship. Tooby and Cosmides (1996) note that friendships are often most valued when altruistic acts are done without promise of reciprocation, and that perhaps our intuitions on this matter (that friendship is not just another sort of reciprocal altruism) may be correct.