Evolutionary biologists have long since been at odds with each other regarding the explanation of the emergent property of altruism in nature. One seemingly easy way to deal with the problem of altruism is multilevel selection (MLS) theory. This idea claims that there are multiple forms of selection that act on particles of a collective at all levels of hierarchy from genes to ecosystems (this idea is not proposed to deal with altruism exclusively, but for many traits and adaptations). This idea can also be referred to as group selection, or selection “for the good of the group.” This idea had been widely accepted until kin-selection and inclusive fitness—stressing a gene’s-eye-view of natural selection—along with a new branch of biology called game theory, came to fruition. These new developments claim to fully explain all domains of adaptation, including altruism, in nature without a retreat to MLS, contending that natural selection acts exclusively on genes. An end to this debate seems far off, but the question is clear: Can gene selection explain all aspects of cooperation, or must we resort to a multileveled selection acting on all levels of the biological hierarchy?
The principle of parsimony is crucial to science. It states that when explaining any natural phenomenon one should eliminate all assumptions and choose the explanation of least resistance, the one requiring the fewest logical leaps. However, disparate facts often cannot be easily synthesized by all-encompassing theories, and models must be made for special, complex cases. One such complex case had come to the attention of evolutionary biologists when attempting to explain the evolution of cooperation, or altruism. Altruism, or traits that are harmful to an individual but benefit a common good, are extremely common in nature from microorganisms up through the most complex eukaryotes (Zahavi 2005). Since the days of Darwin, the common explanation had been to attribute such adaptations to inter-group selection, claiming that such-and-such an adaptation had arisen because it was good for the group, and hence good for the individual. Many biologists found a great flaw in this reasoning. They argued that altruistic traits such as food sharing or grooming could not remain stable in a population because such traits would benefit all individuals in the group including defective cheaters. “Genes associated with cheating would therefore spread through the group, and the propensity for cooperative resource management would be undermined” (Wilson & Wilson 2008). These ideas were put forth in the 1960s by George C. William’s publication of Adaptation and Natural Selection, followed by W. D. Hamilton’s development of kin-selection, and John Maynard Smith’s work in game theory (Sober 2008). The model for interactions between individuals that could give rise to an altruistic population was the Prisoner’s Dilemma model, a product of game theory, which will be explained in detail later.
However convincing and widely accepted this new model was, many were not completely convinced of the exclusivity of gene selection resulting from the interaction of individuals. One particular dissenter, Richard Lewontin, argued that one could not ignore the role of particles of a collective with regard to natural selection. The genes that make up an organism, organisms that make up a population, and populations that make up a community, etc., can all be considered as particles of a collective. His main point was that collectives contain particles that vary in their relative fitness (survival and reproduction) to other particles, and that if the fitness of a parent particle is passed on to their offspring, then the overall structure of the collective will change. This selection between collectives would then favor behaviors that are beneficial to the collectives in relation to each other, and hence selecting for traits that are for the good of the group (Sober 2008). There is much evidence that this process takes place in nature with collectives colorfully named superorganisms (ants, bees, and termites to name a few). This is an area of research that is extremely valuable because one cannot proficiently evaluate effects if the causes have not been determined. Whether or not we can decide on the exclusivity of either kin-selection or group selection will require an investigation of the central tenants of each.
The Prisoner’s Dilemma is a product of game theory and is the model for kin-selection within a group. It is based upon a series of interactions, that is, reencounters with individuals within a population where individuals can either cooperate (invest in a mutual good), or defect (exploit the other’s investment) (McNamara et al. 2004). Think of this model as it may relate to a type of bird that is known to groom other birds, combing them for harmful parasites. Cooperating in this example relates to an individual whom exerts valuable time and energy in removing parasites from a member of his population. Defecting means that the individual gladly accepts the removal of his parasites, but fails to reciprocate and remove the other’s parasites in return. In this model, the highest payoff for an individual is to defect while the other cooperates because the individual increases its fitness by getting parasites removed, but pays nothing in terms of time or energy in removing the parasites of the cooperator. It logically follows that the worst payoff would be to cooperate while the other defects, since you’ll waste time and energy removing parasites without getting yours removed. The payoff in this case would be negative for the cooperator. Now, if both defect, the payoff is zero—nobody exerts energy and nobody gets parasites removed. And lastly, if both cooperate they each get a high payoff, though not quite as high as defecting towards a cooperator. It must be noted that in a population that consists predominantly of cooperators, defector traits enjoy a high fitness because the chances of them meeting a cooperator is extremely high, therefore defectors will begin to dominate. This seemed to say that cooperation is not a stable strategy in populations. Scientists then made individuals follow a set of rules, an algorithm that an individual would follow when it met other individuals. Scientists made numerous programs and played them against one another to see which algorithm surfaced with the highest fitness at the end. Of the algorithms presented, some were especially cooperative, some were particularly defective, and many were tricky, throwing in random defects among their cooperations. The algorithm that finished with the highest marks each time was one of the simplest algorithms named TIT-FOR-TAT. This algorithm would start off nice by cooperating, and would then mimic the last move of its competitor (Axelrod 1981). So, while initially nice and never the first to defect, it punishes defection, yet is always willing to forgive and cooperate if the other chooses to do so (McNamara et al. 2004). These results were so striking because TIT-FOR-TAT is nearly a perfect representation of our own interactions in the world (Aoki 1983). It had been shown in this experiment that even though each individual had been looking out for their own interest, a cooperative and forgiving algorithm showed to be the most evolutionarily stable strategy. Parsimony has seemingly been achieved with this view because if all adaptation can be explained by selection acting only on genes, postulating higher levels of selection become superfluous.
Many real world example of TIT-FOR-TAT have been found, one particularly interesting example being with the hermaphroditic fish, sea bass. These individuals form monogamous pairs and switch off playing male and female roles. It is in the best interest of each individual to cheat and play the male role each time, since this role requires the least investment. Evolutionary biologist Eric Fischer made a prediction that sea bass must be playing TIT-FOR-TAT and keeping track of cheaters. Fischer observed that monogamous pairs alternating unevenly were more likely to break up, showing that a TIT-FOR-TAT strategy is in the works.
Harvard entomologist E. O. Wilson denounces the simplest form of group selection, which states that behaviors can evolve for the good of the group at the cost of personal fitness. He does, however, believe that adaptations with public benefits do evolve by natural selection, though the selection is acting between groups rather than between individuals. Units of selection, he claims, “are like Russian matryoshka dolls nested one within another” (Wilson & Wilson 2008). He claims that at each level in the biological hierarchy (doll) natural selection will favor different adaptations. Evolution proceeds by a combination of within-group and between-group tug-of-wars in which individuals face the problem of dividing up their energy between competition within their group and competition between groups (Reeve & Holldobler 2007). There is much evidence that this, indeed, takes place in nature.
Portuguese man-of-wars, and other siphonophores, are particularly interesting to evolutionary biologists because they seem to blur the concept of individual and group alike. Siphonophores are interesting because they can’t be considered a single organism, they are actually colonies made from asexually reproducing individuals that have specialized their form and function to roles such as locomotion, digestion, and even a makeshift nervous system (Gould 82). These individuals came together and now survive in a new type of organism, a superorganism. The individual organisms can be seen as specializing themselves for the good of the group. These siphonophore colonies grow via asexual reproduction, so the members constituting the entire superorganism are genetically identical, which eliminates any in-group selection. The only way these complex groups could have evolved such astounding cooperation of its subunits would be by selection between other such groups (Wilson & Wilson 2008).
Superorganism status is also commonly ascribed to ants and other hymenoptera such as bees and wasps. These organisms do not reproduce asexually and, therefore, group selection alone can’t account for all of their adaptations. These species show tremendous synchronicity and division of labor within the colony. However, it has been observed that dominants in the colony have limited control over the subordinates within the colony. This makes sense, since each member is not a genetic clone of one another. Reeve and Holldobler have devised a tug-of-war model for working out this special case where neither gene selection nor group selection alone can account for the observed phenomena. In a tug-of-war, each member of a group will selfishly diminish the group’s total output in order to increase its own output. They call this a “selfish investment.” Each member’s share is dependant on the significance of its selfish investment in relation to the summed selfish investments of other members in the group. This relationship models a phenomenon called “mutual policing” within a group. But tug-of-wars are also taking place between rivaling groups over resources, and those groups that have optimized their mutual policing will be selected for over groups that have not achieved such levels of cooperation (Reeve & Holldobler 2007). This is a view that offers a synthesis of both intra-group and inter-group selection. These two forms of selection interlock because “a group’s ability to out compete other groups declines the more total energy its group members expend in the within-group tug-of-war” (Reeve & Holldobler 2007).
We have seen that both camps have much to contribute to theories on the evolution of cooperation. Gene selection offers a simplistic look at the evolution of cooperation without resorting to what it views as superfluous logical leaps, which violate the principle of parsimony. If all adaptations can be explained with selection acting only on genes, there is no need to resort to hierarchical selection. However, the MLS theorists call for selection acting simultaneously on every rung of the hierarchy to explain the many emergent features observed in nature. The implications of this area of research range well outside mere explanations of cooperation, but also morphological adaptations that result from MLS theory, such as the evolution of stingers in bees that rip off once used, killing the bee in the process. From a thorough survey of this field it seems evident that both camps are guilty of attempting to bend nature and reality to fit within their own preconceived models of how nature should work, rather than how it actually is. It appears that the true stance to take with regard to levels of selection is to realize that selection pressures lie on a wide spectrum, and that nature is rarely ever split into dichotomy. Nature needs to be evaluated on a case-by-case basis. The leaps of logic occur when either camp calls for exclusivity of their own model, with complete rejection of the other. Parsimony will only truly be reached when exclusivity is abandoned and synthesis obtained. Nature is never so black and white, and rarely so accommodating.
Aoki K. 1983. A quantitative genetic model of reciprocal altruism: A condition for kin or group selection to prevail. Genetics. 80:4065-4068.
Axelrod R, Hamilton WD. 1981. The evolution of cooperation. Science. 211:1390-1396.
Dawkins, R. (2006). The Selfish Gene (30th Anneversary ed.). New York: Oxford University Press.
Gould, S. J. (1987). The Flamingo’s Smile. New York: Norton Press.
McNamara JM, Barta Z, Houston AI. 2004. Variation in behaviour promotes cooperation in the Prisoner’s Dilemma game. Nature. 428:745-748.
Reeve KH, Hölldobler B. 2007. The emergence of a superorganism through intergroup competition. Proceedings of the National Academy of Sciences. 104(23):9736-9740.
Sober E. 2008. A philosopher Looks at the Units of Selection. Bioscience. 58(1):78-80.
Wilson DS, Wilson EO. 2008. Evolution “for the Good of the Group.” American Scientist. 96(5):380-390.
Zahavi A. 2005. Is group selection necessary to explain social adaptations in microorganisms? Heredity. 94:143-144.