Richard D. Alexander (2005)
"Ronald A. Fisher (1930) used the adjective runaway to apply to a hypothetical selective process that he assigned to sexual selection (see also Trivers 1972; Miller 2000). I suggest that sexual selection and reciprocity selection are special forms of the broader concept of social selection. Social selection is a consequence of competition among individuals for rewards arising out of various kinds of social or mutualistic beneficence.
Fisher’s runaway sexual selection has two special features. First is the tendency of the choosing parties to begin favoring individuals with traits that are extreme on some axis of desirability, rather than favoring some particular condition (other than extremeness) of the individuals to be chosen. This form of choosing can only occur in an organism that is able to compare an array of individuals and identify desirable extremes (e.g., Alexander, Marshall, and Cooley 1997). It can evolve to become the standard method of choice only if selection continues for a long time in one direction and the best choices continue to be beyond the previous expressions of the trait. Thus, if females that choose extreme males outreproduce those that do not, the choosing of extremes will spread. So will extremeness in males; otherwise the females choosing them would not gain. Spreading or fixing the choosing of extremes will inject a certain degree of inertia into the process. For this reason, once an “extreme-choosing” tendency is in place, extremes in the traits of the chosen individuals can pass beyond the form in which they are adaptive in any other sense and indeed can become disadvantageous in other contexts. Some such conflict actually exists among all the compromises of selection on different traits of the organism because the effect of selection can only be enhancement of the reproductive integrity of the organism as a whole rather than the state of any of its individual traits. When selection is social, however—when it is a matter of individuals choosing other individuals in a mutualistic or reciprocal social interaction rather than, say, competition to detect or capture food, or to escape enemies more effectively—overshoots in adaptiveness in other respects, because of choosing extremeness, will be more prominent.
The second special feature of Fisher’s runaway sexual selection is the feedback resulting from the genetic partnership between males and females in jointly produced offspring. Trivers (1972, 166) described it as follows:
>[I]f there is a tendency for females to sample the male distribution and to prefer one extreme (for example, the more brightly colored males), then selection will move the male distribution toward the favoured extreme. After a one generation lag, the distribution of female preferences will also move toward a greater percentage of females with extreme desires, because the granddaughters of females preferring the favoured extreme will be more numerous than the granddaughters of females favoring other male attributes. Until countervailing selection intervenes, this female preference will, as first pointed out by Fisher (1958), move both male attributes and female preferences with increasing rapidity in the same direction.<
We can note, first, that some aspects of human evolution that intrigue us, such as brain functions that result in cleverness in social interactions, including scenario building and testing the social future by weighing alternatives internally, could have evolved partly through sexual selection. In view of the tendency of the human brain to become larger across history, and of human behavior to become more complex—and seemingly ever more rapidly after these features had exceeded their counterparts in other species—we might be concerned to examine the likelihood that runaway sexual selection is involved. A related question is whether some or all features of runaway selection can also occur in nonsexual social interactions such as the high-risk forms of social reciprocity that are prominent only in the human species.
At first one may imagine that there are no parallels in social selection allowing it to become runaway. But the way an individual gains by selecting its social partners parallels the ways an individual can gain from cooperative interactions with a mate and from a mate’s parental care. In both cases there is likely to be genetic change in both the ability to choose good partners and the background of the favored phenotypic attributes because a mutually beneficial interaction can be maintained only if, on average, both interactants gain.
Sexual selection is a distinctive kind of runaway selection because joint production of offspring by the interacting pair causes the process to accelerate (see above Trivers quote). The defining feature of runaway selection is not acceleration, however, but the tendency of the process to go significantly beyond adaptiveness in all contexts except within the particular selective race—much further beyond adaptiveness in other contexts than is ordinarily the case in the myriad compromises among the conflicting adaptive traits that create and maintain the unified organism.
This aspect of runaway selection may hold for reciprocity selection, in which, unlike in sexual selection, both parties can carry tendencies not only to choose extremes but also to display extremes. In social selection, again unlike in sexual selection (except in simultaneous hermaphrodites), an individual can play both roles, of chooser and chosen, with respect to the same traits, and alternations of roles can occur during extended interactions between particular partners. To the extent that social success in ecologically dominant humans (see earlier) becomes the central determinant of reproductive success, runaway reciprocity selection may be a more viable possibility than Fisherian runaway sexual selection.
Fisherian runaway sexual selection presumably begins with a likelihood of heritability (i.e., genetic variability) in both variations in choices and variations in chosen traits. It is easy to see that in this circumstance such competitions, or races, will lead to genetic changes, at first changing both the ability and tendency to choose extremes and the nature of the extremes available for choice. Continuing mutations and genetic recombination (outbreeding among temporarily isolated groups) will tend to offer the choosing parties increasingly extreme possible choices, though to a reduced degree as selection continues to remove heritable variations in trait expressions and the trait becomes sufficiently maladaptive in contexts other than sexual selection. From earlier arguments it is not easy to understand the extent to which relevant heritability is likely to disappear, or even become trivial.
Diminution (or disappearance) of heritability in variations will not necessarily change the tendency to choose extremes: in effect, if an ability and tendency to choose extremes in any of a variety of social races (at least in humans) could become genetically fixed in the population (including the ability and tendency to learn from others, or from observation, the values of such choosing), it could still offer advantages to the chooser of extremes (without evolutionary change in the trait per se) because of the usefulness of even nonheritable trait variations chosen in a social partner. Of course, there may be further evolutionary improvements in the ability to identify and use extreme traits even after all choosers are already choosing extremes.
On the other hand, heritable variations in what is chosen will result in a continued march toward greater extremes because this relative quality will not be fixed in the way the ability to identify and favor extremes can be fixed; extremes can be identified only by comparing whatever is available. In Fisher’s version of runaway selection, extremes win reproductively at first because they are ecologically superior, meaning functionally superior outside the choice situation itself. Later they continue to win because they are sexually (or, here, socially) superior, even though, as a result of the progress of selection involving choice of extremes, they may have be come so extreme as to be otherwise functionally (ecologically) inferior. Here, “sexually or socially superior” means that the choosing parties will have acquired a genetic composition that will cause them to choose the extreme, the resulting choice of the extreme individuals itself causing the chosen individuals to outreproduce.
Extremes, however, will be chosen whether or not, as extremes, they represent heritable variants. If extreme social performances yield special social opportunities to those displaying them (e.g., via their reputations as achievers or winners), then regardless of the basis for the superior performances (i.e., genetic variant or not), winning performances can yield benefits to the kin or other social associates of the individual with the extreme traits. Thus merely joining social competitions or races can pay, though only heritable variations, including the (variant) ability to choose the best from among multiple available races that might be entered, will yield evolutionary change. Heritable variations in the ability to choose appropriate races—including not only those generally likely to be profitable but those in which the choosing individual, because of his or her special traits, has a special likelihood of competing successfully—may be all that is needed to drive runaway reciprocity selection. The more different ways that success can be achieved through reciprocity competition, the more robust will be this type of social selection. This is a kind of selection that will yield at least part of the human type of social intelligence, perceptiveness, and perseverance. As we all know, status (or “reputation”) can exist independent of the adoption of a particular extreme in behavior, so the importance of any behavioral extreme in changing some aspect of culture can also depend on whether a prestigious person adopts, favors, or approves of it.
Since Fisherian runaway sexual selection in nonhuman organisms has remained controversial (or theoretical), but social parallels to it may be robust in human society, one may wonder if Fisher actually derived his idea from observing human social situations rather than from thinking about sexual selection in the birds he ultimately used as his examples. If so, it is likely not the only instance in which an evolutionist was inspired by human traits and tendencies to develop a general evolutionary explanation (e.g., Darwin’s observations on human selection of variations in domestic animals and the fact that Hamilton’s rule applies in its fullest extent as extensive differential nepotism across multiple levels of relatedness only in humans). This suggestion is ironic in another way, in view of the success with which academic biology departments have managed to exclude the human species from their consideration, leaving its analysis to the almost exclusively nonevolutionary approaches of social science and medical departments, and surely delaying acceptable explanations of the human species in evolutionary terms.
When extremes in particular directions are being chosen in social selection, new extremes never before experienced may be chosen above all other expressions of a trait. This behavior was originally referred to in European ethology as a “superoptimal stimulus” effect and is not specifically related to runaway selection. Whenever superoptimal stimulus effects can be identified, however, they suggest a history of choosing extremes, therefore the possibility of some form of runaway selection. If, for example, we perceive that during the past several centuries art— or even human sociality in general—has flourished and become an enormous and diverse enterprise compared to any “ancestral” condition it might have exhibited, we might suspect that this is evidence of a history of choosing extremes continually in short supply and consequently of some kind of runaway process in however artistry affects the securing of the resources of reproduction. Rather than evidence that evolution cannot be used to explain, say, the arts, explanations of recent flowerings of diversity and complexity in human enterprises can be sought by expanding our understanding of the various subprocesses of evolutionary selection. We should not be discouraged because the pathways to explanation become progressively more difficult and call for expansions of our understanding of the workings of the evolutionary process."