«Chapter 14: STYLE AND ETHNICITY: THE EVOLUTION OF SYMBOLIC TRAITS “. what are the advantages which we propose by that great purpose of human life ...»
Indirect bias is an important force on symbolic traits and has very interesting properties. Because symbolic characters are used in communication, they are a natural locus for evolutionary forces based on choice of whom to imitate or mate with. People often seem to act like they want to pass on their genes and culture to others. The epigraph from the archrationalist utilitarian Adam Smith reflects his acknowledgment that people's fundamental desires are to be imitated and mated. We can easily imagine that people choose mates and cultural teachers with an idea to acquiring good genes and good culture to pass on to genetic and cultural offspring. Ideally, we would be able to survey potential mates and “role models” (let's use “role model” as a shorthand for any type of cultural model we might have the chance to choose rather than be stuck with) and estimate as precisely as possible what their genetic and/or cultural quality is. However, this is not easy to do. On the other hand, in order to attract mates and imitators, individuals are perhaps willing to advertise.
The simple step of trying to choose mates and role models based on limited information that includes advertising leads quickly to complex but fascinating evolutionary dynamics. As a mate or model chooser, honest communication will greatly help our choice. Those who seek mates or “proteges” may be quite willing to give signs of their cultural or genetic quality. These signs may or may not be honest. That is, mate or protege seekers with less than ideal genes or culture may well benefit from false advertising.
It matters what other people think. Once a system of communicating mate or model quality becomes widespread, it matters not only what you or I as potential mate or role 14-248 Evolution of Symbolic Traits model choosers desire for our own advantage, but also what others are likely to choose. For example, a female choosing a mate may think that a colorful tail (or the habit of smoking) is a stupid display, poorly correlated with genetic fitness (or cultural achievement). However, if she is convinced that most females in the population do think that colorful tails or smoking are associated with success, she may choose a colorful male or a smoker in order that her sons, inheriting colorful tails or smoking from their fathers, have a better chance of being attractive to their potential mates. This was Darwin's basic intuition about mate choice. Once an “aesthetic capacity” developed in (usually) females, successful appeals to female taste could counterbalance quite considerable ordinary fitness costs. (Cronin, 1992, gives a good analysis and history of the sexual selection issue, going back to Darwin and Wallace.) Indirect bias has a lot of similarities to mate choice sexual selection and, general conclusions that apply to one are also likely to apply to the other. Recall, cultural transmission is indirectly biased when people use some traits displayed by potential role models, such as indicators of prestige, such as dialect, to bias the imitation of other traits, such as subsistence technique. In other words, some attributes of a model are used by naive imitators as a basis for choosing to imitate a more general class of traits for the same model. It is useful to distinguish three classes of characters when thinking about this mechanism of cultural evolution: (1) Indicator traits are displayed by models and used as a basis for weighting their importance by imitators. For example, suppose that imitators are inclined to admire and then to imitate successful individuals, and that success is estimated using particular indicator traits--number of cows, number of children, or style of dress. (2) Indirectly biased traits are acquired as a by-product of choices based on indicator traits. Once a particular model is given a large weight in enculturation, naive individuals might tend to acquire animal husbandry lore, beliefs about appropriate family size, or a set of study habits from this role model. (3) Preference traits are the criteria by which naive individuals evaluate indicator traits of potential models. In some cases, a simple more- is-better rule (e.g., the wealthier, or the better student, the better) might be used. Other times, intermediate values of an indicator trait may result in the strongest weight. For example, contemporary middle-class Americans seem to most admire people with intermediate-sized families, not the childless or those with very large families. Later on, we'll look at a case in which people prefer as models someone who is simply like themselves on a stylistic trait--birds of a feather flock together.
Language evolution shows indirect bias in action. Sociolinguists have shown that patterns of the spread of linguistic variants are under the influence of an indirect bias mechEvolution of Symbolic Traits 14-249 anism (Labov, 1980). One of Labov's most famous examples is the development of a distinctive phonology (way of pronunciation) on Martha's Vineyard, an island off the coast of Massachusetts. People on the island are mostly involved in the tourist business, but they consider catering to the whims of tourists to be rather low status. Most Islanders admire the independence of the few fishermen left working on Martha's Vineyard. They see them as exemplifying the individualism and independence of Old Yankees. Preferring independence, and treating fishermen's salty talk (including frequent disparaging remarks about tourists) as an indicator of an independent frame of mind, Islanders put fishermen at the top of the local prestige hierarchy. Have done so, they tend to imitate the pronunciation of fishermen, who have thus become the leaders of dialect evolution on Martha's Vineyard. Interestingly, those young individuals whose ambitions will lead them to leave the island are responding to the wider New England prestige system, and conform to mainland rather than Island phonological patterns. The pronunciation traits changing on Martha's Vineyard seem to be indirectly biased (rather than being important indicators) because data suggest that people are not very self-aware concerning these particular small-sale dialect differences.
There is considerable evidence that people use indirect bias in a number of situations
similar to the linguistic example to choose whom to imitate:
(1) Evidence from social learning theory. Laboratory studies of human imitation have shown that naive individuals often use indicator attributes of role models to bias their attention to models and their acquisition of other behaviors (Rosenthal and Zimmerman, 1978:251-254). For example, Yussen and Levy (1975) exposed preschool children and third graders to warm and neutral adult models. Warm models increased student's attention, reduced their susceptibility to distraction, and enhanced their recall of modeled behaviors.
(2) Evidence from the diffusion of innovations. Rogers with Shoemaker (1971:ch 6) review how patterns of information flow during the adoption of innovations are affected by sociological attributes of adopters and models. They discovered that a class of individuals whom they label “opinion leaders” play a disproportionate role in the spread of innovations within a local community. These individuals are usually higher in status than average in the local community, and seem to be picked as role models based on local prestige indicators.
If opinion leaders adopt an innovation, it spreads to the rest of the community; if not, few adopt the innovation (more on this topic in Chapter 20).
It is easy to imagine that indirect bias is a functional mechanism for acquiring cultural traits (Flinn and Alexander, 1982). By imitating successful role models, naive individuals can increase the chance that they will acquire the beliefs and values that lead to success. In the case of diffusion of innovations, for example, Rogers suggests that copying 14-250 Evolution of Symbolic Traits opinion leaders is a sensible way for potential adopters to decide whether to adopt an innovation. Potential adopters of new techniques have a wide range of abilities and resources to devote to judging the utility of new techniques, and it makes sense for adopters with moderate resources to use opinion leaders with more resources as models of what to adopt. On the other hand, a choice of models of very different status is unlikely to be an effective strategy because the circumstances of life of such a model are likely to be too different to provide a good guide for optimal techniques. Thus, rural people use the best farmers in the village rather than rich city based amateur farmers as opinion leaders for farming techniques. Foreign “experts” are also often viewed skeptically by 3rd World aid recipients. In our experience such experts do sometimes show an unfortunate carelessness about local constraints and conditions.
C. The Runaway Process Indirect bias can also result in maladaptive runaway evolution. As we noted earlier, one of the most interesting questions is what leads to the apparently maladaptive elaboration of stylistic displays. In this section, we briefly describe the structure of a very simple model that we think captures the bare bones of the problem. A detailed description of the model is given in the reading.
We begin by assuming that each individual can be characterized by the values of two cultural traits. The first trait is an indicator trait that affects the individual's attractiveness as a model, and the second is a preference trait which determines which variant of the indicator trait that the individual finds most attractive. To keep the model simple we do not include any indirectly biased traits other than the preference trait itself. We assume that different variants of the indicator trait are characterized by different genetic fitness. The model also lacks any explicit details about the genetic system. We merely assume that selection on the (cultural) indicator trait favors a variant that is optimal in terms of genetic fitness.
The two traits themselves are modeled as quantitative characters. That is, it is assumed that they can be measured as real numbers rather than taking on discrete values. Such a characterization is quite apt for traits like wealth that do vary continuously, but might not be a very good representation of a trait like class that may have only a few discrete variants in some social systems.
We assume that the life cycle of cultural transmission begins with an episode of simple unbiased transmission from parents to children in which children acquire both traits.
This is followed by an episode of indirectly biased transmission in which adolescents may modify one or both traits after choosing a number non-parental adults as role models. The extent to which a particular role model influences a particular adolescent is affected by the Evolution of Symbolic Traits 14-251 preference trait that the adolescent acquired from his or her parents and the adult role model's indicator trait. That is, adolescents use the preferences learned from their parents to select a set of non-parental models based on these models' easily observable characteristics, their indicator traits. For example, children raised in religious households will be more prone to learn from people who are evidently pious than from those that are not. The adolescents then modify their original indicator and preference traits on the basis of the models chosen. This is not an either/or choice, rather the adolescent imitators weight the influence of the models in accord with their preferences and the models' indicator trait values.
Finally, there is the episode of natural selection favoring the value of the indicator trait that maximizes genetic fitness. Then the next generation begins with an episode of parental transmission to the next generation, completing the life cycle.
Based on these assumptions, one can construct a mathematical model for the distribution of indicator and preference traits in the population (details, again, in the reading).
The model suggests that cultural evolution under the influence of indirect bias has two distinct modes:
(1) Stable fitness maximization, honest advertising. If the strength of indirect bias acting on the preference trait is weak compared to the combined adaptive forces of selection and direct bias on the indicator trait, then the preference trait will eventually reach a stable equilibrium at the value that maximizes genetic fitness. In other words, both direct and indirect bias will evolve so that naive individuals tend to imitate models with the optimum value of the indicator trait. This occurs when selection is strong enough to ensure that the indicator trait remains a good index of fitness, and individuals are usually able to imitate the indicator trait they prefer.
(2) A runaway case, costly exaggerated advertising. If the strength of indirect bias acting on the preference character is strong compared to the combined adaptive forces of selection and direct bias acting on the indicator character, then according to the model, the values of both the indicator trait and the preference trait will run away, becoming indefinitely larger or smaller depending on the initial condition. Clearly, this cannot be literally true; nothing can really grow or shrink without bound. Some process not accounted for in the model will eventually restrain the evolution of the population. The correct qualitative lesson to be drawn is that when the evolution of preference trait is affected by indirect bias, the resulting process may be inherently unstable. Where it exists, such instability is likely to result in preference and indicator traits that are some distance from their genetic fitness optimizing values. Figure 14-1 illustrates the behavior of the model graphically. Thus, when the situation of the runaway obtains, a slave-of-fashion effect can indeed arise. An 14-252 Evolution of Symbolic Traits Figure 14-1. Cultural evolution can “runaway” due to indirect bias forces. If the strength of bias for the preference trait is stronger than the combined strength of selection and bias for the indicator trait, the value of the indicator trait in future generations (t+1, etc.) can runaway to increasingly maladaptive values. Here the two extremes might be everyone wears only skimpy, sexy clothing (Madonna’s underwear regardless of the weather) and the other everyone wears completely concealing unisex clothing (regardless of weather).