Philosophical Aspects of Evolution Research Paper

Charles Darwin (1859) provided the broad outlines of the modern theory. There are traits that are inherited. There is some process that produces natural variation in these traits. The traits may affect the ability of the organism to reproduce, and thus the average number of individuals bearing the traits in the next generation. Therefore, those traits that enhance reproductive success increase in frequency in the population, and those that lead to reproductive success below the average decrease in frequency. The three essential factors in Darwin’s account are (a) natural variation, (b) differential reproduction, and (c) inheritance. The questions of the existence, extent, and nature of these factors are empirical scientific questions.

Could these processes, as they naturally occur, account for large changes over time? Darwin argued that they could by calling attention to the large changes that had come about from the selective breeding of plants and animals. Selective breeding is just a particularly severe form of differential reproduction, imposed by the breeder. Darwin emphasized this argument by calling his theory a theory of Natural Selection:

I have called this principle, by which each slight variation, if useful, is preserved, by the term Natural Selection, in order to mark its relation to man’s power of selection. But the expression often used by Mr. Herbert Spencer, of the Survival of the Fittest, is more accurate, and is sometimes equally convenient. We have seen that man by selection can certainly produce great results, and can adapt organic beings to his own uses, through the accumulation of slight but useful variations, given to him by the hand of Nature. But Natural Selection, as we shall hereafter see, is a power incessantly ready for action, and is as immeasurably superior to man’s feeble efforts, as the works of Nature are to those of Art. (Darwin 1859 Origin of the Species, 6th edn., Chap. III)

The status of Spencer’s description has been a matter of controversy. On superficial reflection on the slogan, Survival of the fittest, more than one commentator has asked: ‘If fitness is fitness to survive, is not the theory of survival of the fittest simply a tautology?’ At this point we can give a preliminary examination of this idea in a highly idealized setting. Let us take Darwinian fitness just to mean expected number of offspring. Let us suppose that the population is effectively infinite, so that the expected number of offspring for a type can be taken to equal the average number of offspring. Let us suppose that reproduction is asexual and transmission of a type is perfect. If we had one generation every time unit, these assumptions specify a dynamics of population proportions called the discrete time replicator dynamics. Taking generations close together, or taking overlapping generations, it is possible to idealize to a system of differential equations which represent the continuous time replicator dynamics (Taylor and Jonker 1978, Hofbauer and Sigmund 1988). This is an idealized model of differential reproduction. In this idealized model, we can ask whether it can be proved that the fittest type will go to fixation in the population.

To get any kind of affirmative answer, we must make more assumptions. Suppose that the fitness of a type is fixed. It does not depend on the frequency of types in the population. The environment is considered constant. Suppose that all the types of interest are initially present in the population and that one of them has the greatest Darwinian fitness. Then it is possible to prove that the dynamics will converge to a state of fixation of the fittest strategy, a state where its proportion of the population is 100 percent. Convergence means just that the population gets arbitrarily close to fixation in an arbitrarily long time. In this sense, survival of the fittest can be mathematically demonstrated. In this sense it is a tautology.

But the demonstration depends not only on the conception of Darwinian fitness, but also essentially on a battery of idealizations and assumptions. If the Darwinian fitness of a type is not fixed, but dependent on the frequency of other types in the population, then—even with all the other idealizations in place— the dynamics may not converge to any state at all. (It may cycle or even exhibit chaotic behavior.) If there is a fitter type not present in the population, then differential reproduction provides no way for it to get in. The model can be modified in various ways to allow for mutation, which may put the type in question into the population, but then the theorem is no longer true. If the population is finite, then only probabilistic conclusions can be reached. These depend on factors other than simply Darwinian fitness, such as the distribution number of progeny for a type and the size of the population. Moreover, in a finite population mutation cannot keep every possible type in play, but rather must be modeled as a chance process that introduces some new types from a very large space of possible types. With the slightest move towards realism the tautology goes away.

In the first half of the twentieth century, Darwinian theory was enriched by Mendelian genetics. Gregor Mendel, an Augustinian monk, had discovered basic principles of genetics by a series of experiments in plant breeding carried out in the garden of the monastery (Mendel 1866). Mendel first bred pure strains of peas that differed in seven selected traits, for instance, green vs. yellow seeds. He crossed these, self-fertilized the hybrids, and studied the frequency of the traits in the hybrids. He found a ratio of 3:1, from which he concluded that each individual carries two genetic units for the trait, one from each parent, chosen at random from the parent’s two genetic units. (This is haploid reproduction.) Thus from hybrid parents, carrying color genes <gy> we would expect to get homozygotes <gg> and <yy> one-quarter of the time each and heterozygotes <gy> one-half of the time. This 1:2:1 ratio then explains the observed 1:3 ratio on the hypothesis that one of the possibilities is dominant, so that this possibility is expressed by the heterozygotes as well as by the appropriate homozygote. Mendel opened the door to genetic models of inheritance.

Mendelian genetics introduces a new source of variation in such sexually reproducing haploid species. Recombination of genes creates rapid variation. But the variation created has its limits. Can recombination of fixed genetic units account for the origin of new species and the whole sweep of Darwinian evolution? Darwinians thought not, and at the beginning of the twentieth century Darwinism and Mendelism were enemies. Only when mutation is included in the picture, so that more fundamental sorts of variation are possible on a longer timescale, is it evident that Mendelism and Darwinism can be thought of as reinforcing each other. This ‘modern synthesis’ of evolution and genetics was created in the first half of the twentieth century (Fisher 1930, Haldane 1932, Dobzhanski 1937, Wright 1968, 1969, 1977, 1978).

Mendelian genetics also enriches the account of differential reproduction. We now need to consider reproduction at more than one level. Individuals reproduce but so do genes. Fitness on one level is connected to that on the other, but in a way that can be complicated. The most obvious case is one in which expression of genes works in such a way that an individual who is a heterozygote is then fitter than either of the homozygotes. A famous example is that of sickle cell anemia. The two possibilities of the gene are sickle, s, or not, n. Heterozygotes, <s,n> are protected against malaria; <n,n> are not, and <s,s> die from sickle cell anemia. But heterozygotes cannot take over the population, because of the mechanics of genetic reproduction. Here, frequency-dependent selection reemerges at the level of the gene.

Mendelian genetics is a mathematical model (or a set of mathematical models). In 1953 Watson and Crick opened the door to understanding the physical basis of genetics at the molecular level. The dynamics of genetic replication, recombination, mutation, and expression can and are being studied at this level. Already the study of molecular genetics has provided many surprises such as the action of retroviruses that can transcribe their own information into the DNA of the host cell, the fact that bacteria can exchange genetic information, and the existence of plasmids— rings of DNA in bacteria that exist and replicate independently of the bacterial chromosome. With the theory of largely adaptively neutral mutations (Kimura 1968, 1983), molecular genetics has also provided the basis of a molecular evolutionary clock which can be used to reconstruct evolutionary history.

The history of the theory of biological evolution has been—and will continue to be—a story of the gradual amplification of Darwin’s insights by empirical research. The three essential factors in Darwin’s account—natural variation, differential reproduction, and inheritance—remain at the center of the theory, but in a form that Darwin could not have imagined.

By social evolution, we mean not the evolution of sociality, which may be one aspect of biological evolution, but rather a social process in some way analogous to the process of biological evolution. But what is the analogy to fix upon? Different thinkers have had different aspects of biological evolution in mind (and sometimes, different conceptions of the nature of the biological process). In its most minimal sense, social and cultural evolution can just be thought of as social and cultural change. So interpreted, the theory of social evolution would be such a truism as to lack all interest. Are there deeper and more informative analogies that are worth exploring?

Some thinkers have thought of ideas that can spread through, or disappear from, a population as analogous to genes (Campbell 1965, Cavalli-Sforza and Feldman 1981, Lumsden and Wilson 1981, Boyd and Richerson 1985). These memes (Dawkins 1976) can recombine in various ways. (A comprehensive discussion of this topic would have to range from nineteenth century associationalist psychology to modern genetic programming.) They can replicate themselves as they spread from mind to mind. They can be thought of as having a sort of Darwinian fitness in terms of the replicator process.

Others have focused on adaptive dynamics as the appropriate analogy. Darwinian evolutionary models fall, more or less, within a broad class of adaptive models which also includes various kinds of learning models. In psychological theories of reinforcement learning, if it assumed that there is some ‘reward’ that is reinforcing, and the theory models the dynamics of behavior under various learning schedules. (The determination of what sort of thing is reinforcing is empirical, but extrinsic to the theory.)

We might then model various dynamical processes of social learning or adaptation instead of trying to model social evolution so as to have a close resemblance to biological evolution.

One common way for individuals to adapt in a social setting is to imitate others whom they observe to be successful according to some generally shared criterion of success. It is a remarkable fact that the replicator dynamics, introduced by Taylor and Jonker (1978) as a mathematical model of differential reproduction, has been derived from a number of different models of learning by imitation in a large population (Binmore et al. 1995, Bjornerstedt and Weibull 1996, Schlag 1998). The replicator dynamics also has deep connections with classical psychological models of positive reinforcement learning (Borgers and Sarin 1997).

The replicator dynamics is a member of the broader class of adaptive dynamics according to which all behaviors with greater than average success increase their population proportion, which is a subset of the still broader class according to which the most successful behavior increases its population proportion. These broader classes of adaptive dynamics may also come into play in theories of social evolution (Weibull 1995, Samuelson 1997).

Populations may not be large, and encounters may not be random but rather structured. Learning by imitation in such settings may lead to different results than it would lead to in large random mixing populations, just as biological evolution may be affected by population structure and population size (Hamilton 1964, Axelrod and Hamilton 1981, Frank 1998, Sober and Wilson 1998, Epstein 1998, Lindgren and Nordahl 1994, Nowak and May 1993, Grim 1996, Pollack 1989, Anderlini and Ianni 1997). It is of particular interest that population structure can favor evolution or learning of cooperative behavior in interactions where it would not survive in large random mixing populations.

Classical reinforcement learning is the kind of learning that demands the least thought. A more sophisticated learner might use some kind of inductive reasoning to form beliefs about the aspects of the environment that, together with her actions, influence payoffs and then act optimally according to these beliefs. If the relevant aspects of the environment include the actions to be taken by other agents, then the reasoning involved may be complicated, depending on the strategic sophistication of the agents and the character of the payoffs. (I am trying to predict what she will do. Is she trying to predict what I will do? If so what will she predict and what will she do and what is my best response? But might she be imagining that I am thinking this way? And so forth.)

Here there is a whole spectrum of learning models, whose appropriateness depends on the inductive and strategic sophistication of the agents being modeled (Fudenberg and Levine 1998).

If small probabilistic perturbations are added to deterministic dynamics of learning or evolution, then a new kind of equilibrium concept is appropriate, that of a stochastically stable equilibrium. In some cases, nature of the underlying deterministic dynamics does not matter, and the stochastically stable equilibrium is determined by the payoff structure (Foster and Young 1990, Kandori et al. 1993). A small population with local interaction may reach stochastically stable equilibrium much faster than a large population with random interactions (Ellison 1993).

The stage of development of the theory of social evolution is comparable to the earlier stages of the theory of biological evolution. There is no analog of molecular biology in view for social evolution. Rather, there are various mathematical models of learning that can be put to empirical test in special domains of application (Suppes and Atkinson 1960, CavalliSforza et al. 1982, Hewlett and Cavalli-Sforza 1986, Roth and Erev 1995, van Huyck et al. 1995, Friedman 1996, Camerer and Ho 1998, 1999).

The processes of biological evolution and social evolution can interact (Cavalli-Sforza and Feldman 1981, Boyd and Richerson 1985, Crawford 1989). The biological evolution of humans as a social species sets the stage upon which social evolution is carried out. Social structure can feed back to biological evolution by imposing or modifying selective pressures. Basic cultural norms, for instance, the prevailing system of agriculture in a society, can be influenced by the biological evolution of other species. As both Darwin and Mendel observed, they can impose selective pressures on domesticated plants and animals. When cultural changes allow humans to survive in new or modified environments, selective pressures on humans themselves can be modified, as emphasized by Baldwin (1902). As humans continue to transform the planet, the biological effects of human cultural evolution become more and more extensive.

Bibliography:

1. Anderlini L, Ianni A 1997 Learning on a torus. In: Bicchieri C, Jeffrey R, Skyrms B (eds.) The Dynamics of Norms. Cambridge University Press, New York
2. Axelrod R, Hamilton W D 1981 The evolution of cooperation. Science 211: 1390–6
3. Baldwin J M 1902 Development and Evolution: Including Psychophysical Evolution, Evolution by Orthoplasty, and the Theory of Genetic Modes. MacMillan, New York
4. Binmore K J, Gale J, Samuelson L 1995 Learning to be imperfect: the ultimatum game. Games and Economic Behavior 8: 56–90
5. Bjornerstedt J, Weibull J 1996 Nash equilibrium and evolution by imitation. In: Arrow K, Columbotto E, Parlman M, Schmidt C (eds.) The Rational Foundations of Economic Behavior. St. Martin’s Press, New York, pp. 155–71
6. Borgers T, Sarin R 1997 learning through reinforcement and the replicator dynamics. Journal of Economic Theory 77: 1–14
7. Blume A, DeJong D V, Kim Y G, Sprinkle G 1998 Experimental evidence on the evolution of meaning in messages in sender– receiver games. American Economic Review 88: 1323–40
8. Boyd R, Richerson P J 1985 Culture and the Evolutionary Process. University of Chicago Press, Chicago
9. Camerer C, Ho T 1998 Experience weighted attraction learning in coordination games: probability rules, heterogeneity and time-variation. Journal of Mathematical Psychology 42: 305–26
10. Camerer C, Ho T 1999 Experience-weighted attraction learning in normal form games. Econometrica 67: 827–74
11. Campbell D 1965 Variation and selective retention in sociocultural evolution. In: Barringer H R, Blanksten G, Mack R M (eds.) Social Change in Developing Areas: A
12. Reinterpritation of E olutionary Theories. Schenkman, Cambridge, MA Cavalli-Sforza L L, Feldman M W 1981 Cultural Transmission and Evolution: A Quantitative Approach. Princeton University Press, Princeton, NJ
13. Cavalli-Sforza L L, Feldman M W, Chen K H, Dornbush S 1982 Theory and observation in cultural transmission. Science 218: 19–27
14. Crawford V P 1989 Learning and mixed strategy equilibria in evolutionary games. Journal of Theoretical Biology 140: 537–50
15. Darwin C 1859 On the Origin of the Species by Means of Natural Selection. J Murray, London
16. Darwin C 1874 The Descent of Man and Selection in relation to Sex. John Murray, London
17. Dawkins R 1976 The Selfish Gene. Oxford University Press, Oxford, UK
18. Dobzhanski T 1937 Genetics and the Origin of Species. Columbia University Press, New York
19. Ellison G 1993 Learning, local interaction and coordination. Econometrica 61: 1047–71
20. Epstein J 1998 Zones of cooperation in demographic prisoner’s dilemma. Complexity 4: 36–48
21. Fisher R A 1930 The Genetical Theory of Natural Selection. Clarendon Press, Oxford, UK
22. Foster D, Young P 1990 Stochastic evolutionary game dynamics. Theoretical Population Biology 38: 219–32
23. Frank S A 1998 Foundations of Social Evolution. Princeton University Press, Princeton, NJ
24. Friedman D 1996 Equilibrium in evolutionary games: some experimental results. Economic Journal 106: 1–25
25. Fudenberg D, Levine D K 1998 The Theory of Learning in Games. MIT Press, Cambridge, MA