Stone Age Minds and Group Selection – What difference do they make? Jack Vromen EIPE [email protected] (first draft; please do not quote) Paper prepared for the Workshop ‘The Nature and Evolution of Institutions’, Jena, 11-13 January 2001 1 1. Introduction Over the last years evolutionary psychology (EP) and Sober and Wilson’s multiselection theory (MST) attracted much attention. These two currents in evolutionary theorising did not only gather adherents. Both currents provoked massive and sometimes heated controversy. But they succeeded in establishing themselves as ‘centres of gravitation’ pulling together much of present-day debate within evolutionary theory. Evolutionary psychology and multiselection theory also caught the eye of practitioners in the social sciences. There also they are received in many different ways. Not only range responses from utterly dismissive all over to wholeheartedly affirmative. There also seem to be much confusion about what evolutionary psychology and multiselection theory are and what they could mean for the social sciences. This paper sets out to do three things. First an attempt is made to understand the basic ideas in both theories. This already is not an easy task. This is so for several reasons. Both EP and MST are relatively recent endeavours. They both are ‘moving targets’ that are still in their infancy. Their basic claims and ideas are often also ambiguous, as we shall see. And in EP there seem to be internal disagreements. The second thing I want to do is to compare the two theories with each other. Significant agreements and disagreements between the two will be spelled in some detail. And finally some first steps are made to find out of what possible relevance EP and MST are for economics. 2. Evolutionary psychology (EP) Evolutionary psychology (EP) often is depicted as sociobiology in disguise (or 1 sociobiology garbed in a new dress). Edward O. Wilson, for example, one of the founders of sociobiology (Wilson 1975), argues as if EP is identical to sociobiology (Wilson 1998, 165 and 185). But this, I think, slides over at least one significant difference between the two. Sociobiology has often been said to endorse genetic determinism. Sociobiology holds, it is argued, that genes code for brain archtecture and brain processes, and that these in turn would directly determine behaviour. EP differs from sociobiology in that much more attention is paid to an intermediate link in between neurophysiological processes and brain architecture on the one hand, and behaviour on the other. Indeed, EP has been introduced by its pioneers as an attempt to provide the missing link in evolutionary theory (Cosmides and Tooby, 1987). EP situates this link at the level of the human mind. To be more precise, EP postulates the existence of a multitude of psychological mechanisms (or modules) working at the level of the human mind (Buss, 1995 and 1999). In order to fully comprehend EP, the crucial distinction between ultimate and proximate causes has to be well understood. Indeed, it is no exaggeration to state that the whole of current evolutionary theory cannot be understood if this distinction is disregarded (see, for example, De Waal, 1996). The distinction goes back to Mayr (1961). Mayr argues that there are several perfectly legitimate answers to the question why for example warblers migrate to the south when winter is about to 1 It has to be noted that EP is far from uncontroversial. There also seem to be different views within EP. And central notions do not seem to be unambiguously defined. For more on this, see Vromen (1999a). 2 come. What is the cause of bird migration? One legitimate answer is that migration is linked with photoperiodicity. Warblers are equipped with some internal physiological machinery that allows them to sense decreases in day length and that allows them to respond by migrating southwards. Another legitimate answer is that during the evolutionary history of the species, warblers have gained a genetic constitution making for such internal physiological machinery. Here the causes are the evolutionary forces that impinged on the ancestors of the warblers living now. Mayr calls the first class of causes proximate causes and the second class ultimate 2 causes. Proximate causes refer to the decision-making machinery (or behaviour-instigating mechanisms) inside the behaving entity. They are internal to the behaving entity and they operate just before the behaviour is displayed. Ultimate causes on the other hand refer to circumstances outside ancestors of the behaving entity responsible for the behavioural profile the behaving entity has now. Thus, while proximate causes refer to causes internal to the behaving entity working now, ultimate causes refer to causes that were external to its ancestors then (that is, to causes working way back in time). The link between the two types of causes is that proximate causes are effects of ultimate causes. Ultimate causes impinging on behavioural entities in the past have produced the proximate causes that govern the behaviour of their descendants now. One of the central (if not the central) ultimate causes in evolutionary theory is natural selection. But what are the paradigmatic species of proximate causes in evolutionary theory? The example of the migrating warblers given above suggests that neurophysiological processes inside organisms are prime proximate causes in evolutionary theory. If we add to this the widespread belief that neurophysiological processes are coded for in the organism’s genes, then the image that emerges is one of a direct, straightforward chain running from genes over neurophysiological processes to overt behaviour. It seems we are back then at genetic determinism. Agency appears to have vanished from the evolutionary scene. Not we ourselves are in charge of what we do. It is our genes that, via the neurophysiological processes that they code for, determine our behaviour. But this need not be the case. EP identifies psychological mechanisms (or modules) at the level of the human mind as the proximate causes of human behaviour. And although genes undoubtedly play an important role in the construction of psychological mechanisms, their workings need not be (and probably are not) fully determined by them. Indeed, EP tries to be as neutral as possible on ontogenetic and heritability issues (Symons 1992, 141). What exactly are psychological mechanisms and with what psychological mechanisms are we equipped according to EP? EP is often circumscribed as 'cognitive science meets evolutionary theory'. From cognitive science (see particularly Fodor, 1983) EP takes over the idea that the human mind consists of modules. Typically, when confronted with some task not all of the human mind is mobilised. Only one module, as a part of the mind, is activated. What part is activated depends on the domain to which the task belongs. Now, what domains are there and, relatedly, what modules does our mind consist of? Here evolutionary 2 Actually, Mayr discusses two more classes of causes. They too can be classified in terms of proximate and ultimate causes. But for the purpose of this paper discussion of the two classes of causes mentioned suffices. 3 theory comes in. EP holds that in order to tell what modules we have and in what domains they are triggered, we have to look at the evolutionary (sub) problems our ancestors had to cope with. The idea here is that although it is clear that individual differential reproductive success was decisive in the evolution of the human species, our ancestors were never directly faced with the problem of how to maximise their individual reproduction rate. What they were directly faced with instead was some component of this. In order to achieve a favourable reproduction rate several evolutionary subproblems had to be solved. Obviously, finding a suitable individual of the opposite sex to mate with was such an evolutionary subproblem. And so are successes in withstanding climatological changes, in staying away from predators and in solving co-operation problems in groups. A central hypothesis in EP is that domains correspond to such evolutionary (sub)problems and that our modules evolved to solve these problems. Perhaps the best worked out example of a module is the cheater-detection algorithm 3 discussed by Cosmides and Tooby (1992). Cosmides and Tooby argue that in order to have ‘TIT FOR TAT’ sustain a social contract, the people in question must be good at spotting cheaters. If people were not good at this, they would be unable to punish cheaters. And the latter of course is crucial to TIT FOR TAT. In Cosmides and Tooby's opinion, the existence of a specialised cheater-detection algorithm is confirmed by the findings in the Wason Selection Task. Peter Wason devised his Selection Task to find out whether people perform well in applying modus tollens, the logical rule underlying Popper's falsificationism. To his surprise, people tended to perform quite poorly. The only exception to this tendency was when a specific type of propositional content was given to the problem. When the problem was framed in terms of a social contract problem, the performance rate increased dramatically from less than 25% to 75%. Apparently, then, Cosmides and Tooby conclude, people avail of a content-specific algorithm, an algorithm that is activated only if the problem has the right propositional content. EP advances a Swiss army knife model of mind on which the human mind consists of a multitude of such special-problem devices. Each of these devices evolved to solve a specific evolutionary (sub)problem in times long foregone, EP argues. But they can still be triggered if people are provided with the right informational input. In this sense, our skulls still house a stone-aged mind. EP revitalises the view that people all around the world share the same basic repertoire of modules. Human nature is said to be the same everywhere. But it is important to recognise that human nature, as EP sees it, is not of a piece. What module is activated, to repeat, depends on the informational input provided. This explains why the same people who behave altruistically in some setting (as in following TIT FOR TAT in social contract settings) behave much more self-regardingly in other contexts. Do our modules prompt us to behave rationally? Among other things this depends on how well adapted modules were to ancestral circumstances and to what extent our present-day circumstances resemble these ancestral circumstances. Behaviour that was optimal then may well be far from optimal now. The Maladapted Mind (Baron-Cohen 1997), for example, discusses current psychopathologies as adaptations to the Environment of Evolutionary Adaptedness (EEA) that in relevant respects was rather different from environmental circumstances we are now confronted with. Cosmides and Tooby (1994) advance the puzzling suggestion that 3 The most controversial example of a module no doubt concerns mate preference. 4 behaviour prompted by our modules may make for better than rational behaviour. What they mean by this is that having a rich repertoire of special problem-solving devices (as EP claims) may be more efficient than having just one general problemsolving device. Modules as special problem-solving devices function as shortcuts, facilitating 'fast-track' learning. Hence they save on time and energy. The flipside of this is, however, that our stone-aged skull may be poorly adapted to solve today's problems. Our modules may prevent us from quickly learning certain useful things. Thus modules enable and facilitate. But they also inevitably constrain and confine (what can be learned or done). Criticism Over the last years EP has been criticised from several vantagepoints. For example, EP has been criticised for endorsing an extreme form of adaptationism. EP is said to entertain the view that all enduring life forms are optimal solutions for problems produced by natural selection. In the preceding paragraph I already touched upon this issue. I believe this criticism is cogent only up to some point. EP abides to the adaptationist program. But it does not take for granted that all features of our minds are perfectly adapted to the EEA (let alone that all features are perfectly adapted to our present circumstances). EP has also been criticised for embracing genetic determinism. But although at least some EPists seem to be wedded to the view that the ontogenetic development of our mind’s modules is genetically specified (in some sense; see below), no one of them goes as far as to claim that this development is fully genetically determined. More reasonable critiques seem to come from those who criticise both the way Cosmides and Tooby conducted Wason selection task experiments and the conclusions they draw from them (see, for example, Evans and Chang 1998). Contrary to what Cosmides argues, critics hold that it is not at all clear whether her account of these experiments in terms of biologically evolved algorithms is superior to what this was supposed to supersede: the permission and obligation schema proposed by Cheng and Holyoak. In fact, some even fail to see a real difference between the two accounts (Lloyd 1999, 221). Several commentators argue that transmitted culture and cultural evolution in general are given short thrift in EP. Although there may be some problems with identifying exactly what (as an analogue of the gene) is transmitted in processes of enculturation, there seems to be little doubt that ‘things’ like values, norms, customs, conventions, notions, ideas and tunes spread and decline (in ways that differ from genetic inheritance) in populations. ‘Evoked culture’, the notion EP introduces, may fall short of capturing the most significant elements in processes of cultural evolution. EP tends to explain cultural differences by referring to different ecological circumstances evoking different behavioural responses from shared modules having a conditional structure. Accounts in which more significance is allotted to ‘transmitted cultural evolution’ include Lumsden and Wilson (1981), Boyd and Richerson (1985) and Durham (1991). More recently, in The Cultural Origins of Human Cognition, Tomasello (1999) argues that some 250.000 years are too short for uniquely human cognitive modules to have evolved in processes of biological evolution. Tomasello argues that it is much more likely that in this relatively short period of time (in terms of phylogenetic time) only one uniquely human biological adaptation accrued: the capability of intentional and causal thinking. The capability of intentional thinking includes the essential capability to understand others as intentional agents (also having intentions, beliefs, plans and the like). It is this 5 capability, Tomasello argues, that has allowed humans to obtain ratchet effects: the maintenance of useful creative inventions. The problem with other primates is not so much that they are much less creative than humans, but that newly invented practices and artefacts are not faithfully transmitted socially so that slippages backward occur time and again. Buller (2000) contains what seems to be a sustained and well-informed criticism of EP from the perspective of contemporary neuroscience. Buller introduces a useful taxonomy of different claims made by EP, ranging from very general to specific ones. At the most general level, EP’s claim that the human mind is the product of evolution seems to be uncontested. At a somewhat less general level, EP’s claim that our brain houses some specialised psychological mechanisms is not controversial either. But in the light of contemporary neuroscience, Buller argues, EP’s claims that go beyond this, its more specific claims, are questionable. Cosmides and Tooby’s contention that the human mind consists of hundreds or thousands modules seems to be unwarranted (see also Samuels 1999). And so are their claims that all of these modules are uniquely human, stem from the Stone Age, and are universal and uniform. The most interesting point Buller makes, however, is that EP may well be right that our brains produce specialised information-processing solutions to recurrent evolutionary problems, but that they presumably got the way wrong in which the brains produce the solutions. EP claims that adaptations are to be sought among the products of brain development – that is, among the special-purpose brain structures that emerge during the course of brain development. Recent research in neuroscience suggests, however, that the fundamental adaptation rather is the brain’s developmental plasticity. The relatively stable problem-specialised structures EP posits emerge as by-products of interactions of the developmentally plastic brain with particular environments. The picture emerging from this is that, contrary to what EP claims, what is ‘innate’ is not a multitude of genetically specified modules, but rather a relatively small number of learning biases in the brain that heighten the responsiveness to certain classes of stimuli. A process of gradually branching domain specificity, Buller argues, unfolds only later. Buller’s discussion here no doubt is very informative. But one is left wondering how damaging it is for EP. For if Buller is right, contemporary neuroscience agrees with EP that in the end, after the brain has gone through a developmental process, many special-purpose modules do come about (see also Cummins and Cummins 1999). It seems that this vindicates one of EP’s central claims. Buller argues that the developmental path leading to the emergence of the modules is quite unlike how EP envisions it. But perhaps with the exception of a few proponents, EPists do not make claims about what drives brain development. What is more, they are happy not to burn their fingers on how brain and mind developments hang together. And, furthermore, as we have seen in section 4, many EPists prefer to remain agnostic 4 about hereditary and ontogenetic matters. One last point of criticism is that our mind’s modules may not be informationally rich (Mallon and Stich 2000). That is, they do may not provide us with clearly 4 On the other hand, it can be argued that closing their eyes for these issues and for developments within adjacent disciplines deprives EPists not just of valuable sources of (possibly disconfirming) evidence and information, but also of allying with possible bystanders. It seems, for example, that EP can find some support in the neurosciences in the work of LeDoux, Gazzaniga and Damasio. 6 specified biases and heuristics in our learning efforts. Knowledge of the modules we have then falls far short of predicting how we will behave in various circumstances. Environmental factors, social and natural, then can be argued to have a greater effect on how we learn and behave. Sometimes EPists engage in some sort of transcendental argument. They argue, for example, that every learning effort requires, as some sort of a conditio sine qua non, innate capabilities that allow us to learn. The logic behind this may be impeccable. But if the capabilities are informationally poor, this need not tell us very much about how we actually learn. 3. Sober and Wilson’s Multi Selection Theory (MST) In order to understand Multi Selection Theory (MST), let us first have a look at the 5 notion of group selection. Group selection became a discredited notion for many soon after Wynne-Edwards (1962) invoked the notion as an explanation of ultrasocial behaviour in nature. There seem to have been two major reasons for this. The first reason is empirical in kind. Starting with Williams (1966), many biologists (and evolutionary theorists in general) have not so much doubted that 'group selection' denotes a genuine possibility in nature, as that the conditions for the force of group selection to be stronger than that of individual selection are so severe that they are not likely to be met in reality. The second reason is conceptual in kind. Biologists increasingly felt that, on the basis of Hamilton's (1964) inclusive fitness theory (based on kin selection), and of Trivers' (1971) theory of reciprocal altruism, they could explain all kinds of ultrasocial behaviour in what was regarded as orthodox Darwinian terms of individual selection (or even in terms of gene selection; see Dawkins, 1976). Hence there was no need for the notion of group selection. Sober and Wilson (1998; see also Wilson and Sober, 1994) want to rehabilitate the notion of group selection. In fact, Sober and Wilson want to establish at least two 6 different points. The first is that the longstanding eschewal of group selection in evolutionary biology is ill founded. Sober and Wilson want to rehabilitate the notion of group selection. They argue in particular that if individual organisms are accepted as paradigm cases of vehicles in evolutionary processes, as all evolutionary biologists do, then there cannot be well founded reasons not to accept groups as vehicles also. The second point Sober and Wilson want to make is that evolutionary altruism requires group selection to evolve. Evolutionary altruism cannot possibly be the outcome of individual selection. I will argue that in their attempt to establish the two points Sober and Wilson put forward two notions of a group that are not only different from one another, but that are even incompatible with one another. First consider Sober and Wilson’s argument that evolutionary altruism requires 7 group selection to evolve. As a starting point of their argument, Sober and Wilson assume that individual selection is omnipresent and will inevitably lead (unless counteracted by other evolutionary forces) to evolutionary selfishness. Sober and Wilson define ‘evolutionary selfish behaviour’ as behaviour that enhances the 5 For a more detailed analysis and discussion of Sober and Wilson’s notion of group selection, see Vromen, (1999b). 6 Another point Sober and Wilson want to establish in their book is that no fully compelling arguments can be made for or against the existence of psychological altruism. In this paper I only deal with evolutionary altruism. 7 The converse does not hold, however: group selection may also lead to non-altruistic, ‘nasty’ behaviour. 7 (inclusive) fitness of the behaving entity at the expense of the (inclusive) fitness of other entities. Evolutionary altruistic behaviour is taken to be the opposite of this: behaviour that enhances the (inclusive) fitness of others at the expense of one’s own (inclusive) fitness. The argument here is the familiar one (familiar, that is, in evolutionary theory) that no matter how well (evolutionary) altruists are represented in a population, and how beneficial this is for all the individuals in that population (in terms of their individual inclusive fitness), altruists will always be at an evolutionary disadvantage compared to selfish individuals. For selfish individuals will engage in free loading or free riding. They will ruthlessly exploit the selfsacrificing behaviour displayed by altruists. Altruists cannot withstand treachery or mutiny from within. That is why in every population individual selection, if it is the only evolutionary force at work, will eventually wipe out altruists. Thus it takes some other evolutionary force (or forces) to prevent this from happening. Sober and Wilson argue that group selection exactly is this evolutionary force. Sober and Wilson carefully hedge this claim, however. When individual selection does not act alone, but is joined with group selection, altruism can survive. This is not to say, Sober and Wilson hasten to add, that individual selection is always accompanied by group selection. Group selection is not (or at least need not be) as omnipresent as individual selection. Furthermore, if there is group selection, it may not be strong enough to counter the altruism-eroding effects of individual selection. What will happen depends on the relative strengths of the two evolutionary forces. And, finally, even if group selection is powerful enough to overrule individual selection, altruism may still not evolve. Group selection can also generate other types of evolutionary behaviour than altruism. In sum, Sober and Wilson claim that group selection is a necessary, but not a sufficient condition for altruism to evolve. Now what exactly is group selection here? How does it work? And how can altruists survive as a consequence of it? The key idea here is temporal isolation of interaction clusters. If the individuals in a population have an equal chance of interacting with one another, the eventual demise of altruism cannot be avoided. But if a population is structured into groups of individuals only interacting with one another (with respect to the evolution of some trait), then altruists have better prospects. To simplify maters, say a population consists of two such groups, one consisting only of altruists and the other consisting only of selfish individuals. Now if the individuals in the groups only interact with other individuals in the same group for a while, the first ‘altruist’ group will do better (in the sense of all its members leaving more offspring) than the second ‘selfish’ group. As a consequence, the first group’s share in the population will grow at the expense of the second group’s share. It may appear that this already is enough to make Sober and Wilson’s point. After all, it seems that if the process just described goes on long enough, the second groups’ share in the population will eventually be reduced to nil. But this is not the case. And Sober and Wilson acknowledge it is not. For we should not forget to take individual selection into account also. No matter how successful the first group is, it cannot resist invasion of selfish mutants. Sooner or later a selfish mutant will make its appearance in the group, exploiting the altruists. Selfish individuals can have no better soil to thrive on than a group of altruists. So the share of selfish individuals within the group will rapidly increase at the expense of the altruists. This will go on until the share of altruists within the group has fallen down to nil. Since this will happen in all groups, the inevitable outcome still seems to be that in the end we will be left with only selfish individuals in the population. 8 Sober and Wilson argue that what is needed next to temporary isolations of groups are re-assortments of groups now and then. Only if altruists in different groups occasionally ‘join forces’ (that is, form new groups) can the eroding consequences of individual selection be compensated for. In sum, as Sober and Wilson see it two things are both needed to thwart the eroding effects of individual selection: temporary isolation of groups and periodic re-shufflings of groups (resulting in the formation of new groups). These two things should suitably alternate in temporal sequences in order to forestall elimination of altruists in the larger population. The notion of a group involved here is a very minimal (or thin) one. A group is 8 defined here solely in terms of interactions. It is enough for some individuals to interact only with each other for some period of time to call them members of the same group. Indeed, as other commentators also have noted (see, for example, Maynard Smith 1998), Sober and Wilson are prepared even to call two individuals that interact only once a group! What is more, the individuals interacting need not have anything in common with one another except that they interact with one another. Individuals can be radically different from one another in all respects, as long as they for whatever reason only interact with each other they are part of the same group. Let us henceforth call this minimal notion of a group the interaction notion of a group. This is not the only notion of a group Sober and Wilson entertain. In the argument they develop to establish their other point – that groups can be vehicles just as individual organisms– they invoke a different notion. This can be called the shared fate notion. The key idea here is that the evolutionary fate of individual group members can be as dependent of the evolutionary fate of the group they are part of as the fate of genes is dependent of the individual organisms they inhabit. The individuals in a group are ‘in the same boat’. They are integrated so tightly and tied together so strongly in a group that when the group expires in competition with other groups its individual members go with it. At first it may seem that the unity of a group cannot be as tight as that of an individual organism. After all, it seems that, contrary to genes within an organism, individuals within groups can go their own way and can act against the interests of the group as a whole. Sober and Wilson argue that sometimes the bonds between individuals in a group are so strong that the individuals can be regarded as inseparable parts of a comprising unit. More importantly, the above comparison between genes within an organism and individuals within a group falsely assumes that there is some sort of self-evident harmony between genes and the organism housing them. From an evolutionary perspective the harmony between genes and the organism (if there is one) is itself an adaptation. Far from taking this for granted, evolutionary theory seeks to explain the internal integration and co-ordination of an individual organism as a solution to the problem of how to cope with freeloading selfish genes in the organism. Thus the internal integration and co-ordination of individuals in groups is no less natural or elementary than that of genes within a well-functioning organism. Both can be regarded as adaptations. 8 To be fair, Sober and Wilson also demand that interactions should have different fitness consequences with respect to some trait. But I think that taking this additional requirement into account does not affect our discussion here. 9 The latter is granted by proponents of the Dawkins’s ‘selfish gene’ perspective. Yet, Sober and Wilson rightly observe, the selfish gene proponents accept individual organisms as paradigm instances of vehicles in evolutionary theory. Sober and Wilson argue that selfish gene adepts are bereft of a sound argument to refuse accepting groups as vehicles also. The selfish gene adepts inist that what ulimately counts in evolution is the relative success of repicators to propagate themselves in a population. Sober and Wilson agree. But this does not establish that there can only be individual selection. In arguing this selfish gene adepts commit the averaging fallacy: by (rightly) insisting that the effects of selection always are to be counted in terms of frequencies of replicators in some population they (wrongly) brush genuine differences in causes under the carpet. A change in frequencies of replicators registers the average effect of possibly different evolutionary forces. Such a change may be caused by individual selection working alone, for example, or by the joint operation of individual and group selection. This shows that Sober and Wilson believe the group selection debate is about vehicles, not replicators. They also urge evolutionary theorists to distinguish carefully between theoretical perspectives and actual causal processes. They furthermore argue that evolutionary theory, unlike the selfish gene perspective, should respect and reflect differences in processes. Their own Multilevel Selection Theory (MST) purports to be exactly that: an attempt to clearly bring out differences in processes in evolutionary theory. As indicated above already, the notion of group selection associated with the shared fate notion of a group is that of groups competing with one another as behavioural units. One can think here of tribal warfare, for example. The type of group selection involved here can be called group competition by means of group interaction. The notion of group selection associated with the interaction notion of a group is quite different. This notion comprises both what can be called group competition without group interaction and selective individual interaction without group competition as indispensable ingredients. ‘Group competition without group interaction’ is intended to bring out the idea that groups can compete for shares in a population without ever getting in touch with one another. It is simply a matter of different growth rates (in terms of propagating replicators in a population) how large a group’s population share turns out to be. ‘Selective individual interaction without group competition’ it meant to convey the idea that periodic re-assortments of groups (that, as we have seen, are also needed to sustain altruism in a population) are brought about not by groups, but by individuals that selectively pick out each other to form new groups. It sometimes seems that Sober and Wilson are looking for a notion of group selection that covers more than just group competition by means of group interaction. But it also seems that Sober and Wilson run into conceptual problems here. It is not just that the second notion of a group is thicker, or more substantial than the first one. Strictly speaking, the notion of group selection associated with the interaction notion of a group rules out the notion of group selection associated with the shared fate 9 notion of a group. Consider tribal warfare or platoons in combat situations. If we take the interaction notion, we should not speak of different groups competing with one another, but rather of some individuals competing with other individuals of the same group. For on the interaction notion the distinguishing feature of a group is that its individual members only interact with each other (for some period of time). It is clear that when it comes to tribal warfare, even though tribes can be considered to be behavioural units, actual encounters between individual members of different 9 Other terms for the specific type of groups I have in mind here are clans, cliques and gangs. 10 10 tribes are very likely to take place. But if this is so (and they do have fitness consequences for the individuals involved!), then the interaction notion of a group implies that the individuals involved do not belong to different groups. In conclusion, Sober and Wilson succeed in making a convincing case for what can be called the structured deme theme: population structure matters. But they fail to demonstrate convincingly that taking population structure into account forces one to take group selection in the more substantive ‘shared fate’ sense on board as well. What outcomes emerge from evolutionary processes depend on whether there is or is not population structure and, if so, what population structure there is. It matters whether all individuals in a population interact with one another. If individuals for some period of time only interact with some of the other individual in a population, then outcomes can accrue that cannot obtain if there were no such groups. This seems to be the whole point of ‘interaction’ group selection. It remains to be seen, however, whether it is really meaningful to speak in all such cases of group selection taking place alongside individual selection. I do not think it is. Especially if one’s theoretical outlook is to reflect genuine differences in causal processes, as Sober and Wilson argue it should, then this does not seem to make much sense. It seems that different causal processes are going on only if there are groups in the more substantive ‘shared fate’ sense. Thus only if there are such groups it seems to make sense to talk of group selection. In other cases it seems to be more appropriate not to invoke the notion of group selection and instead to talk of individual (and genic) selection under different conditions. Individual selection can take place under conditions of random pairing, for example, or under conditions of selective pairing. 4. Agreements and disagreements Both EP and MST tentatively accept the adaptationist programme in evolutionary theorising. The central presupposition guiding the adaptationist programme is that persistent and widespread instances of organised complexity are adaptations brought about by natural selection. It is not denied in this programme that other forces than natural selection can be (and de facto are) driving evolutionary processes. And it is not denied either that features or traits well adapted to circumstances prevailing in the evolutionary past can be ill adapted to circumstances prevailing today. What is maintained only is that natural selection is the only known evolutionary force capable of producing enduring instances of organised complexity. EP and MST do seem to have different reasons for accepting the adaptationist programme, however. While EPists seem to embrace the adaptationist programme wholeheartedly, Sober and Wilson seem to accept it only for the sake of argument. What Sober and Wilson primarily set out to argue is that even if natural selection were the only or all-powerful evolutionary agent, it need not work only on individual organisms. In endorsing the adaptationist programme EPists rely on Williams (1966). Williams (1966, 1971) also has been one of the most ardent opponents of group selection. Yet, although EPists seem to be somewhat sceptical of the relevance if not cogency of group selection, they seem to leave open the possibility that group selection is a cogent and significant force in evolution (Buss 1999). Edward O. Wilson argues that 10 An exception would be that all individuals of the one tribe surrender to those of another one without there being any actual clash between any individuals. 11 research on epigenetic rules has pointed out that choices are group-dependent, but he immediately adds to this that the extent to which this is the case varies strikingly from category to category of behaviour (Wilson 1998, 227). Wilson does not go into the issue, however, whether recognition of the group-dependence of choices entails a theoretical commitment to group selection à la Sober and Wilson. When discussing the possibility of psychological altruism, Sober and Wilson accept folk psychology. They argue that assuming that their desires and beliefs (as the proximate causes of their behaviour lead persons) is fully compatible with evolutionary theory. By contrast EPists believe that modules of the mind are the proximate causes of human behaviour. EPists do not seem to accept folk psychology. What they tend to stress is that our mind’s modules predispose our behavioural patterns in ways that we are most of the time not aware of. No doubt more agreements and disagreements between EP and MST can be discerned. Sober and Wilson emphasise the importance of cultural transmission in human behaviour much more than EP does, for example. Here I want to focus on what I take to be one of the central agreements. EP and MST, it can be argued, share to some extent the same research agenda: how to explain (various forms of) social cooperation between non-kin? In particular, why do individual human beings that are not genetically related to one another engage in forms of mutual co-operation even though it is not in their evolutionary interest to do so? What is more, both prominent EPists such as Cosmides and Tooby (1992) and Sober and Wilson relate to Axelrod’s Evolution of Cooperation to make their respective points. Axelrod’s analysis of Tit For Tat proves to be an interesting point of departure to discuss differences between 11 EP and MST more thoroughly. Cosmides and Tooby (1992) embed postulating the existence of a cheater detection module in the context of a discussion of Axelrod’s Tit For Tat. The strategy Tit For Tat implies that cheating (or defecting) by some individual in some pairwise encounter is reciprocated by the other individual in the next encounter between the two. Thus Tit For Tat essentially assumes reliable and accurate detection of cheaters. Their findings in the Wason Selection Task suggest that this assumption is realistic. What is especially germane for the present discussion is that Cosmides and Tooby hypothesise that the module of cheater detection evolved to solve social contract problems. The module evolved in situations in which it was in the interest of all to uphold some social contract. In such situations all had an interest in finding out whether or not the confidence put in others (by co-operating themselves) was not violated. Cosmides and Tooby further hypothesise that the cheater detection module only is activated if people think they are in a social contract setting. In other settings quite different modules may be activated. People may then be paying attention to quite other things. Sober and Wilson do not believe that by Axelrod’s analysis of Tit For Tat exhausts all possible varieties of evolutionary altruism. But they put a lot of effort into showing that Tit For Tatters are altruistic and that Axelrod’s explanation of the success of Tit For Tat amounts to a specific invocation of group selection. The key to understand Sober and Wilson’s position here is provided by the way in which they identify 11 Axelrod’s treatment is far from uncontroversial (for a cogent criticism from a game-theoretic perspective, see for example Binmore 1994). The points EP and MST want to make here do not seem to fully dependent on the validity of Axelrod’s treatment, however. 12 groups. In Sober and Wilson’s discussion each box of the payoff matrix denotes a separate group (Sober and Wilson 1998, 80). In other words the population always contains as many groups as there are boxes in a matrix. Based on this identification of groups, Sober and Wilson focus on the following two conditions an altruistic strategic type has to meet. First, the type must have the relatively lowest fitness in mixed groups. And, second, the group performance should be better (again in fitness terms) the more the altruistic type is represented in a group. They argue, for example, that the C-strategy in a PD is altruistic because it meets both conditions. In order to see that C meets the first condition, we have to look at the box representing an interaction between a C- and a D-strategist. We then see that the fitness (increase) of the C-strategist is lower than that of the D-strategist. Thus the first condition is met. In order to see that the second condition is also met we first have to add up the two figures in each box. We then see that the sum of the (C, C) interaction is higher than that of the (C, D) interaction, which in turn is higher than that of the (D, D) interaction. Thus the second condition is met as well. This argument only makes sense, I submit, if the pairs of interactions remain isolated for some time (for example, if there is assortative pairing). Only then, only if there is some endurance of group competition without group interaction, does it make sense to compare the relative ‘group’ performance of the (C, C), (C, D) and (D, D) pairs. For only then does it make sense that the gene frequency of the C-strategy increases in the population because the ‘group’ (C, C) grows faster than the ‘group’ (C, D), which in turn grows faster than the ‘group’ (D, D). It is clear that this latter condition is not met if there is random pairing (which is assumed by Axelrod). Sober and Wilson’s analysis here once again shows, I think, how facile their interaction notion of group selection can be. It does not make much sense to argue that group selection is invoked in Axelrod’s account of how Tit For Tat can come to dominate a population. It is quite contrived to distinguish two selection processes in this case. Pace Sober and Wilson I would argue that instead of individual and group selection jointly being at work here only individual selection is taking place. And it also does not seem to make sense to artificially distinguish primary and secondary behaviour here (Sober and Wilson 1998, 142-149). One cannot meaningfully argue 12 that primary and secondary behaviour evolve independently of each other here. Primary and secondary behaviour are inextricably tied up with one another in Tit For Tat. The threat of ‘punishments’ is built into Tit For Tat. If there is one element of group selection in Axelrod’s account of the evolution of Tit For Tat, it is to be found in the earliest stages of Tit For Tat entering a population of defectors. Axelrod (and Hamilton) clearly saw that Tit For Tat has to enter a 13 population of defectors in a cluster. In this cluster Tit For Tatters must for some time be shielded from encounters with defectors. Only after this cluster has gained some 12 Curiously enough, Sober and Wilson themselves acknowledge that primary and secondary behaviours often evolve as a package (Sober and Wilson 1998, 146). See also Binmore for similar criticism. 13 It is easy to verify that (given the PD matrix Axelrod uses) Tit For Tat can only invade a population of Defectors in random pairings if the following condition is met: x > (1 - w)/(3w - 1), where x stands for the frequency of Tit For Tatters in the population and w for the discount parameter. Conversely, in order for a population of Tit For Tatters to resist invasion of Defectors, w must exceed 1/2. If w = 2/3, for example, x must exceed 1/3. In this numerical example, then, the cluster of Tit For Tatters only interacting with each other must first occupy a share in the population of at least 1/3 for Tit For Tat to be able to spread further in the population via random pairings. 13 non-negligible size (some critical value) can Tit For Tatters spread the population in random matings. One can say that in the earliest phase by selective pairing only with other Tit For tatters, Tit For Tatters formed a group within the encompassing population. On the basis of the foregoing discussion it may be useful to introduce a taxonomy of different forms of altruism. Let us define altruism somewhat loosely as a behavioural disposition that, if activated, prompts individuals to do things that enhance the welfare (or well-being, or…) of other individuals at the expense of one’s own welfare (or well-being, or…). Note that I do not define altruism here in terms of antecedent intentions for, or causes of behaviour. The definition does not tell why or how altruistic behavioural dispositions are triggered. Then we can distinguish at least 14 between the following four forms of altruism. 1. Unconditional altruism: an individual is unconditionally altruistic if it always displays altruistic behaviour regardless of the individual(s) it interacts with. 2. Like-seeks-like altruism: an individual is like-seeks-like altruistic if it always displays altruistic behaviour regardless of the individual(s) it interacts with, but that prefers to interact only with like-minded altruists. 15 3. Groupism: an individual is groupish if it only displays altruistic behaviour in interactions with fellow group members. 4. Content-specific altruism: an individual is content-specific altruistic if it only displays altruistic behaviour when provided with stimuli with the appropriate propositional content. Both unconditional and like-seeks-like altruism relate to fixed behavioural types (or to individuals following fixed strategies). In both types a disposition to behave altruistically is activated regardless of what properties the individuals interacted with have. No matter whether the individuals interacted with are genetically related, friends, like-minded altruists, fellow group members or anonymous strangers, unconditional and like-seeks-like altruists always behave altruistically. The only difference between like-seeks-like altruists and unconditional altruists is that the former prefer to interact only with like-minded altruists. This means that if likeseeks-like altruists are given the chance, they will only interact with like-minded altruists, leaving other types to themselves. Note that the latter does not imply that like-seeks-like altruists are always able to act on this preference. Opportunities for selective play (or pairings, or matings) may be absent temporarily or permanently. Groupish individuals do not always behave altruistically. Groupish individuals only behave altruistically towards fellow group members. Note that groupish individuals need not prefer to interact only with fellow group members or like-minded groupish individuals. The behavioural disposition involved in this type of altruism has a conditional structure. When activated, the disposition prompts altruistic behaviour if there is interaction with fellow group members, and non-altruistic behaviour if other conditions obtain. A strong form of groupism is displayed if the disposition to 14 I do not claim that the list provided exhausts all possible forms of altruism. Far from that. But I believe that it allows us to clarify relevant differences between positions taken in EP and MST. 15 It can also be argued that groupism is not a particular form of altruism. It can also be called a separate third possibility next to egoism and altruism (Gilbert 1994). 14 behave altruistically is always activated whenever a fellow group member is encountered. Contrast this with content-specific altruism where the disposition to behave altruistically is activated only under certain circumstances. Here the disposition itself need not have a conditional structure. It is rather that it is only activated if the right conditions are met. Only if the stimuli have the right content is the disposition to behave altruistically triggered. This leaves open the possibility that the ‘altruistic disposition’ is not triggered in some interactions with fellow group members, whereas it is triggered in encounters with individuals not belonging to one’s group. Both Axelrod and Sober and Wilson, in their interaction notion of group selection, seem to invoke like-seeks-like altruism. Axelrod’s analysis of the evolution of Tit For Tat assumes that in the earliest stages Tit For Tatters can act on their preference to interact only with other Tit For Tatters. Only after the frequency of Tit For Tatters has exceeded some critical threshold value can Tit Fir Tat further invade a population of Defectors through random pairings. In their interaction notion of group selection Sober and Wilson assume that individuals of different ‘groups’ first do not interact with each other for some period of time, but that after that the barriers are lifted. When the barriers are lifted, altruists are assumed to seek other altruists to form new groups. Then the newly formed groups are assumed to be isolated again for some period of time, and so on. As far as I can see, the most plausible interpretation of this ongoing sequence is that altruists prefer to interact only with fellow altruists, but are sometimes temporarily prevented from acting upon this preference. If they are temporarily locked in groups, they cannot escape from interacting with whatever types (non-altruists included) are present in their group. But if the borders open up, they can escape being exploited by free riders by acting upon their preference to interact only with like-minded altruists. But if my interpretation makes sense, then Sober and Wilson owe us an explanation of why individuals are temporarily are stuck within their own ‘group’, why after that they are perfectly free to find like-minded individuals to form new groups, and why and how after that groups once again are closed off from each other. With certain types of animals, plausible stories can perhaps be told about why we would observe such sequences or cycles unfolding. The story Sober and Wilson tell about the brain worm, for example, is fascinating and convincing in this respect (Sober and Wilson 1998, 18). But with human beings, the situation may be different. Even if we assume that persons are of a fixed behavioural type, it seems that we cannot ignore the possibility of migration. Only under very extreme circumstances, it seems, can persons be prevented from moving over to other ‘groups’ (or to found new ones). Groupism seems to be quite close to Wilson and Sober’s shared fate notion of group selection. The notion of groups competing with one another as well-integrated units (or as ‘superorganisms’) suggests the combination of in-group solidarity and outgroup hostility characteristic of groupism (Gilbert 1994). Individuals here need not prefer interaction with fellow group members. Indeed, they may even be more eager to engage in combat with individuals of competing groups than to seek contact with fellow group members. Content-specific altruism, finally, seems to be depicted in EP. If people perceive situations and problems in terms of upholding some social contract, EP argues, then a disposition to behave altruistically is triggered. But if they perceive situations differently, for example as a situation in which their position in the pecking order in a group is endangered, quite a different disposition may be triggered. They may engage in competitive status-protecting and status-seeking 15 behaviour within the group then. What this shows, I think, is that EP allows for the possibility that group members need not always behave altruistically towards one another. Conversely, EP in principle also seems to allow for the possibility that altruistic behaviour is directed towards individuals that do not belong to one’s own group. It all seems to depend on whether individuals are faced with the right stimuli. This raises the interesting question whether EPists believe social contracts can be selfsustaining also on larger scales than relatively small-sized groups. 5. What difference does it make for economics? EP and economic theory Several economists have reflected on the possible significance of EP for economics. Not surprisingly, perhaps, economists of different persuasions reached different conclusions (for a more detailed discussion, see Vromen 1999 and 2001b). In this part of the paper I want to focus on economists who argue that EP is of interest for economics mainly because it holds out the prospect of providing a non-arbitrary identification and explanation of the preferences people actually have. I am thinking here, for example, of Ben-Ner and Putterman (1998 and 2000), Zywicki (2000), Gintis (2000) and work done in the so-called ‘indirect evolutionary approach’. Although no explicit references are made to the EP literature in this latter approach (as far as I know), I believe it is very much in the same spirit. In order to see what message the ‘indirect evolutionary approach’ entails for economics, let us have a look at Güth and Yaari (1992), the paper in which it was introduced. Güth and Yaari (1992) set out to show how a genetically determined preference for reciprocal behaviour could possibly have persisted in ruthless selection processes. In order to do this, Güth and Yaari consider a simple game in extensive form with two players. They confine their attention to an analysis of two different ‘gene types’ (as players): one player is assumed to be genetically disposed to reciprocate harm with harm, whereas the other player is assumed to be genetically disposed to yield to a harming opponent. Güth and Yaari then show (assuming in addition complete information and individual rationality) that the reciprocating type is the only Evolutionarily Stable Strategy in the game. The interesting thing about this result is that, appearances notwithstanding, a preference to reciprocate is reproductively successful, not despite but precisely because of its determined reciprocating nature. Precisely because people genetically endowed with a preference to reciprocate are determined, regardless of the personal costs involved in doing so, to harm others if these others harm them (and the others know this), can 16 they avoid being exploited by others. What makes Güth and Yaari’s evolutionary approach indirect rather than direct is that evolutionary forces and pressures are assumed to affect our behaviour indirectly. By contrast, evolutionary game theory assumes that selection pressures mould (or, rather, moulded) our behaviour or our behavioural strategies directly. The basic idea underlying the indirect evolutionary approach is that selection pressures determine what preferences (or preference profiles) evolve. And our 16 The crux of the matter here, it can be argued, is that ‘subjective’ valuations (determining what the players do) are different from ‘objective’ results (determining what type has survival value). 16 preferences in turn guide our behaviour. The indirect evolutionary approach’s basic logic, Huck writes, is that “… preferences guide behavior, behavior determines fitness, and fitness drives the evolution of preferences” (Huck 1997, 773). It may appear that we come full circle here. But for three reasons there is no circularity involved here. First, there is temporal order. The preferences guiding our behaviour now evolved in the past. Second, what we prefer need not coincide with what is selected for. As Güth and Yaari’s example illustrates, what subjective valuations people attach to certain options have may differ from the options’ objective (expected) reproductive success. “While choices are motivated by the preferences, which may include an internal commitment to norms, objective evolutionary success of preference types depends on "external incentives.”” (Güth and Ockenfels 2000, 396). Third, and perhaps most importantly, preferences do not fully determine behaviour. To determine what behaviour follows from certain preferences, additional information has to be provided (or: additional assumptions have to be made). Güth and Yaari explicitly state that “… we assume individual 17 rationality for given preferences” (ibid., 23). What is the rational thing to do depends not just on prevailing preference profiles, but also on prevailing constraints and opportunities. Thus in the indirect evolutionary approach we are not taken to be prisoners or puppets of these evolutionary processes. Even though we may be called prisoners of our preferences, the indirect evolutionary approach “… does not neglect individual’s ability to adjust their behavior in different environments when studying a selection process” (Huck, 1997, 777). Proponents of the indirect evolutionary approach acknowledge that their approach is reminiscent of Becker’s views on how economics and sociobiology could be merged (see Huck 1997 and Bolton 2000). Becker was one of the first economists to spot the potential use the then emerging field of sociobiology could have for economics. In Becker (1976) an attempt is made to merge sociobiology with economic theory. One of the things economists can learn from sociobiology, Becker argued, is that people with altruistic preferences (such as ‘Big Daddy’) can survive in a harsh, ‘material’ world. Becker believed sociobiology would not in any way contravene standard economic analysis. They rather complement each other nicely. Economists can teach sociobiologists that rational individuals sometimes can act quite unexpectedly. Thus altruistic ‘Big Daddy’ can evoke co-operative behaviour from a selfish ‘Rotten Kid’. Conversely, sociobiology can tell us, Becker hoped, what preferences our evolutionary past has passed on to us, and economic theory can tell us how to act rationally upon them. One may say that evolutionary psychology, as an heir of sociobiology, now is cashing in this hope. We may even go further back in time and say that with the advent of evolutionary psychology, psychology finally delivers what Robbins (1932) once hoped psychology could offer to economics: firmly established knowledge about our preferences. This ‘Beckerian’ mix of sociobiology and standard economic analysis, we have seen, can also be observed in the indirect evolutionary approach. Proponents of this approach also set out to demonstrate why in the one situation a single selfish individual is capable of inducing non-selfish individuals to behave non-cooperatively, while in another situation a single non-selfish individual is capable to let selfish individuals behave in a co-operative way (Fehr and Schmidt 1999, 819). All 17 Witness also Bolton’s characterisation of the approach: “…conceptualizing evolution as working on preferences, and then applying rational choice to explain actual decision making” (Bolton 2000, 286). 17 this leaves the basic analytical structure of economic theory intact. As Huck argues, there is no need “… to forsake the elegance provided by the tools of neoclassical economics” (Huck 1997, 773). It may well be, however, that the real lesson evolutionary psychology entails for economics is less consoling for received economic theory. For evolutionary psychology may imply a somewhat different picture of the ‘given’ determinants of our behaviour and of how they interact with the environment to produce behaviour. Evolutionary psychology may undermine the picture of fixed preference profiles that are invariant across different situations. The point is not so much that evolutionary psychologists object to the idea that we have fixed preferences. The innate modules, mechanisms and rules they are talking about can be called (and sometimes are called) preferences. The point rather is that the modules do not constitute a unitary whole that is operative across different situations. Evolutionary psychologists may believe that there is a universal and uniform human nature. But they do not believe human nature to be unitary. The different modules cannot be thought of as different terms in an all-embracing and all-purpose utility function. If one would nevertheless want to cling to the notion of a utility function, each module represents a separate utility function, the relevance of which is restricted to a special purpose only. Human nature, as depicted by evolutionary psychology, is fragmented. Invariant preference profiles and all-embracing, all-purpose utility functions suggest a unity or homogeneity in the determinants of human behaviour that according to evolutionary psychologists simply is not there. Contrary to what the notions of an invariant preference profile and of an all-embracing and all-purpose utility function suggest, there are no tradeoffs between the mind’s modules. There are no tradeoffs between modules because not all modules are operative in all situations. Typically, only one module (or a few modules at most) is activated in a situation. The other modules remain inactive. They are in the switch-off mode, as it were. The economists discussed above assume that the motivation profiles of individuals are invariant across different contexts. Behaviour may still be context-dependent, because different contexts provide different opportunities for individuals. By contrast, evolutionary psychology seems to hold that motivation profiles themselves are context-dependent. To be more precise, individuals are assumed to have several behavioural dispositions, each of which is ‘switched on’ in suitable environmental contexts. Behaviour is context-dependent not because opportunities differ from the one context to the other. Behaviour is context-dependent because different contexts bring different stimuli with them triggering different modules. Moreover, modules do not bear only on the evaluative aspects of our behaviour. They do not relate only to what we desire, value, cherish and so on. They also relate to how we think and learn and how we arrive at our behaviour. In other words, cognitive aspects are also involved. Evaluative and cognitive aspects, it seems, are inextricably interwoven in the operation of modules. As Wilson argues, epigenetic rules do not only determine what things we value and how much we value them. They also determine what options we consider in our attempts to obtain the things we value most. And they also determine the heuristics we entertain in our ‘prepared’ search behaviour. In short, they also affect the cognitive aspects of our behaviour. Sometimes our epigenetic rules enable us to act rationally. But in other cases they do not. Our behaviour is too diverse and heterogeneous to be subsumed under the parsimonious general laws of rational choice theory (see also Wilson 1998). Seen in 18 this perspective, the fault the economists discussed above make is that they erroneously believe that economic theory’s basic assumption can be retained that in analysing behaviour affective or evaluative and cognitive aspects can be neatly distinguished. Group selection and methodological individualism Sober and Wilson argue that rehabilitating group selection also provides an effective antidote to the influential doctrine of methodological individualism (MI) in the social sciences (Sober and Wilson 1998, 133, 329). MI has many staunch advocates especially in economics. In this subsection I examine whether Sober and Wilson’s argument should convert these staunch advocates. MI is the doctrine that explanations of phenomena at the social or aggregate level should refer only to individuals, their properties, actions and interactions. Explanations in terms of properties, actions and interactions of supra-individual entities are discarded. Thus the doctrine rules out that, for example, (changes in) la volonté generale (the ‘general will’) or das gesundenes Volksempfinden are referred to in order to explain changes in the legal system of some country. If there exist things such as the general will at all, advocates of MI insist, then they cannot be but outcomes of actions and interactions of individuals. And if so, they also have to be ultimately explained in terms of the actions and interactions of individuals. What the example just given also indicates is that the refusal of MIsts to accept supra-individual entities as explanatory variables is to a large part extent based on the belief that supra-individual entities are ontologically suspect. Can something like the general will be said to exist at all? Is it not just a shorthand way of speaking about things that really do exist: the citizens of some country and their values, desires, beliefs and causal powers? Invoking groups as explanatory variables raised the same suspicion with proponents of MI. Can groups be as real as individuals? Or is the existence of groups only derivative (on the existence of individuals as the elementary units) and talk of groups only a convenient way of speaking without real ontological 18 underpinning? And if the latter is the case, then should we not refer to the 19 individuals that form groups rather than groups in our explanations? Do Sober and Wilson prove MI to be mistaken? Once again, I think, we should distinguish between the interaction and the shared fate notions of groups and group selection. Proponents of MI would find nothing to quarrel about in the interaction notion, I submit. The ideas that population structure matters and that there can be a combination going on of group competition without group interaction and selective individual interaction without group competition are fully compatible with MI. Proponents of MI would hold (and rightly so, I think) that explanations in terms of groups and group selection here are misleading. In fact, individuals, acting under different circumstances and with different opportunities, are the only active agents here. Things may be different with the shared fate notion of groups and group selection. Here the leading idea is that if we accept individual organisms as elementary units in our explanations (as proponents of MI do), then there is no good reason to reject 18 Perhaps work in ‘social ontology’ will shed more light on this. See also Sugden’s interesting work on team preferences and team reasoning (Sugden 1993 and 2000). 19 But we must be careful here, since many professed proponents of methodological individualism have seen no problem in accepting ‘supra-individual’ firms and households as explanatory individuals. 19 well-integrated groups of individuals having a shared fate as elementary units in our explanations. For groups liken individual organisms in that they also have to cope with the permanent danger of mutiny from within. If we are willing to accept wellintegrated individual organisms, that do reasonably well in coping with this internal problem, as legitimate explanatory units, then we should also accept well-integrated groups of individual organisms. What matters is how well entities (at any level of organisation) are integrated internally. On this criterion, groups acting as superorganisms qualify. Maybe so. But if the notion of superorganism is meant to convey that the bonds between group members are so tight and strong that they will never be broken, then such groups rarely exist if at all (Ridley 1996). Basing one’s theory on such exceptional groups would surely imply an unduly restrictive view on groups. The shared fate notion of a group seems to imply a sort of fatalism on the part of group members that many (not just proponents of MI) will find hard to accept when discussing human behaviour. The shared fate notion essentially says that if a group dies off then its individual members go ‘through the drain’ with it. It is as if individuals do not have any escape option. Or if they do, individuals are portrayed here as passive creatures patiently waiting what happens to their group all up to the point that they themselves pass away. They do not take any initiative to curb their own fate. This seems to strain credulity. It seems obvious, for example, that individuals can leave their group, join other groups, found new groups and so on. At least Sober and Wilson owe us an explanation why individual members of groups as superorganisms do not seize these opportunities. Why would people not try to leave a group, if they think it is a ‘sinking ship’, before it is too late? What prevent them from doing so? I do not think Sober and Wilson offer an explanation for this. But they do mention secondary behaviours and social norms as mechanisms, unique to human groups, which at least suggest possible explanations. Explanations based on these mechanisms would assume that group members are punished if they do not display (their part in) mutually beneficial primary behaviours or if they do not respect the group’s social norms. The most severe form of punishment, Sober and Wilson suggest, is ostracism or expulsion from the group. This seems to presuppose that group members perceive membership of the group as something they want to retain at any cost. But exactly this presupposition can be called into question, I think. If group members believe their whole group is about to be eliminated, they may well want to leave the group. This is not to argue that there are no possible ways in which individuals may be prevented to leave the group also in such critical situations. One may think here of groups having implemented severe punishments on desertion, for example. I do not want to argue either that plausible notions of groups and group processes cannot be found that seem to contradict MI. Sometimes it seems that groups, even if they are formed or founded by individuals, can get to live ‘a life of their own’. They may exhibit ‘emergent properties’ that are not fully reducible to the properties, actions and interactions of individuals. Groups may ‘survive’ replacements of one set of individuals by another one having quite different personal properties, for example. Such groups may also have a profound influence on essential properties of their members. My point here simply is that Sober and Wilson do not advance compelling 20 arguments, neither in their interaction nor in their shared fate notion of a group, to 20 reject MI. Co-evolution and economic processes Even if we were to conclude from the previous discussion that human groups are not entities sui generis, this does not imply that groups need not be taken into account in analyses of economic processes. Human groups can be ephemeral. Yet the ways in which ‘populations’ are structured in groups and in which groups interact with one another may have a profound impact on how economic processes unfold. Let us first return to Sober and Wilson’s discussion of the function of culturally transmitted norms in human groups. Sober and Wilson argue that each human group has developed its own set of norms. They believe that evolutionary game theory sheds an interesting light on how such evolutionary processes unfold. Once we have several groups with their own norms in the population, processes of group selection take off, favouring the most functionally adaptive ‘altruistic’ ones (Wilson and Sober 1998, 151-2). The overall picture emerging here is the following. Before group selection starts to operate there is individual selection going on, leading to groups all having their own 21 distinct set of norms. Only after each group has established its own norms, group selection starts to work on them. And once group selection takes off, the norms of the groups do not change anymore. Thus this picture essentially involves two things. First a temporal order is implied. Processes of individual selection are assumed to precede processes of group selection. And, second, a causal asymmetry between the two types of processes is implied. Group selection is causally dependent on individual selection in the sense that group selection works on the products of individual selection. In this sense there is scaffolding (or cumulative causation, if you like). Conversely, individual selection is in no way causally affected by group selection. Processes of group selection are assumed not to have a feedback on processes of individual selection (in this particular case: on the evolution of norms). The picture is neat. But is it also realistic? Work in social psychology on group dynamics suggests it is not (Tajfel 1981). What happens within groups, how groups evolve over time seems to depend both on intragroup and on intergroup processes. In particular, the social identity of group members seems to evolve as a consequence of both processes. The social identities of individual group members get their shape (and are reshaped) partly in response to exchanges and encounters with (menbers of) other groups. This point can be generalised by saying that there is co-evolution going on. Once there are groups, individual and group selection work simultaneously mutually affecting each other. In this complex combined process there are no ‘fixed points’. It cannot be assumed that individual selection yields products that are unaffected by the dynamics of intergroup interaction. Bringing the discussion closer to home (economics, that is) now, it seems that in analysing economic processes different levels (or layers) of organisation are to be reckoned with. What is more, what is going on at some level of organisation seems to depend both on what is going on at lower levels and at higher levels. Let me try to 20 It also remains to be seen whether individual selection is compatible with MI. See Vromen (2001a) for discussion. 21 As indicated earlier, I believe that applying evolutionary game theory in this way implies that it is assumed that individual selection is going on. 21 elaborate a bit on this. In economic theories of organisation it has become commonplace to argue that three levels of organisation should be taken into account: the individual, the organisation and the market (or industry). Thinking about groups and group selection now, can the firm be treated as a group? It seems that in some respects it can. It can be argued that firms display some unity and compete with one another in the market. But, as Campbell (1994) has argued convincingly, especially in larger firms it seems more plausible to locate groups in between individuals and the firm as a whole at the level of face-to-face cliques. Within a firm, individuals who regularly interact with another in non-anonymous ways tend to form a clique with 22 its own shared beliefs, codes of behaviour and rituals (Campbell 1994, 30). What beliefs, codes of behaviour and rituals evolve does not depend only on the types of individuals entering the clique (at ‘one level below’), but also on how the clique is situated, and interacts with other cliques within the same firm (at ‘one level above’). In other words, there is co-evolution going on. What is interesting in Campbell (1994) is not just the suggestion that in order to get a better understanding of firm processes a level of face-to-face groups has to be inserted in between the levels of the individual and of the (large scale) firm. Interesting is also the suggestion that face-to-face group selection tends to undermine rather than support the market performance of firms. Face-to-face groups tend to display a combination of ‘clique selfishness’ and outgroup hostility (this combination I called ‘groupism’ earlier on). The resulting antagonisms within the firm tend to undermine the firm’s functioning as a whole. This is a refreshing antidote, I think, to those organisational theorists who over-optimistically argue that selection processes working at different levels mutually reinforce each other in the direction of Paretooptimal outcomes. Such a perfectly harmonious picture is drawn, for example, by socalled agency cost theorists (such as Jensen, Meckling and Fama; see Vromen 1995 for further discussion). The point Campbell makes here is perfectly compatible with one of the main things Sober and Wilson argue for: selection processes working at different levels of organisation tend to work in opposite directions. Thus conceived, Sober and Wilson’s MST can serve as some sort of reminder in organisational studies: don’t assume too easily that selection processes operating at different levels of organisation work in the same direction. 6. Concluding remarks The harvest of this paper may seem poor. Its results are programmatic and promissory rather than substantial and ready-made for ‘consumption’ in economics. A lot work is still to be done both within EP and MST and in efforts to flesh out their implications for economics. Those insisting on immediate gratification in interdisciplinary learning are likely to be disappointed. On the other hand it can be argued, I think, that the prospects are promising. MST primarily holds out the hope, it was argued, of coming to grips with a vexing issue that has long intrigued many economists: how to comprehend co-evolution at different levels of organisation? And the contribution of EP to economics may well lie, I argued, in its ability to tell what 22 Experiments in social psychology suggest that groups that are formed arbitrarily ex nihilo immediately start developing their own codes of behaviour and rituals. And among its members immediately ingroup feelings arise spontaneously. 22 23 behavioural patterns are enticed under different conditions and stimuli. Once we have a better understanding of this, we may also know better not only what sorts of behaviour are inevitable, but also how to avoid behaviour that we want to avoid and how to stimulate behaviour that we want to promote. Thus EP could also prove to be of interest for normative and policy issues. The latter also indicates how far contemporary evolutionary theory already has distanced itself from the idea that we, human beings, are puppets of our genes and memes. No doubt many will find that culture and cultural evolution are given short thrift in EP and MST. And many will also hold that it is better to turn to other social sciences and theories about cultural evolution and start there. But I believe we better start with improving insights into our biological heritage and adaptations that enabled the human species to get its specific types of culture and its specific types of cultural evolution off the ground (in phylogenetic time). I think it makes sense to start there and see how far we can get with it. Probably the picture will turn out to be more complex than we can now envisage. We may still have a long way to go, but I think it is worthwhile to continue travelling it. 23 EP may well be able to explain why we tend to behave self-regardingly in some situations while in other situations we tend to behave other-regardingly. EP also may be able to shed some light on otherwise puzzling findings in experimental economics (Hoffman, McKabe and Smith 1998). 23 REFERENCES Axelrod, R. (1984), The Evolution of Co-operation, London: Penguin Books. Baron-Cohen, Simon (1997) (ed.) The Maladapted Mind: Classic Readings in Evolutionary Psychopathology, Psychology Press. Becker, Gary S. (1976) Altruism, egoism and genetic fitness: Economics and sociobiology, Journal of Economic Literature 14, 817-26. Ben-Ner, A. and L. Putterman, (1998), “Values and institutions in economic analysis”, in A. Ben-Ner and L. Putterman, Economics, Values and Organization, Cambridge University Press, p.p. 3 - 69. --- (2000), “On some implications of evolutionary psychology for the study of preferences and institutions”, Journal of Economic Behavior & Organization, Vol. 43, p.p. 91 - 99. Binmore, Ken (1994) Game Theory and the Social Contract. Volume I: Playing Fair, Mass.: MIT Press. --- (1998), Book review of Sober and Wilson (1998), Managerial and Decision Economics 19, 53940. Bolton, Gary E. (2000) Motivation and the games people play, in Leonard D. Katz (ed.), Evolutionary Origins of Morality, Imprint Academics. Bolton, G. and A. Ockenfels (2000), “ERC: A Theory of Equity, Reciprocity, and Competition”, The American Economic Review, vol. 90, No. 1, p.p. 166 - 193. Boyd, Richard and Peter Richerson (1985), Culture and the Evolutionary Process, Chicago: University of Chicago Press. Buller, David J. (2000), Evolutionary Psychology (http:/www.uniroma3.it/kant/field/ep.htm). (in preparation) Adapting Minds: Evolutionary Psychology and the Persistent Quest for Human Nature, MIT Press: Bradford Books. Buss, D.M. (1995), Evolutionary psychology: A new paradigm for psychological science, Psychological Inquiry, 6, 1-30. --- (1999) Evolutionary Psychology: The New Science of the Mind, Boston: Allyn & Bacon. Campbell, Donald T. (1994), How individual and face-to-face-group selection undermine firm selection in organizational evolution, in Joel A.C. Baum and Jitendra V. Singh (eds) Evolutionary Dynamics of Organizations, New York: Oxford University Press, 23-38. Cosmides, L. (1989), The logic of social exchange: has natural selection shaped how humans reason?, Cognition 31,187-276. Cosmides, L. and J. Tooby (1987), From evolution to behavior: evolutionary psychology as the missing link, in J. Dupré (ed.), The Latest on the Best, Cambridge: MIT Press, 277-306. --- (1992), Cognitive adaptations for social exchange, in J.H. Barkow, L. Cosmides and J. Tooby (eds), The Adapted Mind, Oxford: OUP, 163-228. --- (1994), Better than rational: evolutionary psychology and the invisible hand, AER 84, 327332. --- (1996), Are humans good intuitive statisticians after all? Rethinking some conclusions from the literature on judgment under uncertainty, Cognition 58, 1-73. ---(1997), Evolutionary Psychology: A Primer, http://www.psych.ucsb.edu/research/cep/primer.htm. 24 Cummins, Denise Dellarosa and Robert Cummins (1999), Biological preparedness and evolutionary explanation, Cognition 73, B37-B53. Durham, William H. (1991), Coevolution: Genes, Culture, and Human Diversity, Stanford, Ca.: Stanford University Press. Evans, Martin G. and Young Chul Chang (1998), Cheater detection and altruistic behavior: an experimental and methodological exploration, Managerial and Decision Economics 19, 467-480. Fehr, Ernst and Klaus Schmidt (1999) A theory of fairness, competition, and cooperation, Quarterly Journal of Economics 114, 817-68. Fodor, J. (1983), The Modularity of Mind, Cambridge: MIT Press. Frank, R.H. (1988), Passions within Reason, New York: W.W. Norton & Company. ---- (1999), Luxury Fever: Why Money Fails to Satisfy in an Era of Excess, New York: The Free Press. Gazzaniga, Michael (1997), What are brains for? in Robert L. Solso (ed.), Mind and Brain Sciences in the 21st Century, Cambridge, Mass.: MIT Press, 157-171. Gilbert, Margaret (1994), Me, you, and us: distinguishing “egoism”, “altruism” and “groupism” (Open Peer Review of Wilson and Sober 1994), Behavioral and Brain Sciences 17:4, 620-1. Gintis, Herbert (2000) Game Theory Evolving, Princeton: Princeton University Press. Güth, W. and M. E. Yaari (1991), “Explaining Reciprocal Behavior in Simple Strategic Games: An Evolutionary Approach” in U. Witt, Explaining Process and Change: approaches to evolutionary economics, Ann Arbor: University of Michigan, p.p. 23 - 34. Huck, S. (1997) Institutions and preferences: An evolutionary perspective, JITE 153 (4), 771-9. Hoffman, E., K.A. McCabe and V.L. Smith (1998), Behavioral foundations of reciprocity: experimental economics and evolutionary psychology, Economic Inquiry 36, 335-52. Lloyd, E. (1999), Evolutionary psychology: the burdens of proof, Biology and Philosophy 14, 211-33. Lumsden, Charles J. and Edward O. Wilson (1981), Genes, Mind, and Culture: The Coevolutionary Process, Cambridge, Mass.: Harvard University Press. Mallon, Ronald and Stephen Stich (2000), The odd couple: the compatibility of social construction and evolutionary psychology (http://ruccs.rutgers.edu/Archive Folder/…20Group/Publications/Odd/OddCouple.html) Marr, D. (1982), Vision: A computational investigation into the human representation and processing of visual information, San Francisco: Freeman. Maynard Smith, John (1998), The origin of altruism (Book review of Sober and Wilson 1998), Nature 393, 639-40. Mayr, E. (1961), Cause and effect in biology, Science 134, 1501-6. Markóczy, Lívia and Jeff Goldberg (1998), Management, organization and human nature: an introduction, Managerial and Decision Economics 19, 387-409. 25 Nicholson, N. (1997), Evolutionary psychology: toward a new view of human nature and organizational society, Human Relations, 1053-1078. Ridley, M. (1996), The Origins of Virtue, London: Penguin Press. Robbins, Lord Lionel (1932) An Essay on the Nature and Significance of Economic Science, London: MacMillan. Samuels, R. (1998), Evolutionary psychology and the massive modularity hypothesis, British Journal for the Philosophy of Science 49, 575-602. Sober, E. (1987), What is adaptationism? In J. Dupré (ed.), The Latest on the Best, Cambridge: MIT Press, 105-118. --- (1998), Evolutionary psychology – a manifesto (Introductory remarks to a symposium on evolutionary psychology in Madison, Wisconsin). Sober, E. and Wilson, D.S. (1998), Unto Others: The Evolution and Psychology of Unselfish Behavior, Cambridge, Mass: Harvard University Press. --- (2000) Summary of ‘Unto others: The evolution and psychology of unselfish behavior, Journal of Consciousness Studies 7, 185-206. Sugden, Robert (1993), Thinking as a team: towards an explanation of nonselfish behavior, Social Philosophy & Policy 10, 69-89. --- (2000), Team preferences, Economics and Philosophy 16, 175-204. Symons, D. (1979), The Evolution of Human Sexuality, New York and Oxford: OUP. --- (1992), On the use and misuse of Darwinism in the study of human behavior, in Barkow, Cosmides and Tooby (eds), 137-159. Tajfel, Henri (1981), Human Groups & Social Categories, Cambridge: CUP. Tomasello, Michael (1999), The Cultural Origins of Human Cognition, Cambridge, Mass.: Harvard University Press. Tooby, John and Leda Cosmides (1995), Mapping the Evolved Functional Organization of the Mind and Brain, in Michael S. Gazzaniga (ed.), The Cognitive Neurosciences, Cambridge, Mass.: MIT Press, 1185-1197. Vromen, Jack J. (1995), Economic Evolution: An Enquiry into the Foundations of New Institutional Economics, London: Routledge. ---- (1999), 'Evolutionary psychology and economic theory' (paper presented at annual ISNIE Conference, Washington D.C., 17-19 September 1999). --- (2001a) Ontological Commitments of Evolutionary Economics, in U. Mäki (ed.), The Economic World View, Cambridge: Cambridge University Press. --- (2001b) The Human Agent in Evolutionary Economics, in John Laurent and John Nightingale (eds), Darwinism and Evolutionary Economics, Cheltenham: Edward Elgar. Waal, Frans de (1996), Good Natured, Cambridge, Mass.: Harvard University Press. Williams, George C. (1966), Adaptation and Natural Selection, Princeton: Princeton University Press. ---- (1971) (ed.) Group Selection, Chicago: Aldine Atherton. Wilson, David Sloan and Elliot Sober (1994), Reintroducing group selection to the human behavioral sciences, Behavioral and Brain Sciences 17:4, 585-654. Wilson, Edward O. (1998), Consilience: The Unity of Knowledge, New York: Alfred A. Knopf. 26 Zywicki, T. (2000), “Evolutionary Biology and the Social Sciences”, Human studies review, vol. 13, number 1, fall 2000.
© Copyright 2026 Paperzz