While it may be true that the gene is selfish, the organism, a collection of genes, is a different matter. One of the most amazing qualities of the natural world is the ability for cooperation, altruism and selflessness to arise out of a swirling chaos of self-interest. And so, dazzled by this beautiful display, we have set ourselves to work trying to find a an explanation for it all, a kind of theory of altruism that can make sense of the remarkable works of nature. Yet we seem to have been left with a set of theories, each of which seems sound in their own right, yet none of which seem to provide a complete picture on their own. Here are the core scientific theories as things currently stand:
Group-selection – When various groups compete for scarce resources, those that cooperate more effectively are more likely to survive. If the group members have genetics that tend towards altruism or selflessness and reduced internal conflict, their group will thrive, and they will be more likely to survive and pass on their genes. Thus group-selection states that their is a selective evolutionary force for altruistic genes (Wynne-Edwards 1962, 1986). Of course, altruism implies a behaviour that reduces or is at least suboptimal for the fitness of the individual. So, within the group altruistic individuals suffer an evolutionary penalty and logically it seems they would slowly be replaced by free-riders and cheaters, who do not ‘hold up their end of the bargain’ by behaving altruistically, but who reap the benefits of other’s altruism. So, group-selection is offset by an opposite, destructive ‘vector’ of selfishness. This second vector has often been proposed as a reason why group-selection could not explain altruistic behaviour. Consequently group-selection, after dominating the topic for many years, fell from favour in the 1960s (Maynard-Smith 1964) (Williams 1972).
Group selection, despite institutional resistance, has recently seen a new surge in popularity within the field, however (Wilson, Wilson 2007). This has sometimes occured under a new label of ‘multi-level selection’, supported by observations that both multi-cellular organisms and complex cells themselves seem to be exercises in cooperation that can be neatly explaining in terms of group-selection. Another variation to group-selection is the theory of group-augmentation (Kokko, Johnstone, Clutton-Brock 2001). This theory states that by acting to promote the welfare of other members of a group, an organism can increase the size of their group, which in-turn makes the group more successful, providing a fitness benefit to the orginal organism. In this way it avoids references to evolution at a group level, by focusing soley on the return-on-investment for the individual. Here we will treat multi-level selection and group-augmentation broadly under the same label as group selection.
Kin-selection – The fall of group-selection saw the rise in prominence of a ‘rival’ explanation – kin-selection. This theory points out that an organism trying to promote its genes not only has the option of reproducing, but may also further their genetics by promoting the fitness of relatives who likewise share much of the same genetics. If their brothers and sisters survive, for example, they carry on many of the orgnanisms genes even if that organism is unable to reproduce. Thus, genetic tendancies to behaviours, such as kin-altruism, are selected for within the gene pool (Hamilton 1963, 1964) (Maynard-Smith 1964).
Reciprocity – Other theories sought to point out that apparently altruistic acts need not be attempts to actually promote the fitness of the group or other organisms. Instead, organisms with certain altruistic behaviours could actually result in direct fitness benefits within the organism’s lifetime, or even in the short-term. Reciprocity was proposed as one of the central ways in which this would occur (Trivers 1971) (Nowak & Sigmund 1998). This theory suggests an organism adopts a set of behaviours centered around the principle of reciprocity, where kindness, cooperation and altruism is returned to organisms that appear to be behaving in the same way. While sometimes this altruism might be exploited and not returned, the benefits would far outweigh the risks whenever two or more reciprocists got together and could enjoy the many fitness benefits of their reciprocal mutual promotion. Thus reciprocal altruism is selected for within the gene pool. Game theory has often been used in arguing for reciprocity, supported by empirical studies that appear to show reciprocity in natural populations of various species.
The social sciences and humanities are eager, sometimes rightly so, to emphasise the significance of social factors on moral decision making. Let us acknowledge their importance here as a complex set of forces interacting with the biology of the organism.
Social factors – In social organisms, and organisms with a brain designed for extended or life-long learning, the environment in which the organism is embedded becomes very important in activating, shaping, repressing, stimulating or even forging anew behaviours, including an organism’s patterns of altruism of cooperation. For humans, being evolved to adapt ‘on-the-fly’, this tendancy is probably stronger than in any other known organism. It implies the importance of factors such as upbringing, family, culture, education, social location, socio-economics and systems of moral reasoning, to name just a few.
Proponents of each of these explanations are keen to emphasise that their particular explanation is central to understanding altruism, and therefore moral behaviour in humans. Yet the arguments proposed are usually questions of emphasis, rather than hard critiques of the validity of other theories. A sensible view, one which is proposed here, is that all these factors act in parallel, particuarly in the case of humans. Each force varies in relevance according to each individual case, explaining the great variability of human morality. In order to understand humanity and morality thoroughly, we should not restrict our thoughts to one single mechanism, but rather envisage a multi-theory model that allows for multiple mechanisms to come into play simultaneously.
It is also well-known that certain factors undermine altruism.
Free-riding – Where an animal can obtain group membership without engaging in altruistic behaviours, it can receive the fitness benefits from other’s altruism, without paying the fitness costs of the altruistic behaviour.
Betrayal – Where altruistic behaviours involve a reduction in compeititive defences and a friendly orientation is adopted amongst group members, there may be increased opportunity for fitness gain by directly harming other group members who are less able to respond than non-group members. For example, a bird that secretly eats the eggs of other members of the colony is engaged in betrayal. We might also call this direct-harm strategy “in-group predation”.
Consider a model that explains simple aggregate altruism (ignoring situational variance) as a one dimensional function of the above factors. This is not intended as a mathematical model, but rather a conceptual tool by which explains altruism at a brief glance. To begin with, we know altruism is influenced by both nature and nuture, biology and sociology:
Altruistic behaviour =
Cultural, social and personal factors
Let us expand on the evolutionary factors:
Evolutionary factors =
Kin-selection (g) +
Group-selection and Group-augmentation (g) –
( Free-riding (g) + Betrayal (g) – Dynamic Cooperation (g) )
Naturally many other variables influence each of these factors, however we will present the argument that group size is the most profound of these variables and worth including even in a very generalised model such as this one. Here we indicate that a factor is a function of group size here using “g”. We will seek to briefly explain why. We will also seek to introduce the concept of Dynamic Cooperation which serves to reduce free-riding and betrayal.
Naturally social factors are also significant in the overall picture, though we won’t attempt to explore them in this paper:
Cultural, social and personal factors
Altruism as a function of group size
As part of this model we consider group size. This is because group size appears to have a profoundly impact in most theories of altruism.
Let us consider the example kin-selection, which is the most marked example of how group-size changes altruism. In a group of one, there is no opportunities for in-group altruism. The behaviours of the organism are a result of their non-group environment. In a group of two, there are a series of opportunities to altruistically assist the other group members. For example, it might be possible to spend additional time searching for food in order to feed an injured or sick group member. Group-selection and reciprocity aside for the moment, we can the organism’s genetics receives a fitness gain for this act based on the relatedness of the other organism. As the group size increases, so does the opportunity for the organism to advance it’s own through kin-selection. That is, there are more acts of altruism available amongst the subset of choices it is faced with. Whereas perhaps time spent practice fighting was previously the best use of free-time, it might be that now, in a larger group, spending all available spare-time providing altruistic assistance to others is quite a valid strategy.
The overall tendancy, therefore, is that as group size increases, so does the evolutionary incentive for altruistic behaviours and skills (in this case primarily the percentage of time engaged in altruistic acts). A very large group implies a strong kin-selection vector based upon a great variety of altruistic opportunities. Of course, the relationship between group-size and kin-selection forces is not entirely linear, because each new group member is likely to be somewhat less related. However, the greater the group size, the greater the altruistic vector overall. Many species find their group-size is limited by the niche they occupy in the ecosystem and landscape, but for more social creatures, especially for a species like humanity, kin-selection becomes a strong force even discounting close family ties, due to the sheer number of altruistic opportunities available.
We can imagine a similar scenario in the case of reciprocity. That is, as a group becomes larger, there is increasing amounts of opportunties to benefit from reciprocation of altruistic acts. In the case of reciprocation there is probably also a limiting factor requiring intelligence, because the recipient is required to understand the altruistic act and remember who committed it, in order to return the favour at a later time. In a large population this is feasilbly quite a challenge. Humans society is therefore likely to contain a considerable reciprocity vector, whereas eusocial insects may have a reduced potential due to the difficulties fitting sophisticated social memories into a relatively simple neural network.
The difficulties of simple altruism
Critics of all kinds of theories of altruism like to point out, quite rightly, that as the levels of altruism increase, so does free-riding and betrayal. High levels of altruism provide a very strong selective force towards these darker tendancies, because altruistic group members offer easy fitness advantages to the cheaters in comparison to a “honest day’s work”. This is often described by saying that free-riding and betrayal are opposite vectors to altruism, punishing altruism and pushing in the opposite direction to kin selection, group selection and reciprocity. However, in addition to mirroring altruism, these forces are also magnified by group size further. A larger population offers more opportunties for cheating, and victims and witnesses are less likely to have fore-knowledge of the threat in a large group in comparison to a small one. So free-riding and betrayal is also a function of group size, with solitary creatures have none of this darker evolutionary force, a highly social animals having a lot.
Let us suppose that given there are multiple competing vectors at work here, there is some sort of equilibrium achieved in most species that occupy a stable niche in the environment. Individual organisms will adopt a variety of selfish or altruistic paths, but the species overall can be considered to have a basic level of altruism described by our above model. Given that the cheating behaviours are likely to ramp up so drastically with group-size, some other factor seems necessary to describe the significant levels of altruism we see amongst humanity. Here we wish to propose Dynamic Cooperation as a solution to this difficulty that helps to complete our model.
Dynamic Cooperation is the set of behaviours that mitigate or reduce free-riding and betrayal on a social scale. So, while individual victims of betrayal might be prepared to provide a deterrent or defend themselves, predatory group members are usually well prepared for problems and will specifically act at a moment and place where individual retaliation is difficult. However, where group members exihibit systematic behaviours that prevent cheating they are far more effective and can receive a far greater fitness return in exchange for relatively risk-free and low-cost efforts. We propose three components to Dynamic Cooperation:
1. Detection of cheating – The group will adopt behaviours that are likely to systematically detect free riding and betrayal. For example a group might engage careful peer observation and discussion of behaviours, in order to predict free-riding. Or activities that are vulnerable to betrayal might be carried out in view of the entire group, so that betrayal cannot occur unnoticed.
2. Contingent cooperation – Rather than naively extending cooperation or altruistic acts to all parties, contingent cooperation involves the group coordinating to direct the benefits of altruism towards parties that have not been observed free-riding or betraying. For example they may exclude cheaters from a system that provides food to ill patients. While negative reciprocity only occurs between individuals, contingent cooperation implies a systematic exclusion and is engaged in by uninvolved group members. In the most extreme cases group membership could be revoked entirely. In human society, criminal background checks for employment could be considered contingent cooperation, particularly in cases where it goes beyond what might rationally be a problem for that particular role. Note that contigent cooperation is different from reciprocation – a return on investment for the individual organism is not required.
3. Active deterrence – In addition to more passive harm minimisation, organisms coordinate to actively target free-riding or betrayal and provide deterrence. This involves punishing free-riding and betrayal by causing direct harm to the non-cooperative organism. The criminal justice system is the most obvious example of this in human society.
Of course, for such behaviours to become coordinated actions of the group, individual members require a significant level of cognitive ability. Therefore Dynamic Cooperation is likely to be more prominent in species with certain levels on intelligence. Humans certainly qualify as a good example in this regard. Highly sophisticated examples of Dynamic Cooperation and cheaters’ attempts to avoid it are littered throughout history. There is also physical evidence that human motives display characteristics predicted by active deterrence – a strong body of evidence in psychology that shows that endorphins are released when people punish perceived wrong-doing in others, even when they are not the victim. There is also a strong indication that actve deterrence has a significant genetic component (Wallace, Cesarini, Lichtenstein & Johannesson. 2007). It also appears that free-riders and betrayers have tendancies to counter-punish active deterrence should the social opportunity arise, no doubt as part of an ongoing genetic rivalry between the two genetic tendancies (Herrmann, B., Thoni, C., Gachter, S. 2008).
For most who have considered the evolution of altruism, these Dynamic Cooperation behaviours are not neccessarily a new phenomenon, but instead they simply codify well known behaviours into a single theory that fits nicely into the proposed model of altruism. It’s a useful concept because reciprocation, group selection and kin selection all benefit from it, and so it is better codified as a separate concept rather than being conflated into one of these three. Dynamic Cooperation is also useful when we consider the altruistic equilibrium in species with very large group sizes, of which humanity is a perfect example. In such cases the vectors in both directions (altruism and selfishness) are larger than in other species, and are a more significant component of the complete set of forces acting upon the evolution of the species. The crux of the matter is this – high levels of altruism simply aren’t possible without dynamic cooperation.
Humanity continues to evolve, as does it altruism. Rather than attempt to explain a complex process with a single pet theory, we are better served when we have a pluralistic framework that considers a broad range of mechanisms and then allows space for empirical work to identify the importance of each mechanism in specific scenarios. The model proposed here is one that reminds of many aspects of one of the most profoundly important human traits, and gives us an overview that can inform not only evolutionary biology, but also our thinking in social science, economics, politics and our own moral existence. Our decisions matter, because most people are likely to have both darker and lighter lineages in our inheritence, and genetics are ultimately expressed in the choices we make and the lives we live.
Wynne-Edwards, V.C. (1962). Animal Dispersion in Relation to Social Behaviour. Edinburgh: Oliver & Boyd.
Wynne-Edwards, V. C. (1986) Evolution Through Group Selection, Blackwell. ISBN 0-632-01541-1
Group selection critique:
Williams, G.C. (1972) Adaptation and Natural Selection: A Critique of Some Current Evolutionary Thought. Princeton University Press.ISBN 0-691-02357-3
Maynard Smith, J. (1964). “Group selection and kin selection”. Nature 201 (4924): 1145–1147. doi:10.1038/2011145a0.
Hamilton, W. D. (1963). “The evolution of altruistic behavior”. American Naturalist 97 (896): 354–356. doi:10.1086/497114.
Hamilton, W. D. (1964). “The Genetical Evolution of Social Behavior”. Journal of Theoretical Biology 7 (1): 1–16. doi:10.1016/0022-5193(64)90038-4. PMID 5875341.
Smith, J. M. (1964). “Group Selection and Kin Selection”. Nature 201 (4924): 1145–1147. doi:10.1038/2011145a0.
Kokko, H.; Johnstone, R. A.; Clutton-Brock, T. H. (2001). “The evolution of cooperative breeding through group augmentation”. Proceedings of the Royal Society B: Biological Sciences 268 (1463): 187–96. doi:10.1098/rspb.2000.1349. PMC 1088590. PMID 11209890. edit
- Trivers, Q. Rev. Biol. 46, 35 (1971). (The Quarterly Review of Biology) (direct reciprocity)
- S. Nowak, K. Sigmund, Nature 393, 573 (1998). (reputational reciprocity)
Wallace, Cesarini, Lichtenstein & Johannesson. 2007. Heritability of ultimatum game responder behaviour. PNAS doi/10.1073/pnas.0706642104 (genetic pro social punishment)
Herrmann, B., Thoni, C., Gachter, S. (2008). Antisocial Punishment Across Societies. Science, 319(5868), 1362-1367. DOI: 10.1126/science.1153808 (anti social punishment – attacking good)