Friday, 7 August 2015

Powers and Clarke: Insititutions and the development of human sociality

Dr Simon Powers is a member of the Lehmann group in the Department of Ecology and Evolution at the University of Lausanne, although he originally got his PhD in computer science.

Powers is concerned to explain how humans have moved from small-scale, self-sufficient tribes or kin-groups, to large-scale, differentiated exchange economies, a change which has elsewhere been called the 'Holocene transition'.
Powers calls this change a Major Transition, analagous to that of the transition to multicellularity. In doing so he utilises Maynard Smith and Szathmary's 1995 criterion of the emergence of a novel system of inheritance. In the human example this is language. Powers claims, in addition that post-Holocene humans qualify as having transitioned because they have lost their autonomy: they became dependent on one another for their survival and reproduction.

Powers claims that the explanatory task is not one of explaining why humans cooperate. He says this bit, the free rider problem, is easy. Humans cooperate because it pays.....generates direct benefits, thanks to reciprocity. Instead, Powers claims that what we need to do is explain how humans succeeded in converging on one of the many cooperative equilibriums available. To achieve this, Powers calls upon the game theoretic notion of 'institution'. An institution is a nested game form, where individuals communicate strategically to converge on the specification of a game which they then agree to play. The specification sets out the connection between strategies and payoffs, by setting up a system of payments and punishments. As an example, Powers and Lehmann 2013 constructed a model in which they demonstrated that investment in a public good can coevolve with  the carrying capacity of their patch, in a process they call 'social niche construction'. They describe an institution that fixes the level at which participants contribute to a public good, an irrigation system, the proportion of resources that are spent on policing the public good, and the proportion actually invested in it. Once these rules have been agreed, the participants play the game and decide whether to cooperate or to defect. This gives rise to a two-stage structure. In the first stage, the individuals form institution and decide how much of their common resource is used for policing. The remainder is invested. Then in the second stage, the individuals play a public goods game and decide whether to cheat or not. Sanctions are then applied to cheaters. Powers draws on examples from Elinor Olstrom to argue that many human institutions have been successful in stabilising human cooperation. Crucially, institutions transform social interactions into games in which it pays to cooperate.

Finally, Powers argues that another important explanatory task consists in detailing how we evolved the cognitive complexity necessary to support institutions. The cognitive prerequisites for institutions are many…..including language, causal understanding, imagination, theory of mind and prosocial motivations. Powers favours Sarah Hrdy's model of cooperative breeding, saying it led to selection for cognitive complexity which enabled language and institutions, which then enabled the emergence of agriculture.

I was the commentator on Dr Powers' talk. I argued that little is gained by calling the Holocene shift a major transition. I prefer a compositional evolutionary definition, according to which a transition occurs when parts gain a common evolutionary fate – so that there is no opportunity for differential selection between them. This criterion doesn't hold for the Holocene transition. Some humans did and still do well (evolutionarily speaking) at the expense of others. I'm also not confident that language can be profitably thought of as an inheritance system. As for's rather too easy, I think, to argue that autonomy is compromised. Every living thing is dependent on something else. Perhaps, as Stephen C Stearns suggested in 1992, humans are partway transitioned, but we are surely not mere parts of a single societal organism. We still have individual evolutionary fates. Some of us can do well at the expense of others, in contrast to something like a worker ant.

Next I queried Powers’ claim that institutions solve free rider problems. As long as information is imperfect, which it always is, games always admit subgames. All players can choose whether to play by the rules or only pretend to play by the rules, or to try to change the rules, or to attach the player attempting to administer punishment, or to form a coalition with another player in pretending that player punished them. The Powers & Lehmann model assumes that individuals abide by the rules, once they have agreed them. But another possibility is that players merely pretend to play by the rules. For example, a player might spot another cheating, and decide to avoid the costs of punishing him by simply pretending to punish him. Powers says they solve this ‘second-order free-rider problem’ by having the players elect officials who are paid to administer sanctions. Then the other players avoid the costs of punishment and policing altogether, and the official doesn’t mind because for him its now a profitable activity. I was immediately reminded in this of the private companies paid by English councils to clamp illegally parked cars. The election of official punishers doesn’t avoid a second-order problem at all, because the officials are then tempted to impose illegitimate sanctions in order to maximise their returns. They cheat the very system they are elected to police.

We have, sadly, many examples of failed institutions… such as those designed to solve cooperation problems with respect to climate change, fishing, and international peace. Nonetheless, occasionally they work,  and the question we need to answer is why? Institutions on their own are just sets of rules – questions remain about how they become authoritative, compelling. We agree that everyone pays tax to fund the NHS, a public good. But some people cheat and don’t pay tax – why don’t more people do that?  I suspect that we will need to invoke something like norms here. Norms might be effective in getting people to view the game a certain way ie as a public goods game rather than them against the tax inspector, Alternatively, perhaps cooperative endeavours succeed despite free riding, simply because the benefits achieved are so huge. But then the explanatory problem is all concerned with the how the benefit is generated, and free riders are a red herring.

Lastly, I sought to undermine Powers’ reliance on underlying cognitive complexity. According to his account, institutions function to solve coordination problems, such as how much of the public purse ought to be spent on policing a public good.  Institutions don’t help us in explaining the public goods themselves…how to produce them, or why they’re a good idea. Obviously, there is no point setting up an institution to decide how to pay for an irrigation channel unless you already know how to build an irrigation channel.  And of course it is not a criticism of Power’s model that he doesn’t tell us how to build an irrigation channel, but some consideration of the complexities of the task might lead us to doubt Power’s story about the cognitive complexity necessary for setting up institutions. I don’t know anything about irrigation channels, but its strikes me as plausible that their design is something that will happen gradually, cumulatively, rather than being invented overnight by some smart guy. First someone digs a bit by accident, and notices it helps his crops. Later on the activity comes to include more people and gets more complex. How does one cheat an irrigation channel?  Why are group channels better than individual channels? I worry that some of the details I am ignorant of are important. Once you’ve figured out all this stuff, then agreeing on how to pay for one seems like the easy part. And if the hard part can be done without anyone being a genius and thinking it all through, then why can’t the easy part too?

In fact, Powers allowed that institutions could emerge out of a blind cultural evolution process, without anyone having to have intelligent foresight. But then why insist that cognitive complexity had to come into it at all? Why not imagine institutions as emerging gradually and brainlessly out of a trial and error process, in which effective rules were eventually arrived at by accident? My sense is that institutions flourish when they enable individuals to dispense with understanding the whole system. They operate by the summed actions of individuals who each know their own small part well, rather than by any top-down intelligence.

This worry doesn’t undermine Power’s whole story, it merely breaks its dependence on a particular version of historical events. We can keep the vital role of institutions in in modifying payoffs so that cooperation is incentivised, without pinning it on innate cognitive complexity. But we can drop the bet that brains got bigger before agriculture got started: presumably we will have an empirical verdict on that bet eventually anyway.  Of course significant intelligence is still involved. Presumably someone has to have some rough design for the layout of an irrigation channel in his head. But he might have that design in his mind because he watched his father dig one before, rather than because he sat down and thought it through. We can lean on social learning and niche construction, or biased transmission in cultural evolution, to explain the emergence of complex cooperative achievements – it doesn’t need to be about selection of complex innate cognitive capacities.

The audio recording of Simon's talk is here. My response starts 46 mins in.

Next post: Birch and Bentley -Time and relatedness in microbes and humans.

No comments: