“Mind the Gap: Social Affairs between the Micro and Macro”, a talk given at NCH on 22nd November 2016 by Dr. Sebastian Ille, Lecturer in Economics

Those present were: Dr. Sebastian Ille, Dr. Marianna Koli (Faculty of Economics), Dr. Dave Rampton (Faculty of Politics & International Relations), Dr. Naomi Goulder (Faculty of Philosophy), Dr. Mike Peacey (Faculty of Economics), Dr. Lars Kjaer (Faculty of History), Dr. Peter Maber (Faculty of English), Dr. Brian Ball (Faculty of Philosophy) and Dr. David Mitchell (Faculty of Philosophy).


With the help of on-screen simulations more sophisticated than anything previously seen at the Ottoline Club, Sebastian gave a vivid introduction to some recent developments in economists’ methods. We watched unfold the possible consequences of decentralised decisions of many kinds: whether to go to a certain bar in the evening, whether to join a street protest, whether to move to a less ethnically heterogeneous neighbourhood, whom to trade with, which social networks to join. Stag-hunts also came in for consideration, and sectarianism in taxi-drivers, and the triggering of slides by the accumulation of heaps of sand.

Three new methods described by Sebastian promise, he argued, to allow much better prediction of the aggregate outcomes over time of the behaviour of interacting individuals. This has been recognised since the 1980s as an area in which standard microeconomic and macroeconomic methods often fail. One major reason they fail is that they do not register certain important feedback effects between the individual level and the aggregate or institutional level: Sebastian’s discussion focused particularly on individuals’ ‘social learning’, their adjusting their strategies of behaviour in the light of observation of how well or ill other individuals’ strategies are going. Threshold effects and the influence of topology are also among the factors that can now be addressed.

The first of these promising new tools is evolutionary game theory, which differs from standard game theory in studying large numbers of players with limited cognitive abilities. And rather than simply identifying possible equilibria of strategies, it investigates how such equilibria emerge and how resilient they are. The latter aspects, as Sebastian noted in response to a remark of Mike’s, are of great potential interest in relation to development policy, where the question often arises whether a certain intervention will need continual supplementation over time.

This version of game theory can be conjoined with a second new development, network theory, to deliver predictions regarding ‘spatial games’, for instance, which are much more accurate than those of standard approaches, particularly in not underestimating people’s cooperativeness when interacting with peers or in close-knit societies. A third method, agent-based modelling, which involves the programming of interacting individuals who differ in style and learning dispositions, is unlike the other two in depending essentially on simulation. This offers gains in complexity and precision but a certain loss of tractability as regards figuring out by what factors its outcomes are to be explained.

The ensuing discussion was more than usually a matter of different colleagues probing from different angles for a stronger grasp on these novelties. Naomi was interested in whether agent-based modelling can programme unconscious or at any rate unreasoned motives in its agents, or again whether it can express loyalty or commitment to institutions. Sebastian confirmed that it could. In response to a question of Dave’s about ‘practices’, he was more agnostic: Dave mentioned Bourdieu’s appealing to practices so as to avoid relying too much in social explanation on agents on the one hand and structures on the other. In one way, said Sebastian, strategies or rules of thumb could be counted as practices; in another way, practices are perhaps a subset of ‘institutions’ in the economist’s broad sense.

Mike usefully classified the new methods as readily attributing ‘backward-looking’ reasons to agents, in contrast to the exclusively forward-looking conception of rationality to which standard models adhere. He himself felt that evolutionary game theory in particular may be better at explaining the behaviour of animals than of human beings, whereas Sebastian was keen that we acknowledge how often we quite lack the information about options and payoffs that is crucial for forward-looking calculation. A programme that models behaviour in repeated prisoner’s dilemma situations was run several times with different initial conditions: a certain cooperative pattern turned out to be interestingly robust, at any rate given the specific parameters imposed. Lars asked what light could be thrown on how small-scale societies can be deeply cooperative internally but have very violent relations with neighbours. This connected with some of Sebastian’s recent use of agent-based modelling to investigate parochialism. His interest in these topics was spurred by observing some perplexing forms of sectarianism in the Lebanon. The meeting thus concluded, by chance, on a question to which the Club’s next meeting will be in part dedicated, the question of when behaviour whose motivation is ostensibly religious is genuinely so motivated, and when it is due to other factors.

Written by Dr David Mitchell, Senior Lecturer in Philosophy and Convenor of the Ottoline Club, NCH’s faculty club for interdisciplinary discussion