From extortion to generosity, evolution in the iterated prisoner’s dilemma
US Research, 2013
The prisoner’s dilemma (PD) is a scenario based framework used in game theory to understand strategic decision making. In the ‘game’, two criminals are caught and must decide whether to cooperate with each other or turn on each other. If they cooperate and both deny the crime they receive a lower sentence. If one confesses (defects) they will walk free, but the other will serve a longer sentence. Taken more broadly, the scenario emphasises that cooperation results in a larger payoff than non-cooperation.
The original PD evolved into the iterative prisoner’s dilemma (IPD), which sees individuals faced with the co-operate/defect dilemma multiple times. This evolution of the ‘game’ saw the appearance of extortion strategies, where one player is able to achieve a disproportionality high payoff. The complexity arises when shifting the IPD into an evolving population, where not just two people are involved, but multiple players who change over time. In such situations, which would seem to better reflect the way strategic decision making occurs in reality, extortion strategies do not appear to be as successful or popular. In the research by postdoctoral researcher Alexander Stewart (University of Pennsylvania) and Associate Professor Joshua Plotkin (University of Texas), the identification of a group of ‘generous’ strategies has interesting implications for addressing strategic decision making in an organisation.
Generous strategies are characterised by rewarding cooperation and punishing defection, but the focus is on forgiveness rather than harsh punishment. In evolving populations, the authors find these strategies to be successful and favoured, in some cases outperforming more traditional successful IPD strategies.
This research aimed to better understand the evolution of cooperation by exploring the subset of generous strategies arising from IPD scenarios. Additionally, the authors sought to provide evidence supporting their hypothesis that these generous strategies can be just as, if not more, successful in evolving populations (which more accurately reflect reality) than more traditional strategies.
The research is based on a number of mathematical simulations, where the parameters are the population size, the group of strategies available for ‘players’ to use and the rate at which ‘mutations’ are introduced. The mutations are designed to change the payoffs players receive when using each strategy. For the purpose of this research only weak mutations were used.
The simulations supported the author’s hypotheses that generous strategies can be just as successful as more traditional cooperation/defection strategies, and in some cases perhaps more successful. Other findings highlighted that:
- Generous strategies remain robust when the population is evolving (changing in size and mutations)
- Over time, generous strategies will dominate in an evolving population as there is a favourable tendency towards them
- Generous strategies outperform extortion strategies in an evolving population
- The most successful generous strategies are not extreme in that they do not see the cooperating player reduce their payoff by much in response to a defector.
The findings of this research suggests that a decision making approach based on generosity, rather than extortion, is more successful in the long run. For many diversity and inclusion issues in the workplace, the core of the problem lies in the existence of “in” and “out” groups, which may at times be aligned to majorities and minorities but can also be more subtle such as cliques. Recent UK research suggests that the most marginalised within an organisation (in terms of race, gender, disability and age) are likely to be the least well paid. This current research suggests a critical strategy to addressing this inequity is influencing the in-group to care about redressing the imbalance, and explaining that the group benefits of generosity could be helpful to all. A rising tide floats all boats, so to speak. Indeed focussing on this lever is key, knowing that while the out group is able to exert some influence over pay inequity, the extent to which they can do so is limited as they are already disempowered by virtue of being in the out group. Real progress can only be made by motivating the in group to take action to minimise or remove the presence of an out group.
Stewart and Plotkin’s research provides evidence to support approaching issues like pay equity and discrimination through the lens of cooperation and generosity. Their findings suggest that the whole population (whether this is an organisation or society more broadly) can benefit more from working together and cooperating, rather than any players looking to gain more for themselves through ‘defecting’. Ultimately, this would lead to the in group taking action to reduce the occurrence of things like discrimination or unequal pay, even though it may not provide them with a greater payoff or outcome in the short term. Over time, however, the continued use of generous strategies leads to a better outcome for the whole organisation or population.
To read the full article, see Stewart, A. J. and Plotkin, J.B. (2013) “From extortion to generosity, evolution in the Iterated Prisoner’s Dilemma” The Proceedings of the National Academy of Science, published online ahead of print. http://www.pnas.org/content/early/2013/08/28/1306246110.full.pdf html?sid=42128f37-9ca7-4d21-8e82-58f8cf1a2027