Words matter. Images matter. The Scientific Inquirer needs your support. Help us pay our contributors for their hard work. Visit our Patreon page and discover ways that you can make a difference. http://bit.ly/2jjiagi
The roots of entrenched beliefs run deep and strong. Whether they are social, religious, or political in origin, no amount of information can sway people from what they believe. Or so it seems. Attempts at understanding why people hold their on to convictions — in other words, their social behavior — even when confronted with evidence to the contrary, have been fertile ground for the social sciences. Bryce Morsky, a post-doctoral researcher at the University of Pennsylvania, and Erol Akçay, an assistant professor, integrated a game theory approach to analyzing the reasons people prove so difficult to sway.
SCIENTIFIC INQUIRER: What prompted you to study how social norms and beliefs take root and spread?
BRYCE MORSKY: Our research initiated with trying to understand collective identity: how it emerges, and how it can coordinate social interactions. This goal was part of a broader research agenda to understand human social behaviour, funded by DARPA’s NGS2 program (Next Generation Social Science). This work has applications to group dynamics on a variety of scales and in a variety of cases, including political groups, religion, and civilization. From here, we began to focus on social norms as a shared mechanism for coordination. We were influenced by the works of Christina Bicchieri on social norms.
A social norm is an informal rule that prescribes behaviours to an individual, who will follow them given some expectations. To Bicchieri, and in our work, the role of expectations is central in determining whether or not the rule is followed. There has been much excellent work on how such social norms operate. We were interested in furthering this line of research. Where do norms come from? How do conflicting norms interface? What’s the evolutionary story? What’s the role of rationality?
SCINQ: What did using a game theory centered approach offer in terms of analysis that other approaches do not? How is it complimentary?
BM: Game theory is about strategic interactions, and evolutionary game theory is about how frequencies of strategies change over time. Much previous work on social norms uses game theory, which is well suited to studying cooperation and coordination. The Prisoner’s Dilemma, the Hawk Dove game, and the Divide-the-Dollar game are good examples. These games represent conflict between prosocial and antisocial behaviour. Further, there is a concept from game theory, the correlated equilibrium, which is well suited to describing social norms (as we define them).
Originated by Robert Aumann, a correlated equilibrium is a solution to a game that features a third party that signals to the players how to behave. Although these signals are sent privately, the players have an expectation of what their opponents have been instructed to do. A correlated equilibrium is a case where it’s rational for the players to obey the third party given their expectations. Herbert Gintis argued that social norms act as these third parties, choreographing the behaviour of individuals. With the right signalling, such a choreographer can promote prosociality.
SCINQ: According to your findings, how do otherwise meaningless events develop significance in the absence of a “choreographer”?
BM: We used an evolutionary approach to show how a choreographer (Gintis’ view of a social norm) can emerge from meaningless events. Players hold social norms and interact with one another thereby receiving payoffs. Selection then acts through imitation; the social norms of those players with higher payoffs are imitated more often.
Now, consider a collection of natural events that are correlated in some way unknown to the players. These events don’t have any intrinsic meaning; that is to say, they don’t suggest any particular behaviour to a player who observes them. Initially, let the players have no beliefs about what these events mean and therefore play some default strategy, a sort of null norm. Then, permit “mutations” to occur in the population. Over time, players will believe that these events have some meaning when they observe them: what they ought to do, and what others may do. In this way, beliefs can accumulate in the population. However, players will only obey their behavioural recommendations if it’s rational to do so, if their expected payoff from obeying them is greater than disobeying given the expectations of what others may do.
For a broad class of games, norms obeyed in this way can neutrally invade. Amongst these norms, there are those with expectations that are consistent with respect to the behaviour of others. And, in particular, there are norms that are self-consistent; the behaviours they recommend are the best behaviours to play when playing others that share the same norm. These social norms can fix in the population and form a correlated equilibrium. The underlying events have no intrinsic meaning; there is no design. However, normative meaning has formed, which coordinates players’ interactions, and can increase prosocial behaviour. All players can be better off as if the right signalling was designed and enacted. This process works as a blind choreographer, analogous to Richard Dawkins’ blind watchmaker.
SCINQ: Can you define evolutionarily stable and its significance?
BM: A norm is evolutionarily stable when a population holding it cannot be invaded by a different and rare norm. It’s significant because any further “mutations” in the norm cannot become established and are lost from the population (although, a large influx of players holding a different norm could invade). For this model, it’s the end result of the evolutionary story. Of course, this assumes that no other aspects of the model are changing, such as the game being played or the underlying events in the world.
SCINQ: Do participants need to gain some tangible benefit for a behavior to become a norm?
BM: The benefits are tangible in the sense that they determine the imitation dynamics. So, those players with norms that earn them a higher payoff than on average increase in frequency. However, the interpretation of the game’s payoffs is quite open. When players interact, they could be earning some tangible payoffs such as food or money. Or, perhaps, they win something more ephemeral, such as a social reward. As long as the imitation selection mechanism is driven by these payoffs, that players perceive higher payoffs as desirable, the model operates.
SCINQ: At what point does the behavior and its result become self-fulfilling because the group acts uniformly when presented with a situation?
BM: This model doesn’t feature collective action. However, social norms can be self-consistent and the actions of the group can be consistent across all situations. At this point, a self-consistent norm has become established. It’s individually rational to obey the norm. On top of that, it’s rational to obey the norm given that everyone else is obeying it. This state arrives once evolution finds such a consistent norm. When mutations occur that make it rational to obey the norm, a new behaviour emerges, and those norms that are self-consistent tend to become fixed in the population (and remain stable, since they form correlated equilibria).
SCINQ: The way many people directly affected by Ebola outbreaks in Africa have shown how entrenched superstition/cultural beliefs can be. Despite the fact that handling of corpses who have succumbed to Ebola spreads the disease and can often cause them to die, they continue to insist on the practice. The benefit that once reinforced the behavior is no longer as strong as it once was. Yet it persists. Why?
BM: I can think of a few explanations drawn from our model. One is that these now negative behaviours are tied to other things and so cannot be easily disentangled. Perhaps then, the net benefits of the norm are still positive. Another is that the “mutant norm” (safer handling of corpses) receives a low payoff from being taboo in the context of the majority having the current norm. Though it is socially beneficial and perhaps evolutionarily stable if enough individuals believed in it, it hasn’t reached a high enough threshold to invade. Norms needn’t be prosocial.
The correlated equilibrium that emerges from our model frequently results in higher payoffs to all players than if they had no beliefs and played the default strategy. However, there are correlated equilibria that result in lower payoffs. A norm that implements one would be evolutionarily stable, and yet it would make everyone worse off. If everyone held a different norm, they could be better off and that norm could even be evolutionarily stable. However, it cannot be reached by small changes in the population.
SCINQ: Finally, what is next for you in terms of research?
BM: We have some interest in exploring this model experimentally, and we have several ideas on how to extend it theoretically. However, currently, we are working on a related problem, norms of reciprocity, signalling, and the emergence of communities. Given that individuals tend to be conditional cooperators (i.e. they cooperate if a sufficient proportion of others do so too), we are interested in how signalling can induce a cooperative community to emerge even if the threshold to cooperate hasn’t actually been met. Essentially, people wouldn’t be so cooperative if they knew the true degree of cooperation in the population. Due to an inaccurate representation of the level of cooperation, more cooperation is elicited than otherwise would occur. This mechanism could bootstrap cooperative communities. When can this occur? How do such communities emerge and crash? How can they be sustained?
IMAGE SOURCE: Creative Commons; Bryce Morsky; Centers for Disease Control