Robbers Cave: the Network of Intergroup Conflict
In 1954, Muzafer Sherif and Carolyn Wood Sherif conducted a social psychology experiment where 22 eleven- and twelve-year old boys were brought together in to examine the effects of intergroup conflict. All the boys were from similar socioeconomic backgrounds and most importantly, had never met before. In the first stage, the boys were split up into two groups, wholly unaware of the other half. They participated in a number of bonding activities meant to foster group unity over several days and towards the end, were allowed to learn of the other group’s existence. An “us” vs. “them” attitude developed very quickly and thus began stage two, where activities involving direct competition occurred. The two groups became very antagonistic toward one another. In stage three, the experimenters attempted to mend the gap between the two groups. They first applied the “mere contact” theory where the two groups were brought together in neutral situations in the hopes that the equal footing would decrease the friction between them. However this did not work and eventually the presence of superordinate goals that could be achieved more readily by the groups working together was able to bring the groups together. This appeared in the form of a bus breaking down that could only be restarted by everyone working together, paying for a movie that everyone wanted to see, and so on.
This experiment naturally lends itself to a well-defined network in which each group of boys is a cluster with weak/negative ties between the two groups. We can think of one group starting with all A and the other with all B and we would like to find the best way to have both groups adopt the same choice through network diffusion. In this case where both groups started off very antagonistic towards one another, the threshold q for adoption must be very high with potential negative consequences for any one node. For example, if a single boy in a group had decided to switch loyalties, he would have been ostracized from his own group and may not even be accepted by the other group. Thus, lowering the threshold may not be the easiest route. The experimenters were able to create weak ties by putting the groups into direct contact with one another, but as seen in class, this does not work well in high-threshold networks. Additionally, since the groups were so highly unified, it was difficult to add edges to any single node that could cause a cascade. Instead, the experimenters were able to strengthen the weak ties between the groups with the superordinate goals into strong ties, which then made the cascade possible.
When taking PSYCH 2800: Social Psychology here at Cornell, I found this experiment fascinating in how easily the experimenters were able to manipulate the groups into both disliking one another and then cooperating so predictably. Another similar experiment on minimal group paradigm was able to show how easily groups are formed on very little information, implying that clusters in a network would form very easily and from the experiment, those clusters would be very difficult to have participate in a cascade. Extending an example given in class, a direct comparison would be in politics where direct competition arises between parties during elections and it is more difficult to reconcile different parties/clusters through mere contact with one another.
Source:
Revisiting Robbers Cave: The easy spontaneity of intergroup conflict