Cooperation and Social Cues; Ideology and Punishment

The following is my third-week reaction paper from a Fall 2006 class, Biological Approaches to Political Behavior, taught by Professors John Hibbing and Kevin Smith.

This week’s readings suggest a number of promising avenues for future research. Kurzban and Houser’s cooperation experiment, which provided evidence for three stable human cooperative types, should be re-run with social cues in place to settle Pruitt and Kimmel’s question about “dispositional qualities� in light of Ketelaar’s observations regarding social motives in human interactions. This could be accomplished in group games by assigning each group member a photo and a name. Group members besides the subject would be characters programmed to play as free riders, reciprocators or cooperators. This would be the social condition of the experiment. There would also be an asocial condition, run per Kurzban and Houser’s design. I would have enough subjects play enough games in each condition that clear cooperative types would emerge, then compare the playing strategies of each cooperative type in the asocial condition with the strategies of the same cooperative type in the social condition. Pruitt and Kimmel expected that “dispositional qualities,� or in this case cooperative type, would have “little impact in an impersonal setting as represented by most gaming environments.� Therefore, I expect the differences between cooperative types to be pronounced in the social condition than in the asocial condition.

Kurzban and Houser’s design can further be expanded to examine coalition formation. I would run a series of four-person cooperation games using the social cues described above in a social condition and no social cues in an asocial condition. Subjects would be told their aim was to maximize the group’s winnings by choosing an effective team. The three computer-controlled players would play as the three cooperative types, and at the end of each game subjects would get to choose which of the three computer players they wanted to play in the next game. The games would proceed until subjects had selected three teammates, at which time a final game with the selected team would determine that team’s score. The experimental comparison would be between the team scores of subjects in the social and asocial conditions. Following Smith et al.’s argument that humans can predict others’ game-playing behavior after a short period of interaction, my expectation is that social condition teams would have higher scores because the social cues make it easier to determine a potential teammate’s cooperative type.

An interesting question to ask in papers implementing both of the proposed extensions of Kurzban and Houser’s design is this: How many subjects will choose free-riding strategies, in the first case as individuals and in the second case as teams? While Kurzban and Houser’s results on this point were not statistically significant, it is interesting to note that free riders in their experiment earned the highest average amount, with the lowest standard deviation. In other words, they made more with less uncertainty. What does it mean for cooperative efforts, like representative democracy, that free riding is such a successful strategy?

Singer et al.’s discovery that men’s empathic responses are shaped by their valuation of other people’s social behavior also provides an opportunity for further research. Alford, Funk and Hibbing argue one characteristic of the distinct absolutist political phenotype is a preference for swift and severe punishment. While even I could not get the ultimate test of a fondness for swift and severe punishment – experimental imposition of the death penalty – past the IRB, I would propose a modification of Singer et al.’s design whereby the latency between the unfair act and the pain stimuli is varied. Using data from a politics-focused Wilson-Patterson Attitude Inventory administered after the fMRI scans, I would locate the absolutist and contextualist men and compare their responses to punishment of the unfair player, expecting three things. First, absolutist men would express a stronger desire for revenge than contextualist men on the revenge index given in the post-experiment questionnaire. Second, absolutist men would show greater activation than contextualist men in nucleus accumbens when an unfair player was punished. Third, absolutist men who observed swifter punishment would show greater nucleus accumbens activation than absolutist men who observed slower punishment.