Motivational Systems in Political Collective Action: Managing Free Riders, Exiters, and Recruits

Download data and study materials from OSF

Principal investigators:

Andrew W. Delton

University of California, Santa Barbara



Leda Cosmides

University of California, Santa Barbara



Sample size: 1069

Field period: 8/3/2005-8/15/2005


What is the architecture of the motivational systems that underlie human collective action? Based on an evolutionary psychological analysis, we hypothesize that the mind contains separate punishment and reward systems designed to regulate the cooperation of others involved in the collective action. Because these mechanisms have different functions, they should be differentially sensitive to distinct situational cues. In other words, at least in the context of a collective action, punishment and reward are not equivalent. To regulate one’s own and others’ cooperation, the mind also needs a set of internal variables that track, among other things, one’s own contributions and one’s expected benefit from the collective action. We explore how situational cues and internal variables regulate punishment and reward. We test our hypotheses by asking respondents about a series of voters involved in a hypothetical upcoming presidential election.


H1: The more potential free riders will benefit, the more punitive sentiment respondents will feel towards them. Expected benefit will not influence reward toward contributing ingroup members.
H2A: If punishment is designed to induce cooperation, then more uncertainty regarding the success of the collective action will increase punitive sentiment toward free riders (and possibly exiters).
H2B: If reward is designed to induce cooperation, then more uncertainty regarding the success of the collective action will increase reward sentiment toward recruits (and possibly contributors).
H3. Punishment will be used to prevent members from exiting one’s coalition and such punishment will be modulated by cues that the exiting is particularly costly. Similar effects will not be observed for reward towards exiters.
H4. More reward will be used to recruit a potential party member who was formerly in the opposing party
H5A: Respondents’ contributions and expected benefits will be tracked by separate psychological variables.
H5B: The two variables will have independent and unique effects in predicting reward and punishment.
H6: Punishment and reward systems are designed to regulate within-group cooperation. The effects outlined above should be weaker or non-existent when respondents consider members of an outgroup.

Experimental Manipulations

1. The amount of benefit that free riders and contributors (both in the ingroup and outgroup) expected from the collective was varied. The hypothetical targets either expected a great deal of benefit or no benefit at all.
2. The probability that the collective action would be successful (i.e. one’s candidate would win) was varied. The collective action was either likely to succeed, likely to fail, or had a roughly equal chance of succeeding or failing. (In no case was the outcome certain.)
3. The cost potential exiters would impose on their ingroup by exiting was varied. They either left politics altogether (less severe) or joined the opposing party (more severe).
4. The benefit that a recruit provided was varied. Recruits either were not in any previous party (low benefit) or in the opposing party (high benefit).


1. Anger at ingroup free rider
2. Gratitude toward ingroup contributor
3. Anger at outgroup free rider
4. Gratitude toward outgroup contributor
5. Anger at ingroup exiter
6. Positive Encouragement toward ingroup exiter to keep from exiting
7. Gratitude toward potential recruit.
8. Respondents’ contributions to their political party (both previous and projected)
9. Respondents’ expected benefit from success of the collective action (both personally and for the nation as a whole)

Summary of Results

Present results are only preliminary. H1 was supported: Punishment toward free riders was modulated by expected benefit, but reward toward contributors was not. H2A and H2B were not generally supported: Probability of success had little effect on any variable, although there was a trend for lower probability to cause more reward for recruits. H3 was supported: Punishment toward exiters was modulated by the cost of their exiting but reward was not. (However, more reward than punitive sentiment was generally expressed toward exiters.) H4 was not supported: The “origin” of recruits did not influence reward expressed towards them. H5A was supported: Confirmatory factor analyses revealed contributions and expected benefits to be distinct factors. H5B was partially supported: These variables were generally independent predictors of punishment and reward, but there were no circumstances where either was a unique predictor. H6 was supported: Reactions were generally weaker to outgroup members.


The results provided mixed support for our hypotheses. Importantly, as predicted by our framework, there was some specificity in the operation of punishment and reward mechanisms: Punishment was preferentially modulated in situations where reward was not (H1 & H3) and there was a trend for reward to be modulated when punishment was not (for recruits; H2A & H2B). Punishment and reward were also preferentially activated by ingroup members (H6). Contributions and expected benefits do appear to be separate psychological variables, but, based on the measures used here, there was no specificity in their operation. Although some of our hypotheses failed to find support, this may be due to the difficulty of having respondents vividly imagine a hypothetical scenario using internet methodologies. Overall, our results provide additional support for evolutionary psychological theories regarding the motivational architecture of collective action.

Additional Information

We used anger to measure punitive sentiment and gratitude (or positive encouragement) to measure reward sentiment for two reasons. One, we were more interested in the thoughts and feelings that the hypothetical voters generated in our respondents, rather than whether our respondents would actually be willing to act on these thoughts and feelings. Two, because words such as “punishment” and “reward” are often closely associated with overt physical acts, in this context they might be associated with voter intimidation and vote buying. There are norms against these acts in American culture, and so we were concerned that few respondents would be likely to endorse punishment and reward in an election context.
Although we collected data from both members of political parties and independent voters, we summarize only data from members of political parties.