External validity – Cooperation without punishment

Guala argues that the key source of disagreement concerning the external validity of SRT is whether or not costly punishment, in the face of free-riding, supports spontaneous cooperation outside the laboratory. Guala argues that punishment to support cooperation is, in fact, coordinated and cheap. However, it should be acknowledged that there are many forms of spontaneous cooperation that emerge in the absence of punishment, despite free-riding. One example is voluntary blood donation. The donor and recipient remain anonymous and never meet. Blood donors tend not to talk about being a donor (Ferguson & Chandler Reference Ferguson and Chandler2005), and the number of donors they know does not influence their decision to donate (Piliavin & Callero Reference Piliavin and Callero1991). While there are organized blood drives, blood donors tend to donate at drop-in centres at times they find convenient. Although an anonymous, relatively high-cost spontaneous act of altruism, blood donation is marked by a large free-rider problem: about 6% of the eligible population donate (Ferguson et al. Reference Ferguson, France, Abraham, Ditto and Sheeran2007). Evidence shows that feelings of warm glow (Andreoni Reference Andreoni1990) are a key motivation for blood donation (Ferguson et al. Reference Ferguson, Farrell and Lawrence2008; in press) Recent evidence also shows that registering for posthumous organ donation is likewise motivated by emotional regulation through anticipated regret (O'Carroll et al. Reference O'Carroll, Foster, McGeechan, Sandford and Ferguson2011). Finally, under specific conditions free-riders may be tolerated as this enables maximum group benefit (MacLean et al. Reference MacLean, Fuentes-Hernandez, Greig, Hurst and Gudelj2010). Thus, punishment is not needed for many forms of cooperation.

Internal validity – Individual differences

An experiment is internally valid when alternative explanations or additional mechanisms that contribute to the effect are identified or controlled. Given the degree of heterogeneity observed in economic tasks, there is a growing realization that economic models need to consider the role of personality traits as additional explanatory variables (Ferguson et al. Reference Ferguson, Heckman and Corr2011; Wischniewski et al. Reference Wischniewski, Windmann, Juckel and Brune2009). Ferguson et al. (Reference Ferguson, Heckman and Corr2011) propose a model of personality within economics whereby the expressed behaviour on any economic task reflects the motivations associated with the personality trait, as well as task constraints, incentives, and so forth. For example, someone who is motivated to maximize rewards is more likely to show cooperative behaviour when reputation building is possible, but free-ride in an anonymous game: behavioural expression is strategic.

One implication for SRT is that some agents will punish only when it is strategically advantageous to do so. Indeed, this is what has been observed with respect to those high in Machiavellianism, who behave selfishly when punishment is not expected, but cooperate when it is (Spitzer et al. Reference Spitzer, Fischbacher, Herrnberger, Grön and Fehr2007). Individual differences will also influence decisions both to punish and how to respond to punishment. Some individuals (e.g., psychopaths) will be more willing than others to punish across all contexts; however, some others (e.g., the highly anxious) will respond to punishment by cooperating, while others will continue to free-ride or retaliate (e.g., psychopaths). Finally, when individuals have the opportunity to meet and communicate (as is the real world), some (e.g., those high in psychopathy) will be more likely to exploit people whom they identify as possessing exploitable traits (e.g., agreeableness; Buss & Duntley Reference Buss and Duntley2008). Thus, while expressed behavioural styles are often noted in the SRT literature (Ule et al. Reference Ule, Schram, Riedl and Cason2009), these may reflect, in part, the operation of pre-existing individual differences.

Ecological validity

By increasing the ecological validity of laboratory-based tests of SRT, the external validity gap can be reduced (List Reference List2009). Guala highlights studies allowing for multiple punishment options, communication, or retaliation. We highlight two other evolutionary parameters that are important in this regard: (1) resource holding and status and sex; and (2) legitimacy of free-riding.

While resources (e.g., physical, financial) are not equally distributed in the population, this is not captured within standard laboratory tests of SRT, where participants often have access to equal resources. The unequal distributions of resources will influence levels of aggression (punishment). For example, people are less likely to be aggressive towards those with greater physical resources (Archer & Benson Reference Archer and Benson2008). Indeed, people are less willing to punish a transgression by high-status individuals or groups (Eckel et al. Reference Eckel, Fatas and Wilson2010). There are also established sex differences in the use of aggression: Males use direct aggression, and women indirect (Archer Reference Archer2004). Therefore, pronounced sex differences should be observed in both the degree and type of punishment used in laboratory tests of SRT. Women also may be less willing to punish males who hold stronger physical resources. Thus, power, status, and sex may result in the use of punishment for many different reasons other than to enforce norms of fairness.

In real-world contexts, non-contributions may occur for a number of legitimate reasons (e.g., illness), and for sustained cooperation people need to be able to distinguish legitimate from non-legitimate non-cooperation (Lotem et al. Reference Lotem, Fishman and Stone1999). Within laboratory tests of SRT, free-riders do not have any legitimate reason not to contribute (all have an endowment and can contribute). SRT, therefore, needs to examine the role of legitimacy of non-contribution and how this influences the level of punishment adopted. Legitimate free-riders should be treated like cooperators.

To conclude, the generality of SRT is limited by the existence of anonymous spontaneous cooperation, in the face of free-riding, that is maintained in the absence of punishment or even conditions where free-riding is tolerated. Personality, status, sex, and legitimacy of non-cooperation need to be examined in order to increase the internal and ecological validity of SRT experiments and, ultimately, its external validity.