Can Social Projection Solve Social Dilemmas (Any Better Than Social Normative Models)?

Click to increase image sizeClick to decrease image size

Notes

Unless, apparently (see Table 1, row 3, columns 2 & 4), all available choices meet one's obligation to reciprocate equally, in which case one may consider self interest to break the tie of indifference.

I realize that in Kerr and Kaufman-Gilliland (1994) we were considering a moral norm that would not ordinarily arise in the one-shot PD game (without group discussion or some other venue for making commitments). However, the point of the quotation applies equally well to moral norms that can arise in the one-shot PD (e.g., to act pro-socially, to reciprocate, to act in the group's best interest).

All the PD games considered by Krueger et al. (this issue) are simple symmetric games (i.e., T – R = P – S = δ in Figure 1d. The particular values of the “nasty” PD shown in Figure 1c were used in this plot, and the generic outcome labels (T, R, P, S) are also shown.

For that matter, sometimes Krueger et al. (this issue) also consider straw-man versions of the other (non-social-normative) models. A striking instance is for the classic game theory predictions for bilateral switching in Table 1. Krueger et al.'s version of the classic theory holds that initial cooperators always switch to defect, and initial defectors never switch to cooperate, presumably because regardless of one's initial choice or one's initial expectation about the overall rate of cooperation, the defect choice dominates in a PD. But in the bilateral-switching version of the Last Minute Intrigue game, the choice between switching and not switching must consider a different set of expectations than governed the original PD choice. A cooperating Player A who attached a high (e.g., .80) probability to Player B cooperating in the first phase of the game must now attach an equally high probability to Player B defecting if she or he (Player A) switches in the second phase. Any classic game theorist would presume that Player A would use this information, and if the initial probability of reciprocation were sufficiently high, would predict that an initial cooperator would decline to bilaterally switch and that an initial defector would choose to bilaterally switch. Thus, classic game theory would make the same predictions as the projection model at Phase 2 of the Last Minute Intrigue game.

Krueger et al. (this issue) argue that it is plausible to assume that others who are more similar to us are more likely to share our preferences. But it is implausible to assume that the nature of the random process that determines payoff values affects similarity in this way.

The SVO theory does not mispredict behavior in the coordination game, it simply makes no prediction. Hence, it receives a 0 rather than a –1 in the coordination game/fourth inning.

Krueger et al. (this issue) appear to grant that people will act to enhance their reputation but suggest that reputational concerns are or should be irrelevant in the one-shot PD. This begs the question of whether the PM can account for such reputational or other effects in iterated games.

But not an unsurmountable one. For example, one would also predict this finding if chronic cooperators have experienced greater levels of trustworthy behavior from others than chronic defectors, and if all see themselves as relatively trustworthy people. It is also more agreeable for cooperators to project (and thereby discount the risk of exploitation) than noncooperators (who depend on others to act dissimilarly and hence be exploitable).