Charitable giving as a signal of trustworthiness: Disentangling the signaling benefits of altruistic acts

Abstract

It has been shown that psychological predispositions to benefit others can motivate human cooperation and the evolution of such social preferences can be explained with kin or multi-level selection models. It has also been shown that cooperation can evolve as a costly signal of an unobservable quality that makes a person more attractive with regard to other types of social interactions. Here we show that if a proportion of individuals with social preferences is maintained in the population through kin or multi-level selection, cooperative acts that are truly altruistic can be a costly signal of social preferences and make altruistic individuals more trustworthy interaction partners in social exchange. In a computerized laboratory experiment, we test whether altruistic behavior in the form of charitable giving is indeed correlated with trustworthiness and whether a charitable donation increases the observing agents' trust in the donor. Our results support these hypotheses and show that, apart from trust, responses to altruistic acts can have a rewarding or outcome-equalizing purpose. Our findings corroborate that the signaling benefits of altruistic acts that accrue in social exchange can ease the conditions for the evolution of social preferences.

Introduction

Humans frequently cooperate with non-kin others and incur costs to benefit them. The question of how such cooperative behavior can be explained has attracted considerable attention across several decades and disciplines (see West, El Mouden, & Gardner, 2011 for a critical review). A large body of literature has shown that cooperation can be a manifestation of self-interest if it is likely to be reciprocated with a benefit that outweighs its costs in the not-too-distant future (Trivers, 1971, Axelrod and Hamilton, 1981, Nowak and Sigmund, 1998). However, these explanations are restricted to interactions between members of relatively small groups, where cooperators and defectors can be identified and respectively targeted by reward or punishment (Bowles & Gintis, 2011: 63–70; Leimar and Hammerstein, 2001, Panchanathan and Boyd, 2003). Moreover, empirical evidence has accumulated suggesting that cooperative behavior may be motivated by psychological predispositions to benefit others (henceforth, social preferences) (Camerer, 2003: Ch. 2). However, since cooperative behavior is often costly, the evolution of social preferences in humans is difficult to explain in an individual-selectionist framework (although see Delton, Krasnow, Cosmides, & Tooby, 2011). This has led to a renewed interest in models of multi-level selection (Wilson, 1975, Gintis, 2000, Boyd et al., 2003).

Models of multi-level selection assume that there is both competition between individuals (within groups) and between groups, and groups with a higher proportion of cooperative individuals will be more likely to survive inter-group competition and adverse environmental conditions. For cooperation to be sustained in a population, positive assortment of cooperators, i.e., the higher likelihood of cooperators interacting with cooperators than with non-cooperators, must outweigh the ratio of costs c (for the cooperator) to benefits b (for the rest of the group) of cooperation (Eshel and Cavalli-Sforza, 1982, Bowles and Gintis, 2011: 52–59). However, since models of multi-level selection are mathematically equivalent to models of kin selection where genetic relatedness is implied by the limited dispersal of individuals within groups, some authors have argued that it is not necessary to resort to multi-level selection to explain the evolution of cooperation (West et al., 2011). We leave it to others to answer questions regarding to what degree population structures led to positive assortment of genetically related individuals in human prehistory and whether multi-level selection is necessary to explain how social preferences and cooperation have evolved. Instead, we argue that cooperative acts can be credible signals of an individual's social preferences and, through favorable treatment of these individuals in social exchange, ease the conditions for their evolution, whether in a kin or a multi-level selection framework.

Gintis, Smith, and Bowles (2001) show that cooperation can evolve as a costly signal of an unobservable but relevant quality, if this quality is causally related to an individual's ability to cooperate (see also Leimar, 1997, Roberts, 1998, Lotem et al., 2003, Smith and Bliege Bird, 2005). In the simplest case, there are high-quality and low-quality types who incur low costs (c₁) or high costs (c₂ > c₁), respectively, from sending the signal. If the benefits (s) from being interacted with, conditional on having sent the signal, compensate the high-quality types but not the low-quality types (c₂ > s > c₁), only the high-quality types can afford to send it and thus will be identified as such. If, moreover, sending the signal yields a higher net benefit for the sender than not sending the signal, type-separating behavior can evolve in which high-quality types send a signal, low-quality types do not send a signal, and agents are only interacted with if they sent a signal. Gintis et al. (2001) also analyze the evolutionary dynamics of their model and show that cooperation as a type-separating signal is evolutionarily stable under plausible conditions.

These predictions also hold if social preferences are the unobservable quality of interest. Agents with social preferences are the high-quality types, who derive a psychological reward (r₁ > 0) from benefiting others, whereas individuals lacking social preferences are the low-quality types, who only care about their own payoffs (r₂ = 0). Although the material costs are the same for both types (c₂ = c₁ = c), the psychological rewards make it “cheaper” for the high-quality type to cooperate (c > c–r₁). The condition that must hold for cooperation to be a type-separating signal is c > s ≥ 0. In other words, high-quality types cannot be fully compensated in material terms for the costs they incur. In fact, their cooperative acts must be truly altruistic (henceforth, altruistic acts). This requires that the existence of individuals with social preferences is maintained by another evolutionary mechanism (e.g. kin and/or multi-level selection). However, as long as s > 0, cooperators receive partial compensation, which we call signaling benefits. We will show next that altruistic acts can induce signaling benefits through social exchange and that this can ease the conditions for the evolution of social preferences.

Social exchange among unrelated individuals has been an important part of human sociality for tens of thousands of years and arguably a driving force in the evolution of the human mind (Cosmides & Tooby, 1992). The upper half of Fig. 1B shows a Person X and a Person Y engaging in social exchange that is not based on a formally binding agreement (Dasgupta, 1988, Coleman, 1990: Ch. 5). The social exchange can be mutually beneficial if Person X makes a transfer x first and Person Y makes a back transfer y, which is tripled to reflect the gains from trade. While a selfish Person Y has a real incentive to keep x without sending back y, a Person Y with social preferences will make a back transfer y, such that 3y > x. For Person X, a trust problem arises as he or she does not know whether Person Y is cooperative and will make a back transfer that is sufficiently high. Referring to the vast social science literature on social exchange (e.g. Ostrom and Walker, 2003, Fehr, 2009), we call Person Y's cooperative behavior trustworthiness and we call Person X's transfer, which is motivated by the expectation of gain from Person Y's back transfer, trust (see also definitions in bottom half of Fig. 1B).

Now suppose that Person X, before engaging in social exchange with Person Y, observes Person Y in the situation depicted in Fig. 1A. Here, Person Y has the opportunity to perform an altruistic act in the form of a charitable donation. Then, Person X can condition his or her transfer in the social exchange on whether Person Y acted altruistically (Y₁) or not (Y₂). Since only a Person Y with social preferences will both give to charity and make a back transfer in social exchange, Person X can infer Person Y's type from his or her donation to charity or the lack of it. Consequently, while Person Y₂ will be disregarded by Person X, Person Y₁ will be partly compensated for his or her altruistic act by the gains he or she makes from trade (i.e. c′ = c–s, where c are the costs of the altruistic act and s = x–y are the signaling benefits). In addition, since Person X benefits from the social exchange with a trustworthy interaction partner, the benefits for the group (excluding Person Y) increase as well (i.e. b′ = b + 3y–x, where b are the group benefits from Person Y's altruistic act and 3y–x are Person X's gains). Now, as c′/b′ < c/b, signaling eases the conditions for the evolution of social preferences, because it lowers the degree of positive assortment necessary to maintain such traits in the population.

This evolutionary argument implies that what we should observe today is that altruistic behavior and trustworthiness are correlated. Moreover, individuals acting altruistically will be trusted more in social exchange because they will be expected to have social preferences and thus to be trustworthy. Note that the reverse is not true in general. That is, observing someone behaving trustworthily does not necessarily tell us that this person has social preferences – he or she could be selfish and behave trustworthily to acquire a reputation for being trustworthy and to be trusted more in the future (Bolton, Katok, & Ockenfels, 2004). However, here we exclude this possibility with a one-shot (i.e. non-repeated) game design, in which a selfish Person Y has no incentive to act trustworthily and thus altruistic acts can be a signal of trustworthiness via social preferences only.

There is ample evidence for a positive correlation between altruistic behavior and trustworthiness measured in laboratory experiments with economic games (Barclay, 2004, Ashraf et al., 2006, Chaudhuri and Gangadharan, 2007, Albert et al., 2007, Blanco et al., 2011; Fehrler unpublished; Gambetta & Przepiorka unpublished). Four of these experiments were also designed to investigate whether subjects who act altruistically are thereafter trusted more by third parties in social exchange, and they find support for this conjecture (Barclay, 2004, Albert et al., 2007; Fehrler unpublished; Gambetta & Przepiorka unpublished). Barclay and Willer (2007) and Sylwester and Roberts (2010) provide similar evidence from experiments with public good games. However, there is experimental evidence showing that subjects who help others or donate more to charity receive more in return from third parties (Wedekind and Milinski, 2000, Milinski et al., 2002). Thus, observing subjects in social exchange responding positively to altruistic acts does not tell us to what extent these responses reflect trust and to what extent they are mere transfers of resources intended to unconditionally reward the altruistic individual. Moreover, there is compelling experimental evidence that some subjects prefer egalitarian outcomes (Bolton and Ockenfels, 2000, Dawes et al., 2007, Fehr and Schmidt, 1999). Since altruism entails giving away resources or incurring costs, positive responses to altruistic acts could also be a manifestation of inequality aversion. However, while trust is motivated by pure self-interest, rewards or responses based on inequality aversion are not and would thus remain in need of an evolutionary explanation.

In our computerized laboratory experiment, we test whether altruistic behavior in the form of charitable giving is indeed correlated with trustworthiness. Moreover, we test whether a charitable donation increases the observing agents' trust in the donor. Our experimental design allows us to disentangle trust from rewarding and outcome-equalizing transfers as responses to altruistic acts.