Altruism, fast and slow? Evidence from a meta-analysis and a new experiment

Abstract

Can we use the lens of dual-system theories to explain altruistic behavior? In recent years this question has attracted the interest of both economists and psychologists. We contribute to this emerging literature by reporting the results of a meta-study of the literature and a new experiment. Our meta-study is based on 22 experimental studies conducted with more than 12,000 subjects. We show that the overall effect of manipulating cognitive resources to promote the "intuitive" system at the expense of the "deliberative" system is very close to zero. One reason for this null result could be that promoting intuition has heterogeneous effects on altruism across different subgroups of subjects or contexts. Another reason could be that there simply is no real effect and that previously reported single results are false positives. We explore the role of heterogeneity both by performing a mediator analysis of the meta-analytic effect and by conducting a new experiment designed to circumvent the issue of potential heterogeneity in the direction of the effect of promoting intuition. In both cases, we find little evidence that heterogeneity explains the absence of an overall effect of intuition on altruism. Taken together, our results offer little support for dual-system theories of altruistic behavior.

The Dual-Process Approach to Human Sociality: A Review

“…However, a more recent meta-analysis (Fromell, Nosenzo & Owens, 2018) found an interaction between gender and cognitive mode, in a slightly different direction than Rand et al (2016). Specifically, Fromell et al (2018) meta-analysed 19 studies (total N = 10,898) and, in line with Rand et al (2016), they found that there is an interaction between gender and cognitive process manipulation; however, in contrast to Rand et al (2016), they found that promoting System 1 has no effect on women, whereas it has a marginally significant negative effect on men. It is worth noticing that Fromell et al (2018) used broader inclusion criteria than those used by Rand et al (2016).…”

“…Specifically, Fromell et al (2018) meta-analysed 19 studies (total N = 10,898) and, in line with Rand et al (2016), they found that there is an interaction between gender and cognitive process manipulation; however, in contrast to Rand et al (2016), they found that promoting System 1 has no effect on women, whereas it has a marginally significant negative effect on men. It is worth noticing that Fromell et al (2018) used broader inclusion criteria than those used by Rand et al (2016). In particular, they included also dictator games in which the recipient was a charitable organization and those with varying cost-to-benefit ratio.…”

“…Observe that, althoughTable 2seems to suggest that ego-depletion studies trend towards a negative effect of ego-depletion on altruism, the overall effect is not significantly different from zero(Fromell et al, 2018).…”

“…In sum, two meta-analyses of the cognitive basis of altruistic behaviour agree on the existence of a differential effect on women and men, but they disagree with the main effect: Rand et al (2016) argue that System 1 favours altruism for women but not for men ;Fromell et al (2018) argue that System 1 disfavours altruism for men but not for women.…”

“…Second, Rand et al's (2016) meta-analysis includes a relatively small number of studies, as many experiments were published after the meta-analysis. In contrast, although Fromell et al (2018) include more studies, they also include studies using dictator games in which the altruistic option also maximizes social welfare. The inclusion of these studies can create a confound because it is known that men are more altruistic than women in these situations (Andreoni & Vesterlund, 2001).…”