Reply to van Hoorn: Converging lines of evidence

We agree with the comments by van Hoorn (1) on our critique (2): testing causal hypotheses about human behavior is a challenge (1, 3). Making progress requires specifying alternative hypotheses and then testing these hypotheses using diverse and converging lines of evidence. We have defended the hypothesis that social norms, which culturally coevolved with the institutions of large-scale societies including markets, influence economic decision-making. This hypothesis emerged from a larger set that we developed both at the outset of our project and as we went along. Our interdisciplinary team’s initial list of hypotheses included the idea that experimental games might spark an innate reciprocity module that would yield little variation across populations. We also considered the hypothesis that group-level differences might result from individual differences in wealth or income. Nevertheless, what emerged in the data in our first project was (i) substantial variation among 15 populations, (ii) a strong correlation with market integration, and (iii) little relation to individual-level economic or demographic variables. Not satisfied with our first effort, we sampled 10 new populations, replicated these findings with improved protocols (developed based on critiques of the Phase I), and then extended them to two additional experimental games. Along the way, we have explored alternative hypotheses using measures of genetic relatedness, social network position, anonymity manipulations, and framing tools. To our knowledge, no other existing hypotheses can better account for the observed patterns of variation.