| Last Updated:
Creating DOI. Please wait...
The social and behavioral sciences are undergoing a revolution spurred by mounting evidence that many published findings do not replicate. As a consequence, there is growing interest in scientific reform. One such reform, direct replication, has emerged as a leading solution. Despite its importance, the focus on direct replication may divert attention from another problem plaguing the social sciences: lack of generalizability to the full range of human contexts. Here, we use our lab’s recent cross-cultural study of social discounting to illustrate why developing broadly-applicable theories of human nature requires a complementary investment in generalizability. Social discounting is a repeatedly-documented and purportedly fundamental human tendency to sacrifice more for socially-close individuals. In our lab, we consistently document social discounting in U.S., MTurk, and college-student samples. However, when adapting a typical social discounting protocol to low-literacy and resource-scarce settings in Bangladesh and Indonesia, we find no independent effect of social distance on generosity, despite still documenting this effect among typical U.S. participants. Additionally, we document high-levels of “inconsistent” responding in Bangladesh and Indonesia. This finding strongly contrasts with current models of choice in various discounting literatures (e.g. social, probability, temporal, risk). These results highlight the need to complement current replication efforts with investment in strong checks on generalizability. By failing to do so, we risk developing theories of human nature that reliably explain behavior among only a thin slice of humanity.
CC-By Attribution 4.0 International