Replicable evidence for uncalculating cooperation as a signal of trustworthiness

We investigate the replicability of prior work (Jordan et al., 2016) hypothesizing that “uncalculating cooperation” (i.e., cooperation without careful consideration of the costs and benefits) can signal an actor’s commitment to behaving prosocially, and thus trustworthiness. Jordan et al. found that subjects recruited from Amazon Turk make less calculating cooperative decisions when their decision-making process is observable to others, suggesting that uncalculating cooperation is used to signal trustworthiness. Moreover, Jordan et al. found that uncalculating cooperation is perceived as a signal of trustworthiness (subjects preferentially trust cooperators who made uncalculating, rather than calculating, decisions) and actually is a signal of trustworthiness (uncalculating cooperators actually behave in a more trustworthy manner). A recent replication effort (Holzmeister et al., 2024)—investigating the replicability of 26 studies with MTurk samples published in PNAS between 2015-2018—attempted to replicate the finding that observability increases uncalculating behavior, with mixed results (a significant effect in one, both not both, of two batches of data collection). However, subjects in both batches of Holzmeister et al. data, who were recruited directly from MTurk, performed relatively poorly on task comprehension questions compared to Jordan et al., 2016. We thus present results from an additional pre-registered replication of Jordan et al., 2016, conducted through the CloudResearch platform, that recruited a more attentive sample (comparable to the original study), as well as analyses drawing on all studies collected thus far. Together, we find strong evidence that uncalculating cooperation is used as, perceived as, and actually is a signal of trustworthiness, especially among attentive samples.