The experiments reveal a human tendency to treat clustered judgments as stronger signals. When two partners gave nearly the same answer and were believed to be human, participants moved their estimates more than when partners disagreed. When the identical pattern came from computers, the pull of similarity weakened, which suggests people factor in who or what produced the judgments, not only the numbers themselves. The Bayesian modeling ties this behavior to rational belief updating rather than mere copying.
Understanding how judgment similarity guides conformity matters for designing fair information environments. Newsfeeds, group discussions, and decision-support tools all present patterns of agreement that can nudge people in different directions. Follow the link to the full article to see how these findings connect to improving collaboration, designing inclusive systems, and protecting individual judgment when group signals are misleading.
Abstract
Conformity is strongly shaped by perceived consensus, yet prior research has predominantly examined this process in discrete-choice contexts, where consensus is typically operationalized as the size of a uniform majority. In continuous real-world environments, however, social judgments naturally vary, making judgment similarity a potentially important cue for consensus. The present study investigated whether similarity among others’ judgments modulates conformity. Across two perceptual estimation experiments, participants first estimated the location of a briefly presented target and then revised their estimates after viewing two partners’ responses. In Experiment 1, in which partner responses were attributed to humans, participants showed greater conformity when partners’ estimates were highly similar. Computational modeling formalized this process within a Bayesian belief-updating framework, suggesting that judgment similarity increased the evidential weight assigned to social information. In Experiment 2, this similarity-dependent effect was attenuated when the same response structure was explicitly attributed to computer programs rather than human agents. These findings indicate that conformity in continuous judgment spaces cannot be fully explained by a domain-general statistical heuristic or passive imitation alone. Instead, individuals appear to evaluate both the content of social evidence and the attributed properties of its source, using judgment similarity as a cue for integrating information from others.
