Prior research has yielded mixed findings on whether learners’ certainty reflects veridical probabilities from observed evidence. We compared predictions from an idealized model of learning to humans’ subjective reports of certainty during a Boolean concept-learning task in order to examine subjective certainty over the course of abstract, logical concept learning. Our analysis evaluated theoretically motivated potential predictors of certainty to determine how well each predicted participants’ subjective reports of certainty. Regression analyses that controlled for individual differences demonstrated that despite learning curves tracking the ideal learning models, reported certainty was best explained by performance rather than measures derived from a learning model. In particular, participants’ confidence was driven primarily by how well they observed themselves doing, not by idealized statistical inferences made from the data they observed.

Many social and legal conflicts hinge on semantic disagreements. Understanding the origins and implications of these disagreements necessitates novel methods for identifying and quantifying variation in semantic cognition between individuals. We collected conceptual similarity ratings and feature judgements from a variety of words in two domains. We analyzed this data using a non-parametric clustering scheme, as well as an ecological statistical estimator, in order to infer the number of different variants of common concepts that exist in the population. Our results show at least ten to thirty quantifiably different variants of word meanings exist for even common nouns. Further, people are unaware of this variation, and exhibit a strong bias to erroneously believe that other people share their semantics. This points to one factor that likely interferes with productive political and social discourse.

COVID-19 and the 2021 U.S. Capitol attacks have highlighted the potential dangers of pseudoscientific and conspiratorial belief adoption. Approaches to combating misinformed beliefs have tried to “pre-bunk” or “inoculate” people against misinformation adoption and have yielded only modest results. These approaches presume that some citizens may be more gullible than others and thus susceptible to multiple misinformed beliefs. We provide evidence of an alternative account it’s simply too hard for all people to be accurate in all domains of belief, but most individuals are trying. We collected data on a constellation of human beliefs across domains from more than 1,700 people on Amazon Mechanical Turk. We find misinformed beliefs to be broadly, but thinly, spread among the population. Further, we do not find that individuals who adopt one misinformed belief are more likely to engage in pseudo-scientific or conspiratorial thinking across the board, in opposition to “slippery slope” notions of misinformation adoption.

People rely heavily on information from the social world to inform their real-world beliefs. We ask whether the perceived prevalence of a belief, divorced from any direct evidence, serves as an independent cue in belief updating. Using real-world pseudoscientific and conspiratorial claims, our experiment (N = 403 American adults) shows that increases in people’s estimates of the prevalence of a belief led to increases in their endorsement of said belief. Belief change was most dramatic when initial beliefs were most uncertain and when novel prevalence information was most convincing, suggesting that people weigh social information rationally with other information sources. We discuss the implications of our results in the context of online misinformation.