Fig. 4: Overview of the main analyses for experiment 2.

In this experiment participants learned about artificial profiles which did not have the trait similarity structures. Results indicate that participants use a coarse granularity structure when less social information is present. a Model 3 [Coarse Granularity & Population RP] is the best fitting model (n = 41). This model uses the average population as a reference point and coarse granularity for generalization. b Simulated data for the best performing model (n = 41). Unlike participants’ data, Model 2 [Coarse Granularity] was the best performing model, demonstrating that participants could have used a more optimal strategy. c Both plots display the average absolute PEs over time ± SEM. Top) Participants’ data shows a decrease in the PEs over time (ρ:−0.546, least squares line, red), this indicates participants were learning over time. Bottom) simulated data from the best fitting model (Model 3) shows a similar decrease in PEs over time, indicating that the models learned in a similar way to participants. d All three regressors (representing: 1 RW learning, 2 Coarse granularity, 3 Fine granularity), were significant (one-sided t-test), regressor 1: t(40) = −5.4617, p < 0.001, regressor 2: t(40) = −5.7377, p < 0.001, regressor 3: t(40) = −7.7059, p < 0.001, indicating that participants (n = 41) learned over time but also made use of both coarse and fine granularity. However, these regressors were correlated and conclusions regarding this GLM should thus be drawn with caution. Participants’ parameter estimates (for each regressor) are indicating by the individual data points, which are summarized by the adjacent boxplots of the same colour. The boxplots indicate the median (middle line), and the box is formed by the 25th, and 75th percentile. The whiskers extend to most extreme data points not considered outliers (1.5 times interquartile range), outliers are indicated with + signs. [One-sided t-test; * indicates p < 0.05, ** indicates p < 0.001, no correction for multiple comparisons]. CG coarse granularity, FG fine granularity, RP reference point, # number of, PEs prediction errors, SEM standard error of the mean, LSLine least squares line.