Fig. 4: Exploration correlates with socioeconomic disparities in academic achievement.

a Higher socioeconomic status (SES) correlated with higher standardized academic skills (n = 124). Statistical Test: Pearson correlation: t(122) = 4.93, p < 0.001, p-adjusted <0.001, r = 0.407, 95% CI [0.249, 0.545]. The dashed line indicates the population average for academic skills. b Higher SES correlated with better school grades (n = 122). Statistical Test: Cumulative Link Model: (z = 3.22, p = 0.001, p-adjusted = 0.003, b = 0.61, 95% CI [0.243, 0.990]). c Higher exploration, measured by the mean number of pumps on unreliable balloon trials, correlated with higher academic skills (n = 124). Statistical test: Linear regression: t(122) = 2.74, p = 0.007, p-adjusted = 0.009, r = 0.241, 95% CI [0.067, 0.399]. This relationship was observed for lower-SES adolescents (purple line; n = 66; t(64) = 3.92, p < 0.001, p-adjusted = 0.001, β = 0.44, 95% CI [0.220, 0.660]) but not higher-SES adolescents (green line; n = 58; t(56) = −1.01, p = 0.315, p-adjusted = 0.402, β = −0.13, 95% CI [−0.400, 0.130]). d More pumping correlated with higher grades (n = 122). Statistical test: Cumulative Link Model: (z = 2.73, p = 0.006, p-adjusted = 0.009, b = 0.26, 95% CI [0.076, 0.455]). Each data point represents a participant’s academic skill score (a, c) or grade (b, d). Lines represent linear regression fits; grey shading indicates 95% confidence intervals around the fitted mean. In b & d, the color gradient (pink to blue) represents lower to higher grades on a 1–7 scale. All statistical tests were two-sided. P-values were adjusted for multiple comparisons using the Benjamini-Hochberg procedure.