Fig. 2: Human participants’ numbers of samples to decision for all studies.

Significant frequentist pairwise differences between condition means are shown as green horizontal lines (p < 0.05, Bonferroni corrected for the number of pairs in each study) for Study 1 full (n = 50 participants), Study 1 ratings (n = 51), baseline (n = 50), squares (n = 50), timing (n = 50), payoff (n = 51), Study 2 10 options (n = 75) and 14 options (n = 65). Only greater sampling for longer sequences in Study 3 proved significant. Null effects were concluded based on BF₀₁ > 3 (i.e. at least moderate evidence for equal means). Such pairs are connected by magenta horizontal lines. Detail on pairwise statistical tests depicted here can be found in Table 3 and Supplementary Table 3. Boxplots reflect first, second (median) and third quartiles, while whiskers reflect 1.5 interquartile range. Points reflect individual participant mean values.