Data and Statistical Analyses

Several independent variables were of interest as part of an analysis of individual and age differences in foraging behavior under conditions of varying target value and prevalence. Specifically, we analyzed the number of patches observers visited in a block and the number of uncollected targets (targets left behind in a patch). We further assessed optimality of foraging behavior according to MVT by analyzing and the average rate of collection (points per unit time) across the entire block and the instantaneous rate of collection as a function of “reverse click” in the patches of each block. (Reverse clicks count backward from the last click in a patch, rather than forward from the first. Since clicks per patch are not fixed, these are different ways to look at the same data.).

Finally, we analyzed observers’ individual choices of the targets collected in a patch to reveal target preferences that were expected to vary with the experimental manipulation. We used mixed ANOVAs, one-sample T-tests, and paired T-tests to test for differences between age groups and conditions. Whenever the data deviated from the normal distribution, we checked that non-parametric Friedman tests of differences among repeated measures produced the same results as the reported ANOVA.

We further analyzed whether individual differences in the responsiveness to the experimental manipulation were related to individual sensitivity to reward and punishment as measured by the BIS BAS scale. We ran multiple linear regressions with the experimental variable (collection of low-value, high-prevalence target) as dependent variable and BIS score and Age (in years) as predictors and with BAS score and Age as predictors.

For all analyses, we also report the Bayes factor (BF). In contrast to classic hypothesis testing based on p-values and effect sizes, the sample size is less critical to interpret the evidence for or against a given hypothesis based on a BF. Furthermore, the BF gives an estimate of how strongly the data support not only the presence of a hypothesized effect, but also how strongly a null effect is supported. BF01 was computed as evidence for H0/H1 and BF10 as evidence for H1/H0 (i.e. 1/BF01). Thus, BF01>1 indicates support for H0 (null model) and BF10>1 indicates support for the H1.