When Choices Are Not Personal: The Effect of Statistical and Social Cues on Children's Inferences About the Scope of Preferences

Abstract

Individual choices are commonly taken to manifest personal preferences. The present study investigated whether social and statistical cues influence young children's inferences about the generalizability of preferences. Preschoolers were exposed to either 1 or 2 demonstrators’ selections of objects. The selected objects constituted 18%, 50%, or 100% of all available objects. We found that children took a single demonstrator's choices as indicative only of his or her personal preference. However, when 2 demonstrators made the same selection, then children inferred that it generalized to other agents of the same kind as the original demonstrator's, but not to agents of a different kind. Lastly, only when both demonstrators blatantly violated random selection (i.e., in the 18% condition) did children generalize the preference even to an agent of a different kind. Thus, from a young age, social and statistical cues inform children's naïve sociology.

ACKNOWLEDGMENTS

The study was part of SS's master's thesis. We want to thank the teachers, parents, and children for their participation, and Nitzan Maimon for help with data collection.

Notes

¹Note that the present pattern of findings regarding the effect of object population on children's inferences about the preferences of the demonstrating agent differs from the pattern reported by Kushnir et al. (2010). One possible explanation for this divergence has to do with the fact that different from Kushnir et al., here, children in the analogous single-demonstrator condition were exposed to an agent making the same choice twice in a row—rather than only once. To address this possibility, we tested another group of fifty 3- to 4-year-olds in the three object population conditions (ns = 16–18 per condition), in a single demonstrator–single demonstration condition. We analyzed the effect of object population only on children's responses on the inference-about-individual question. Replicating Kushnir et al.: a) A planned contrast comparing the frequency of selection of the target object in the 18% condition (M = 1.22, SD = 0.73) to those in the 50% (M = 0.63, SD = 0.62) and 100% (M = 0.88, SD = 0.89) conditions was significant, t(47) = 2.13, p < .05; and b) analyses against chance confirmed that only in the 18% condition did the frequency of children's choices of the target object significantly differ from chance, t(17) = 3.20, p < .005 (for the other two conditions, ps > .3). It is possible that in the present study, having the same demonstrator select 10 target objects, or repeating the same selection twice, weakened the effect of the relative frequency statistical cue.