Learning the Relevance of Relevance and the Trouble with Truth: Evaluating Explanatory Relevance across Childhood

ABSTRACT

In four experiments, we investigate how the ability to detect irrelevant explanations develops. In Experiments 1 and 2, 4- to 8-year-olds and adults rated different types of explanations about “what makes cars go” individually, in the absence of a direct contrast. Each explanation was true and relevant (e.g., “Cars have engines that turn gasoline into power”), true and irrelevant (e.g., “Cars have radios that play music”), or a false statement that would be relevant if it were true (e.g., “Cars have rockets that speed them up”). Participants of all ages spontaneously indicated that false explanations were less helpful than relevant explanations. However, there was a developmental shift for irrelevant explanations: 4-year-olds only detected irrelevant explanations that did not involve internal features of cars (e.g., “Cars have parking lots that they park in”). Crucially, this shift between age 4 and 5 cannot be explained by 4-year-olds’ lack of knowledge since 4-year-olds correctly indicated that relevant explanations were more helpful than irrelevant feature explanations when given a direct contrast in Experiment 3. These results are further clarified in Experiment 4, in which we provided a different explanatory goal (“where to find cars”) and found that even young children have a nuanced understanding of explanatory relevance that is sensitive to differing explanatory goals. Together, these four experiments suggest an early-emerging ability to understand relevance, but a shift between age 4 and 5 in the ability to spontaneously use this understanding when evaluating individual explanations in isolation.

Acknowledgments

We would like to thank the members of the Cognition and Development Lab, particularly Mariel Goddu, for helpful feedback in designing the studies. Additionally, we would like to thank Candice Mills and Amanda Royka for helpful comments on an earlier version of the manuscript. This material is based upon work supported by the National Science Foundation Graduate Research Fellowship under Grant No. DGE-1122492 and by NSF Grant DRL 1561143 to F. Keil.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ For converging results see Allen (2008). We thank Tania Lombrozo for pointing out the relevance of this dissertation to the current work.

² We find the same pattern of results when we analyze the data according to participants’ first binary choice of “helpful” versus “not helpful” for each item, rather than using the full 4-point helpfulness scale.

³ We find the same pattern of results when we analyze the data according to the first binary choice of “helpful” versus “not helpful.”

⁴ As pointed out by an anonymous reviewer, and as highlighted in Experiment 4, some extrinsic explanations are relevant. For example, “cars have gas stations that give them fuel” would be relevant for the explanatory goal of understanding “what makes cars go.” Thus, the pattern of results in Experiment 2 should not be interpreted as indicating a generalizable difference in the helpfulness of “feature” vs. “extrinsic” explanations, but rather should be interpreted as indicating that 4-year-olds have a more fragile ability than 5-year-olds to detect irrelevant explanations, and that this difference is captured by the specific stimulus items we chose.

Additional information

Funding

This work was supported by the National Science Foundation [DGE-1122492,DRL 1561143].