Who’s Playing a Trick? Children’s Ability to Detect When Evidence is Being Used Deceptively

• In Event: Poster Session 05
  In Poster Session: PS 05 Topic - Social Cognition

Abstract

Children’s learning in pedagogical contexts is often framed as being guided by an assumption that the communicator is both knowledgeable and honest; that their intent is to choose information that will help lead the learner to an accurate conclusion (e.g., Shafto, Goodman, & Griffiths, 2014). This expectation is not unique to pedagogical contexts or even to young children generally, but rather it is the basis of human communication (Grice, 1975). Nevertheless, communicative intent does not necessitate the use of information that is representative of a truth. Children as young as four understand that to instill particular beliefs in others they must tailor their selection of evidence (e.g., Bass, et al., 2018, Rhodes, Bonawitz, Shafto, Chen, & Caglar, 2015). The current study examines whether children can leverage this understanding to detect others’ intent to (mis)inform.
In this study, children ages 5 to 7 (N=144) first explored a novel toy with two pairs of blocks. After confirming that all blocks activated the toy’s function, children were introduced to a naïve puppet who wanted to learn how the toy works. Children were asked to help the naïve puppet decide between two possible teachers, each of whom created a video demonstration for the naïve puppet to watch; one used a representative sample (i.e., one block of each kind) for demonstration, and the other, a non-representative sample (i.e., two blocks of the same kind). Children then identified which of the two teachers wanted to “play a trick” on the naïve puppet, and which wanted to “help teach” them how the toy works. Then, with respect to each video demonstration, children explained what the naïve puppet would be led to believe were they to watch only that particular video. Finally, children reported their own preference between the two teachers in a hypothetical learning scenario.
Results revealed a marginal main effect of age in predicting attributions of deceptive and helpful intent to the non-representative and representative sampler, respectively, (Wald χ^2 (2)=2.66,p=.101). Follow-up analyses demonstrated that only by age seven did children do so more often than expected by chance, (binomial-test, (two-tailed), 79%,p<.001). Next, the quality-ratings of children’s explanations regarding what the naïve puppet would be led to believe were they to only watch the non-representative and representative sampler’s video demonstrations were examined; there were significant main effects of age and performance in the intentional attribution trials, (non-representative: Wald χ^2 (2)=6.61,p<.05,Wald χ^2 (2)=30.46,p<.001, respectively; See Figure 1), (representative: Wald χ^2 (2)=11.69,p<.01,Wald χ^2 (2)=16.46,p<.001, respectively; See Figure 2). The final phase of analysis revealed a main effect of children’s prior attribution of helpful intent to the representative-sampler, but not of age, on their preference for that teacher in the hypothetical learning scenario, (Wald χ^2 (2)=48.24,p<.001,Wald χ^2 (2)=.86,p=.65, respectively).
These findings suggest that by age 7, children can effectively leverage their ability to reason about what selected information is meant to convey in order to detect when there is an intent to deceive. This research has implications for studying the developmental trajectory of children’s capacity to leverage their evidence-based reasoning abilities for the purpose of combatting misinformation.

Authors

• Karen Carmel Levush, University of Maryland, College Park
  Presenting Author

• Lucas Payne Butler, University of Maryland - College Park
  Non-Presenting Author