Preschool Children Use Mentalising Inferences in the Process of Selective Trust

• In Event: Children’s selective trust across domains, agents, and cultures

Abstract

Learning from others grants children access to knowledge that would otherwise be unavailable to them. However, it also leaves them vulnerable to misinformation and manipulation. This poses a great challenge starting in early childhood as the trustworthiness of informants often requires nonobvious inferences about their intentions. Prior research has revealed an age-related development in children’s inferences about intentions when deciding whom to trust. By the age of five, children reason about mental states underlying accurate or inaccurate outcomes (Liu et al., 2013; Nurmsoo & Robinson, 2009a; 2009b).
What is unclear is whether children consider the reason behind the outcome of an informant’s advice when deciding whether to trust them. This is especially the case in a deceptive context. For example, a positive outcome of accepting an informant’s advice reflects an honest intent while a positive outcome of rejecting an informant’s advice reflects a deceptive one. Thus, our primary interest was to examine preschool children’s inference of intentions based on how the outcome was achieved.
In addition to this primary interest, we were interested in whether children generalise their inferences when provided with different types of information. Previous work suggests that they are more likely to generalise their inferences when faced with past semantic inaccuracies than episodic ones (Stephens & Koenig, 2015). However, it remains unknown whether children make the same pattern of generalisation when faced with misinformation from a deceptive informant.
Across two studies, 3- to 6-year-old children observed four informants interact with their opponents in a 2 (Opponent’s Decision: accept versus reject) ✕ 2 (Outcome: win versus lose) within-subjects design. Informants suggested to their opponent the location of a sticker in Study 1 (i.e., episodic context) and the name of a novel object in Study 2 (i.e., semantic context). The opponent accepted or rejected the informant’s advice, leading to a win or lose outcome. Thus, children had to integrate cues of the opponent’s decision with the outcome to infer the informant’s intention. They were then tested on an object location task and an object naming task (Stephens & Koenig, 2015).
Study 1 (N = 43) results showed that children trusted the informants in the accept-win and reject-lose conditions more than those in the accept-lose and reject-win conditions in the object location task. However, there were no statistically significant differences in the object naming task. This suggests that children consider the reason behind the outcome of informants’ advice when deciding whom to trust, but may not generalise their inference of intent when presented with episodic misinformation. Study 2 (N = 73) extended Study 1’s results with an age effect: older children trusted the informants in the accept-win and reject-lose conditions more than those in the accept-lose and reject-win conditions in the object naming task. Additionally, older children made the same distinction between the informants in the object location task. Study 2 showed that children generalise their inference of intent when presented with semantic misinformation.
Overall, our findings reveal that children can use mentalising inferences in the process of selective trust.

Authors

• Nina Ni Ye
  Presenting Author

• Xiao Pan Ding, National University of Singapore