About this srcd poster session

---

Panel information
Panel 20. Social Cognition

Abstract

---

People's minds constantly change over time. Effective social reasoning therefore requires representing how mental states change over time. Despite its significance, children's understanding of the dynamics of mental states remains underexplored. Here we investigated one common type of mental state change: knowledge acquisition.

In a pre-registered study (Figure 1A), we investigated whether five- and six-year-olds (N = 52) distinguish agents who gain knowledge from those who lose it. Children were presented with two novel toys and taught which toy was the “Sarn”. Children then observed two-identical agents try to identify which one of the two toys was a “Sarn”. In the forgetting condition, an agent answered this question incorrectly, followed by an identical-looking agent answering correctly (representing knowledge gain). In the learning condition, children saw the reverse pattern (representing knowledge loss).

As predicted, children in the forgetting condition were more likely to infer that there were two agents (76.92%; p<.005 by binomial test). We also predicted that children in the learning condition would find the evidence ambiguous and that their response pattern would be significantly different across conditions, confirmed by a chi-squared test (X2(2)=6.93; p=.03; Figure 1B). We did not find significant age effects in either condition (β>0.35, p>0.1 in both cases). These results suggest that young children have intuitions about how epistemic states change.

Study 1 demonstrates children's expectations of small knowledge changes—knowing and not knowing which object was the Sarn—within a short timeframe. It remains an open question how children's intuitions about knowledge change might differ in scenarios where acquiring knowledge quickly is unlikely.

To explore this question, study 2 (pre-registered; Figure 2A) tested whether children aged 5-6 (N = 82/120) could distinguish between agents who gained knowledge in an "easy learning" condition and those in an "impossible learning" condition. In the "easy learning" condition, the procedure followed the logic from the "learning" condition from Study 1. In the "impossible learning" condition, children were shown eleven novel toys and told which one was the "Sarn." They then observed two identical agents: one who answered the question incorrectly, followed by an identical-looking agent answering correctly.

Children inferred the presence of two agents in the “impossible learning” condition significantly above chance (72.9%; p < .05 by binomial test), while this pattern was not observed in the “easy learning” condition (23.5%; p = 1.00 by binomial test) . The response distribution differed significantly across conditions (X2(2)=20.06; p<.001 by chi-squared test; Figure 2B). No significant age effects were observed in either condition (β>0.32, p>.09 in both cases). This suggests that children believe rapidly acquiring extensive knowledge or losing knowledge quickly is unlikely. Overall, our work suggests that from a young age, children develop specific expectations about epistemic states change.

Author information

Author	Role
Rui Zhang, Yale University	Presenting author
Mackenzie Briscoe, Harvard University	Non-presenting author
Julian Jara-Ettinger, Yale University	Non-presenting author

⇦ Back to session