First-hand Knowledge Helps 3-year-olds Reinterpret Misinformation From Broken Machines, but not From Deceptive People

• In Event: Cultural Transmission of Information: Evidence of Epistemic Vigilance Across Children, Adults, and Non-human Informants

Abstract

Young children are notoriously bad at correctly interpreting deceptive information (Mascaro & Sperber, 2009). Previous research has shown that children’s ability to correctly interpret deceptive information is influenced by source type (Heyman et al., 2013) and children’s own first-hand knowledge that conflicts with the deceptive information (Jaswal, 2010). How these two factors interact when children are confronted with deceptive information is unclear. Here, two studies explore the effect of different kinds of sources (human vs. machine) and own-knowledge status (having conflicting first-hand information or not) on children’s processing of deceptive information. On 8 trials, 3-year-olds watched as a ball was dropped down one of two transparent criss-crossing tubes and landed in a cup below (e.g., Hood, 1995). Children then heard a deceptive person (Study 1, n = 33) or saw a “broken” machine (Study 2, n = 40) provide incorrect information about the ball’s location, and were asked to search for it. Half of children saw the ball land in one of the cups, which conflicted with the provided information. The other half did not see, and so had no first-hand knowledge about the ball’s location.

In Study 1, children with knowledge searched in the correct location 47.8% of the time, which did not differ from a hypothetical chance distribution, λ2 (1, N = 136) = 0.13, p = 0.716. Children without knowledge searched correctly only 30.5% of the time; in other words, they searched where the deceptive person indicated, λ2 (1, N = 128) = 10.15, p = 0.001. They also searched correctly less frequently than children who had knowledge, λ2 (1, N = 128) = 8.29, p = 0.005. Having first-hand knowledge about the ball’s location prevented children from blindly believing the deceptive person’s claims, but it did not help them develop a successful search strategy.

In Study 2 a “broken” machine replaced the deceptive person to explore whether children’s expectation that humans are typically veridical is the reason they cannot correctly interpret deception. This time, children with knowledge searched correctly 69% of the time, more than would be expected by chance, λ2 (1, N = 168) = 12.65, p = 0.0003 and more frequently than children without knowledge, λ2 (1, N = 320) = 28.20, p < 0.001, who only searched correctly 39.5% of the time, which did not differ from what would be expected by chance, λ2 (1, N = 152) = 3.41, p = 0.065.

Combining studies, the probability of children’s correct searches as a function of source type (Study 1/human vs Study2/machine) and knowledge status (knowledge vs no-knowledge) was estimated using a binary logistic regression. The regression indicated a significant main effect of source (p = 0.002) and a significant main effect of knowledge state (p < 0.001), but no interaction between the two (p = 0.09). These results suggest that children’s difficulty with deceptive information stems in part from their bias to expect truthful/honest information from humans. Moreover, this bias is so strong that having first-hand knowledge is not enough to override it.

Authors

• Carolyn M Palmquist, Amherst College
  Presenting Author

• Robyn Lindsey Kondrad, James Madison University
  Non-Presenting Author