About this paper symposium

---

Panel information
Panel 20. Social Cognition

Paper #1
How Children Incorporate Belief Strength into Intuitive Theories of Belief Change
Author information	Role
Pearl Han Li, University of Wisconsin-Madison, United States	Presenting author
Laura Soter, York University, Canada	Non-presenting author
Tamar Kushnir, Duke University, United States	Non-presenting author
Abstract
Models of rational belief revision incorporate two components: strength of initial belief, and strength of new evidence. Prior work shows that young children are often rational revisers when learning about the world: they integrate novel information proportional to the evidence and their prior belief strength (Kushnir & Gopnik, 2007; Gopnik & Wellman, 2012; Tenenbaum et al., 2011). But do children understand the roles of prior belief strength and evidence when thinking about how other agents update their beliefs? We ran two preregistered studies to test this model by examining children’s understand of the role of prior belief strength and how prior beliefs interact with evidence in belief change. In Study 1, we explore whether U.S. 4-8-year olds (N=100; Mage= 6.52; SDage= 1.44; Range = 4.00-8.98) think it is more possible and difficult for someone to change a strongly held belief versus a weakly held one. Children were presented with two characters who held the same belief about various topics (facts, norms, or preferences)—but one character held the belief a lot, and the other only a little. All beliefs used novel words (e.g., “Noah believes daxing is wrong.”). Children then judged whether it was possible for each character to change their belief, and how difficult (“hard”) it would be to change their belief. Even 4-year-olds thought strong and weak belief change was possible. By age 5, children said strong beliefs were more difficult to change than weak ones; this differentiation increased with age (𝛽 = 0.27, CI = [0.18, 0.35], p <.001). This pattern held across belief-type, suggesting that children understood the role of belief strength as a general feature of belief change. In Study 2, we tested whether children use information about prior belief strength to make predictions about whether someone would change their belief upon encountering counterevidence. We again presented 4-8-year olds (Ncurrent= 88, Mage= 6.54; SDage= 1.3; Range = 4.12- 8.99, full sample expected October 2024) with two characters who held the same belief strongly or weakly. The characters then encounter counterevidence to their belief: a group of peers say the belief is false. We asked children whether the character would keep or change their initial belief. Preliminary results show that children judged weakly held beliefs as more likely to change than strongly held beliefs, and this differentiation increased with age (OR =.66, CI = [0.49, 0.9], p =.01). This differentiation with age was more pronounced for descriptive beliefs (e.g., how many fish in a river) than evaluative beliefs (e.g., if daxing is okay, OR =.72, CI = [0.54, 0.97], p =.03); this may reflect differences in children’s theories about whether testimony is good counter-evidence across domains. Together, these results suggest that even very young children incorporate information about someone’s initial belief strength into their reasoning about whether others can and will change their beliefs–both when thinking about how difficult or possible belief change is, and when predicting how someone will respond to new evidence. These findings offer evidence that children’s intuitive theories of belief change may be consistent with rational models of belief updating.

Paper #2
Children’s and adults’ beliefs about the benefits of talking to someone with a different belief
Author information	Role
Dr. Kirsten H. Blakey, University of Toronto, Canada	Presenting author
Ashley Ransom, University of Toronto Mississauga, Canada	Non-presenting author
Samuel Ronfard, University of Toronto Mississauga, Canada	Non-presenting author
Abstract
Increased societal polarization and the human tendency to confirm one’s biases (Nickerson, 1998; Stanovich et al., 2013) means that we often dismiss views that contradict our own. This is unfortunate because disagreement can be beneficial for learning. For instance, disagreement reduces overconfidence and encourages exploration in 4- to 6-olds (Langenhoff et al., 2024) and promotes critical thinking in adolescents (Kuhn, 2019) and adults (Mercier & Sperber, 2011). However, to benefit from disagreement, one must be willing to seek out individuals with different beliefs and to do so with the intention to learn from them. Do 4- to 8-year-old children believe that someone will learn more about a topic by talking to a person who disagrees with them rather than agrees with them? Participants (4- & 5-year-olds, N = 50; 7- & 8-year-olds, N = 50; and adults, N = 100) watched videos that featured three children. In each video, a target child had a belief that they wanted to learn more about. One child agreed with the target child’s belief, and the other child disagreed with the target child’s belief. Participants indicated who the target child should talk to if they wanted to learn more — the agreeing child or the disagreeing child. We manipulated the type of belief (factual, moral, preference, ambiguous) the children had. On factual trials, the target child wanted to learn more about a factually correct belief (e.g., that dinosaurs lived a long time ago). On moral trials, the target child wanted to learn more about a conventional moral belief (e.g., that it is wrong to steal). On preference trials, the target child wanted to learn more about a personal preference (e.g., what is the tastiest fruit). On ambiguous trials, the target child wanted to learn more about an ambiguous fact (e.g., whether or not an object would float). Participants saw each trial type 4 times for a total of 16 trials. We examined participants’ responses with a multilevel binomial logistic regression that included choice on each trial (agreeing child, disagreeing child) as the dependent variable. There was a main effect of age group on participants’ choices, Figure 1. On average, adults chose the disagreeing child on 61% of trials, which was significantly more often than 4- & 5-year-olds (39%) or 7- & 8-year-olds (39%). There was also a main effect of belief type on participants’ choices. Participants chose the disagreeing child more often on ambiguous (68%) and preference (64%) trials than on factual (30%) or moral (35%) trials. Finally, there was an interaction between age group and belief type. Age was a significant predictor of choice on ambiguous, moral, and preference trials, but not for factual trials. In sum, we find that adults value learning from someone who disagrees more than children do but that the type of belief influences both adults’ and children’s choices. For children and adults, the value of disagreement increases when they do not think that a correct answer exists because evidence is lacking or because there of a lack of consensus.

Paper #3
Children's developing ability to adjust their beliefs reasonably in light of disagreement
Author information	Role
Antonia Frederike Langenhoff, Stanford University, United States	Presenting author
Jan M Engelmann, University of California, Berkeley, United States	Non-presenting author
Mahesh Srinivasan, University of California, Berkeley, United States	Non-presenting author
Abstract
The history of science is often framed as a series of disagreements resolved through empirical research and theory revision (e.g., Kuhn, 1962). Beyond benefiting science as a whole, constructivist learning theorists have long argued that disagreement also offers significant advantages for individual learners: engaging with conflicting viewpoints is believed to challenge learners’ existing knowledge and beliefs, which, in turn, fosters critical thinking and deeper understanding (e.g., Piaget, 1952). Despite this long-standing interest, exactly how disagreement promotes learning is still not well-understood. The current work begins to test the idea that even young children intuitively understand that empirical disagreements often stem from differential access to information and can be resolved by additional empirical evidence, and thus respond to them with enhanced information-seeking, exploration, and learning. Across two preregistered studies (N=218), we tested 4- to 6-year-olds and 7- to 9-year-olds (and adults for comparison) in an interactive online paradigm, in which they had to determine where a pet was hiding. Participants first acquired evidence by asking informants, and were then asked to state their belief about where the pet went. Next, participants learned that another agent—who had also consulted informants on the whereabouts of the pet—disagreed with them, and we participants' belief was assessed for a second time. Between conditions, we varied whether the evidence supporting participants' belief was 1) stronger than, 2) weaker than, or 3) equal to the evidence supporting the belief of the disagreeing agent (see Figure 1). We were interested whether even young children would adopt the disagreeing other’s belief when their own evidence was weaker, maintain their initial belief when it was stronger, and suspend judgment before making a final decision when both had equally strong evidence (see Frances, 2014). Findings showed that all age groups responded to disagreement in these rational ways (ps < .01), although participants’ responses became more nuanced with age. Notably, 4- to 6-year-olds suspended judgment reliably only when this was operationalized as an actual opportunity to search for additional information (Study 2; see Figure 2), but not when it required them to explicitly state their uncertainty (Study 1). Our findings are in line with other recent work showing that experiencing disagreement prompts children’s information-seeking and belief revision (e.g., Langenhoff et al., 2024; Ronfard et al., 2018), and suggest that already young children might view empirical disagreements as indicators of epistemic ambiguity and opportunities for learning. In ongoing work, we are investigating whether the effects of disagreement on children’s information-seeking and belief revision translate into observable learning gains, and whether disagreement has stronger effects on learning than other (e.g., non-interactive) forms of being presented with conflicting hypotheses. In future work, we additionally hope to test whether these effects generalize to different learning mechanisms (e.g., exploration, hypothesis-testing). Implications of these findings for education and learning will be discussed.

Paper #4
How Parents' Testimony affects Children’s Formation of Negative Bias about Social Groups
Author information	Role
Ms. Yeon Ju Suh, University of Minnesota, Twin Cities, United States	Presenting author
Charisse B. Pickron, University of Minnesota, Twin Cities, United States	Non-presenting author
Melissa A. Koenig, University of Minnesota, Twin Cities, United States	Non-presenting author
Abstract
There is a critical need to evaluate how biased beliefs against other social groups are learned in early childhood. This study focused on two mechanisms of social learning—parents’ direct and indirect testimony—in how they can lead 4-to-7-year-old children to endorse and transmit negative biases against novel social groups. In a pre-registered study, we hypothesized that: 1) parents’ direct and indirect testimony will both lead children to endorse negative beliefs and attitudes towards a novel social group (“Zorches”), 2) children in the direct testimony condition will transmit negative information more than children in the indirect testimony condition, and 3) these negative beliefs will persist over 2-weeks, especially for older children. A sample of 4- to 7-year-olds (N=110; Mage = 5.978, 52.7% Female, 85.5% White) and their primary caregiver were invited to the lab where they learned four negative pieces of information about Zorches, either through parents’ direct testimony (following a brief script) or through indirect testimony (pointing in response to experimenter questions). Children were asked to endorse parents’ testimony both verbally and non-verbally, to evaluate their moral worth, and their capacity for friendship. Children were also invited back (via Zoom) after 2-weeks and asked the same questions to assess how their beliefs about Zorches may have changed. To assess our first hypothesis, we found that children who learned through direct testimony endorsed negative beliefs towards Zorches more than those who learned through indirect non-verbal testimony (verbal endorsement (t(67) = 5.068, p < .001). In addition, we found a significant interaction between age and condition for children’s verbal and non-verbal endorsements: In the direct testimony condition, children increasingly endorsed negative beliefs towards Zorches with age, whereas in the indirect testimony condition children decreased their endorsement of negative beliefs with age (see Figure 1 for verbal endorsement (F(3, 78) = 12.95, p <.001); see Figure 2 for non-verbal endorsement (F(3, 91) = 5.79, p = 0.001). Further, children who learned through direct testimony more negatively evaluated the group’s moral worth (t(118) = 6.309, p <.001) and their capacity for friendship (t(111) = 5.342, p <.001) than children who learned through indirect testimony. For our second hypothesis, we found significant condition differences in the valence of information children transmitted such that children in the direct testimony condition shared negative information whereas children in the indirect condition shared more positive or neutral information (Χ²(1) = 27.899, p < .001). Lastly, for our third hypothesis, children’s in-lab verbal and non-verbal endorsement scores predicted their verbal (F(4, 47) = 28.61, p<.001) and non-verbal endorsements (F(4, 73) = 21.41, p <.001) two weeks later, demonstrating that their initial judgment of Zorches have persisting effects over time. Findings underscore the harmful and long-term effects of explicitly biased statements from parents. Furthermore, younger children seem especially vulnerable to both direct and indirect testimony. Findings will be discussed in terms of potential sources of risk and resilience towards parents’ biased beliefs when learning through testimony.

⇦ Back to session