About this paper symposium

---

Panel information
Panel 20. Social Cognition

Paper #1
The Lifespan Development of Biased Information-Seeking: Myside Bias
Author information	Role
F. Ece Özkan, University of Toronto, Canada	Presenting author
Samuel Ronfard, University of Toronto, Canada	Non-presenting author
Abstract
The Argumentative Theory of Reasoning (ATR) argues that the primary function of reasoning is to justify ourselves and evaluate others (Mercier & Sperber, 2011; 2017). According to the theory, when we evaluate reasons provided by others, we are demanding; however, when we produce reasons, we show a myside bias, e.g., we support our perspective while discounting other perspectives (Trouche et al., 2016; Mercier, 2022). ATR predicts that we should observe a myside bias when we are producing reasons in an argumentative setting. But what about the pre-argumentation stage? Are we also biased when we gather evidence before engaging in dialogic argumentation? To find out, we tested 480 5-year-old, 8-year-old, 11-year-old, 14-year-old, 18-24-year-old, and 25-55-year-old (n=80 per age group) English speakers based in Canada (ethnically diverse sample). To manipulate the argumentative context, children were randomly assigned to a learn or justify condition. In the learn conditions, their task was to seek information to see if they selected the correct path. In the justify conditions, we told them that after playing the game, they needed to convince another player that they had selected the correct path. To assess whether knowing that someone else made a different choice made participants less or more biased in their information search, we randomly assigned participants to conflict or no conflict conditions. In the conflict conditions, we told participants that the other player selected the other path. In the no-conflict conditions, there was no conflicting opinion. In sum, each participant was randomly assigned to one of the four conditions: learn, learn+conflict, justify, justify+conflict. Their task was 1) to predict which of two paths a character took to go to a city and 2) to choose positive or negative information about these paths. In the information-seeking phase, participants were shown eight envelopes: 4 envelopes about the one path and four envelopes about the alternate path. Among those, half contained positive information, and half contained negative information (the envelopes were color-coded and contained thumbs-up/down). The participants were told to select 1 to 7 envelopes. Participants’ myside score was computed as the total number of positive information about the selected path (confirming evidence) and negative information selected about the opposite path (counterevidence) divided by the total number of selections the participant made. We found that children as young as five years showed a myside bias but only in the justify conditions. We also found that this myside bias increases with age in the justify conditions but does not increase in the learn conditions. Additionally, we showed that the participants’ selections of counterevidence increased with age, while their tendency to select only confirming evidence decreased with age, supporting the claim that looking for counterevidence is a later developing ability (Kuhn & Crowell, 2011; Domberg et al., 2018). These findings extend the scope of myside bias to the evidence-gathering stage, confirm that it is restricted to argumentative contexts, and for the first time depict its lifespan development.

Paper #2
Group Membership Biases Children’s Evaluation of Evidence
Author information	Role
Joshua Allan Confer, University of California, Berkeley, United States	Presenting author
Hanna Schleihauf, Utrecht University, Netherlands	Non-presenting author
Dorsa Amir, Duke University, United States	Non-presenting author
Jan M. Engelmann, University of California, Berkeley, United States	Non-presenting author
Abstract
How do children learn about the world? A prominent theory characterizes children as ‘little scientists’ who are open-minded, flexible, curious learners (Gruber, 1973; Carey, 1986; Gopnik, 1994). Indeed, children correctly form beliefs based on the available evidence (Gopnik et al., 2001; Kushnir & Gopnik, 2005); revise their beliefs if presented with counterevidence (Lucas et al., 2014; Kimura & Gopnik, 2019); and seek out more evidence if the evidence they have is inconclusive (Gweon & Schulz, 2011; Bonawitz et al., 2012). Yet, previous research in this area has largely taken place in individual settings. In the current project, we examine children’s epistemic practices in a social context. Across two pre-registered studies (N=138), we tested how belonging to a social group influences 4- to 6-year-olds’ evaluation of evidence when forming (Study 1) and revising their beliefs (Study 2). In both studies, children were first assigned to one of two groups (Group Condition), or simply introduced to two groups (No Group Condition). Then, children had to determine the correct response in a reasoning task (e.g., whether there were more elephants or lions in a zoo). Before being exposed to the available evidence, children learned what both groups believed about the answer. Study 1 focused on belief formation. After hearing both groups’ beliefs, children could inspect as much evidence as they wanted, one at a time, until they reached a conclusion. The first pieces of evidence appeared to support the ingroup belief (for children in the Group Condition) or the first group’s belief (for children in the No Group Condition). However, in total, the evidence supported the other team’s belief in both conditions. As predicted, children tested fewer pieces of evidence to reach a conclusion if they belonged to a group (p<.01). This led children in the Group Condition to form the correct belief about the reasoning task less than half as often as children in the No Group Condition (p<.01). Study 2 focused on belief revision. After hearing both groups’ beliefs, children committed to a belief themselves and rated how confident they were. As pre-registered, in the Group Condition, we focused our analysis on children who adopted the group belief (94% of children). Then, children in both conditions were shown two pieces of counterevidence. As predicted, we found that children maintained a higher level of confidence in their initial belief if they belonged to a group (p<.01). This led children in the Group Condition to revise their beliefs about the reasoning task less than half as often as children in the No Group Condition (p<.01). These findings suggest that from a young age, children’s epistemic practices are sensitive to the social context they are in. Specifically, they demonstrate that even preschoolers bias their standards of evidence to believe what their group believes, resulting in children forming and holding onto inaccurate beliefs. These results provide insights into the developing cognitive mechanisms behind belief formation and revision. They also have important implications for interventions aimed at reducing the spread of misinformation and fostering intellectual humility.

Paper #3
Explaining Outdoes Evidence in Supporting Belief Revision
Author information	Role
Anahid S. Modrek, CSUSB, United States	Presenting author
Deanna Kuhn, Columbia University, United States	Non-presenting author
Abstract
Both explanations and evidence contribute to belief revision, leaving it arguable what their distinctive roles are. Indeed, the longstanding emphasis is on guiding people to pay attention to evidence to further support their conclusions. Beyond this are known advantages of explanations in aiding learning and understanding (Chi et al., 1989). A strong narrative, such as an explanatory account, is beneficial as a vehicle for assembling facts (Pennington & Hastie, 1992; Kuhn et al., 1994; Mandler & Johnson, 1977; Stein & Nezworski, 1978). An explanatory account of an event sequence can guide attention to relevant evidence that otherwise would escape attention (Legare & Lombrozo 2014; Legare, Gelman, & Wellman, 2010; Walker et al., 2014). However, there is a line of work suggesting explanations have a downside (Walker et al., 2017; Kuhn & Katz, 2009; Williams, Lombrozo, & Rehder, 2013; Legare & Lombrozo, 2014). That is, articulating and elaborating a theoretical explanation that one becomes increasingly committed to runs the risk of heightened belief and overgeneralization (Walker, Lombrozo, & Gopnik, 2014). A further consequence may be less receptivity to new evidence (Kuhn & Katz, 2009). So, do explanations support or impair belief revision? Given these mixed findings, we undertook the present work with the goal of better understanding how explanation and evidence interact in inquiry learning. To do so, we examined a domain in which young adolescent participants were likely to have existing ideas regarding the effects of different variables on an outcome, such as what factors contribute to a more successful summer camp (pool vs. lake, hiking route vs. running path, etc.). Across three experiments (n=155) participants ages 12-17 were provided data that could help them ascertain whether their current beliefs were supported by this evidence or warranted revision. After eliciting their beliefs, we made available to participants data contradicting some of these beliefs. Participants were then randomly assigned to three conditions: In Condition 1 (pre- and post-explanation; PPE), explanations were solicited prior to and following data investigation. In Condition 2, (post-explanation only; PE) explanations were solicited only following data investigation. In Condition 3 (no explanation; NE), no explanations were solicited. Participants were given ample time over multiple sessions to examine the data, with no instructions or constraints imposed. In a final session, they reported their conclusions and the basis for them. Successfully replicated across the three studies was an effect of experimental conditions. Explanations facilitated belief revisions in PPE and PE conditions compared to the NE condition. Only 10% of participants, or less, applied the highly regarded control of variables (COV) strategy. Instead, they used the error-prone strategy of examining data broadly to identify favorable outcomes and then implicating antecedents often co-occurring with them. Accurate belief revision nonetheless occurred, slightly more successfully among the younger sample, who made more frequent reference to the data in a final justification of their conclusions, solicited at the end of the multi-day sessions. Overall, explaining supported belief revision in young adolescents.

⇦ Back to session