• Synthesis

• Feedback Cycles

• Dealing with diversity

Talk Less to Strangers: How Homophily Can Improve Collective Decision-Making in Diverse Teams Ill-informed consensus or truthful disagreement? How argumentation styles and preference perceptions affect deliberation outcomes in groups with conflicting stakes

Research on learning in diverse groups has produced conflicting findings: on the one hand, socio-demographic differences between peers can improve learning because they create expectations of underlying cognitive potential and novelty (Levine et al., 2014; Phillips, 2003; Phillips et al., 2006). Other studies find that salient dissimilarity worsens learning and hampers social influence, pointing to the destructive identity processes that visible diversity can trigger (Guilbeault et al., 2018, 2023; Rawlings & Friedkin, 2017; van der Does et al., 2022). Successful learning seems to depend on whether diversity awakens associations of cognitive potential or triggers affective devaluation (Page, 2019). However, this evidence is far from conclusive since studies invoking different findings used different experimental tasks, making it hard to pinpoint conflicting results to the precise factors that caused them. In this study, we aim to isolate and experimentally vary a crucial component that can produce associations of either potential for learning or the devaluation of knowledge from others, namely, the category used for comparisons of (dis-)similarity. We hypothesize that when information about the cognitive characteristics of a source is available, dissimilarity should foster social learning more than similarity. Conversely, an identity-related trait such as political ideology should reinforce learning more when source and receiver are similar rather than dissimilar. We test these hypotheses in an experiment using the established approach-avoid paradigm, in which ‘learning traps’ lead individuals to develop suboptimal solutions (Liquin & Gopnik, 2022; Rich & Gureckis, 2018). We then provide social information from other participants that can help finding the optimal solution, but its use is financially risky and cognitively demanding. A 2x2 experimental design varies whether the social information comes from a source that is either similar or dissimilar to the participant, and whether this type of (dis-)similarity relates to cognitive characteristics or ideological leaning. Only the characteristics of the source are varied across conditions and not the information provided by the source.

How does visible diversity affect social learning in complex tasks?

• -

• Experiment

Yes

• .csv

• Structured

• English

01/07/2024

31/08/2024

USA

01/08/2024

30/09/2024

Nothing to declare.