Metacognitive Confidence and Communication in Dyadic Decision-Making

Abstract

This thesis investigated the psychological traits and communicative factors that shape collaboration in dyadic decision-making. Study 1 explored the “two heads are better than one” effect in a dynamic driving simulation with changing operational conditions. The study found that communication quality, but not quantity, predicted dyadic accuracy, and that dyads outperformed individuals but only under specific conditions. Study 2 examined the emergence of collective intelligence in dyads completing static tasks guided by the Cattell-Horn-Carroll model of individual intelligence. The results challenged claims that collective intelligence is primarily shaped by equal participation in discussions, gender composition, and social sensitivity, and instead highlighted the central roles of individual cognitive ability and metacognitive confidence. Latent profile analysis identified distinct dyad types, characterised by unique psychological profiles and performance patterns. Study 3 extended the confidence theory, which proposes that groups share and use confidence to guide their decisions, by examining how individual differences in trait-confidence and communication modality interact to shape dyadic processes and outcomes. Results showed that dyads with low-trait confidence improved equally under passive (visual information sharing only) and active communication (verbal discussion), whereas those with mixed-trait or high-trait confidence benefited most from active communication. Additionally, dyad members’ confidence ratings became more aligned (i.e., confidence matching) under both passive and active communication, but this alignment predicted accuracy improvements only in the passive communication condition. Together, these three studies demonstrate that no single factor guarantees collaborative success. Instead, dyadic performance is shaped by a dynamic interplay between individual differences in trait confidence, communication type and context, and task structure.