Does collaboration improve or impede finding remote associations?

Across two experiments using the Compound Remote Associates Test (cRAT), we examined whether collaboration improves or impedes the ability to find remote associations. Participants worked either jointly in dyads or in parallel. In both experiments, joint dyads solved fewer problems and required more time to reach correct solutions, indicating a consistent performance cost of collaboration. In Experiment 1, joint dyads also generated fewer solution attempts and showed greater semantic similarity between consecutive ideas, consistent with the hypothesis of collaborative fixation. In Experiment 2, we removed verbal interaction while maintaining shared access to each other's solution attempts. Performance was still lower in the joint condition, but evidence for semantic fixation was no longer observed. Our findings indicate that collaboration may, in fact, impede rather than improve the creation of remote semantic links.