When social influences reduce false recognition memory: A case of categorically related information.

Social interactions create opportunities for reminiscence and memory rehearsal but can also lead to memory errors. We tested how the type of information people remember can influence the magnitude of memory errors they make following collaborative discussion. Past findings show that unrelated item lists and emotional salient items reduce false alarms and improve memory discrimination, respectively, on an individual recognition test after collaborative discussion compared to no prior collaboration. In contrast, for associatively related materials with high relatedness (e.g., bed, rest, awake, tired, dream, etc.) collaboration increases false recognition memory for the critical lures (e.g., sleep) on a later individual test. We tested whether the error-pruning benefits of collaboration are restricted to unrelated and emotional information or can also occur for other classes of related information that produce high memory errors. Using categorized stimuli, we created conditions that produced high or low memory errors for the same targets (12 versus 2 target exemplars per category across study lists of equal length). Replicating past research, collaboration increased the accuracy of recognition memory and large category size decreased it. The critical novel finding showed that collaboration pruned individual recognition errors by reducing false alarms not only in the low memory error condition but also the high memory error condition. This study delineates the conditions where collaboration can prune memory errors for related information.