Social interactions can shape our memories. The collaborative recognition task is completed in groups of three and consists of three phases. In an individual study phase, some materials are encoded by all participants, some by two participants, and some by single participants only. Then, an interpolated memory test on all studied information is taken either individually – or collaboratively, together with the other group members. Finally, an old/new recognition test on all studied information is completed individually by all participants. Here, it is typically found that prior collaboration enhances old judgments for information initially studied by all group members, but also distorts memory such that information initially studied by other group members only is also more likely to be incorrectly judged as old. It is unclear, however, if these effects of prior collaboration reflect real shifts in memory or can be explained via differences in guessing. Going beyond prior work, in a new experiment, participants provided confidence judgments at final test, allowing examination of ROC curves. Analyses based on the unequal variance signal detection model suggested that prior social interactions increased memory strength (captured by parameter d_a) for initially studied information as well as for information studied by other group members only. Analyses at the group level moreover showed that prior collaboration enhanced mnemonic overlap regarding what specific contents were remembered across group members. Overlap in confidence judgments was centered around high confidence old responses. Overall, the observed effects of social interactions seem to reflect real differences in memory.