Three experiments investigated the effects of making errors oneself, as compared to just hearing the correct answer without error generation, hearing another person make an error, or being "on-the-hook," that is, possibly but not necessarily being the person who would be "called-on" to give a response. In all three experiments, generating either an error of commission or generating the correct response, oneself, out loud, compared to being a person who heard another's commission errors (or correct responses), was beneficial for later recall of the correct answer. Experiment 1 suggested that the decrement in recall from self- to other-generation could be partially offset by being "on-the-hook." However, this benefit was fragile and did not hold up either at a delay or when the presence of the other participants was downplayed. The beneficial effect of self-generation, both of correct responses and of errors of commission is consistent with reconsolidation theory. That theory holds that retrieval has a special status as a memory process that renders the retrieved traces labile. If the person was correct, reconsolidating the correct trace results in strengthening. If wrong, the malleability of the retrieved trace implied by reconsolidation theory makes it open to enhanced modification and correction. If the person was not the agent who retrieved, though, such as when someone else retrieves information, or when nothing is retrieved as is the case with omission errors (which we argue is truly how the term "unsuccessful retrieval" should be used), the benefit conferred by the special malleability entailed by the postulated reconsolidation process does not obtain. [At the time of submission to ERIC, this article was in press with "Psychonomic Bulletin & Review."]