Introduction: Patients with schizophrenia exhibit deficits in memory performance, particularly when required to bind together disparate items (associative memory). Yet the literature on associative memory is decidedly mixed, with some studies showing large deficits and other showing none.
Methods: The aims of this meta-analysis were to determine an overall effect size for the associative memory deficit in patients with schizophrenia and to examine two potential moderating variables related to this impairment: the nature of the memory being tested (pair vs. source recognition) and the inclusion or exclusion of novel items as part of the recognition test.
Results: We found that the mean effect sizes were large for pair recognition (r=.50) and medium for source recognition (r=.29), with a significant difference between the two recognition types. Contrary to a priori hypotheses, there were no differences in the effect sizes across the various types of source memory (i.e., internal, external, or reality monitoring). There was, however, a significant difference in the effect sizes between those studies that included novel items as part of the memory test (r=.26) and those that did not (r=.44).
Conclusion: These findings suggest that the associative memory deficit in schizophrenia is not specific to self/other distinctions, but is rather a more global effect seen across testing conditions. In addition, memory tests that do not include new items appear to maximise this effect, perhaps by removing a potential response outlet for subjects who lack confidence in the accuracy of their memory performance.