Relationship between magnitude of damage to the hippocampus and impaired recognition memory in monkeys.

Two recent meta-analyses, drawing on data from many of the same studies with monkeys, reached different conclusions about the relationship between hippocampal damage and recognition memory performance. Both studies found evidence of recognition memory impairment following hippocampal damage. However, Zola et al. (J Neurosci 2000;20:451-463) found no significant correlation between extent of hippocampal damage and recognition memory performance, whereas Baxter and Murray (Hippocampus 2001;11:61-71) concluded that the extent of hippocampal damage in monkeys was inversely correlated with impaired performance. Here, we first consider the requirements for carrying out a valid meta-analysis, and point out that the analysis carried out by Baxter and Murray (Hippocampus 2001;11:61-71) is invalid on simple statistical grounds. We then adopt the appropriate statistical procedures (multiple regression analyses rather than simple correlational analysis) to assess the relationship between extent of hippocampal damage and recognition performance across different studies. None of these analyses, including a reanalysis of the data of Baxter and Murray (Hippocampus 2001;11:61-71), revealed a significant inverse relationship between lesion size and behavioral impairment. Most of the variance was explained by differences between the studies that contributed to the meta-analysis, not by lesion size itself. Indeed, analysis of covariance indicated that there were differences among the studies beyond lesion size that significantly affected performance. Finally, we consider what relationship might hold between lesion size and memory performance in the monkey.