The roles of the hippocampus in recognition memory.

摘要

Recognition is our awareness of prior experience and is thought to depend upon the recollection and familiarity processes. Recollection refers to remembering an experience with associated contextual details, whereas familiarity refers to a sense of awareness absent recollection. Some psychological models have proposed that when recollection is available, confidence in a recognition decision is categorically high. Other models propose that recollection and familiarity are both continuous processes and, therefore, differences in memory strength per se do not separate recollection from familiarity. Functional magnetic resonance imaging (fMRI) has been used extensively to examine the neural correlates of recollection and familiarity, yet interpretation of these data and how recognition is supported by the brain remain open debates. The two dual-process views make different interpretations about whether the fMRI results suggest a division of labor between medial temporal lobe (MTL) structures. The typical conclusion in these fMRI studies is that the hippocampus selectively subserves recollection. In this dissertation, Experiments 1 and 2 tested the hypotheses of both dual-process models (i.e., whether recollection is categorical or continuous). The results from both experiments showed that highly confident memory is not categorically based on recollection. Experiment 3 performed a meta-analysis on the relevant fMRI literature and considered the two model-based interpretations of the results. This analysis showed that although these studies intended to dissociate MTL activity on the basis of recognition processes, their interpretations dissociated activity on the basis of memory strength. Experiment 4, which scanned a source memory test during retrieval, took a new approach to compare correct source judgments (recollection) and incorrect source judgments (familiarity) while equating for old/new memory strength by including only high-confidence hits in the analysis. This approach avoided the complication from previous studies (identified in Experiment 3) that confounded recollection and familiarity with strong and weak memories. After equating for memory strength, activity in the hippocampus in association with high-confidence hits was greater than for forgotten items, regardless whether source judgments were correct (recollection succeeded) or incorrect (recollection failed).;The conclusion in this dissertation, drawn from the results of the four experiments, is that the hippocampus serves a broader role than selectively supporting recollection. Two views are discussed about how this role may work. One view hypothesizes that the summation of MTL input associated with item identification, contextual information and arousal is mediated by the hippocampus. In this framework, the role of the hippocampus is to enhance the retrievability of salient experiences. Another view suggests that the hippocampus abstracts bits of information from prior experience. This role does not map onto specific recognition processes (i.e., recollection and familiarity) or measurements of memory strength. Further work is needed to examine the range of hippocampal sensitivity for memory strength. Taken as a whole, these developments elucidate a critical role for the hippocampus in recognition and not solely in recollection.