Delay-dependent contributions of medial temporal lobe regions to episodic memory retrieval

The medial temporal lobes play an important role in episodic memory, but over time, hippocampal contributions to retrieval may be diminished. However, it is unclear whether such changes are related to the ability to retrieve contextual information, and whether they are common across all medial temporal regions. Here, we used functional neuroimaging to compare neural responses during immediate and delayed recognition. Results showed that recollection-related activity in the posterior hippocampus declined after a 1-day delay. In contrast, activity was relatively stable in the anterior hippocampus and in neocortical areas. Multi-voxel pattern similarity analyses also revealed that anterior hippocampal patterns contained information about context during item recognition, and after a delay, context coding in this region was related to successful retention of context information. Together, these findings suggest that the anterior and posterior hippocampus have different contributions to memory over time and that neurobiological models of memory must account for these differences. DOI: http://dx.doi.org/10.7554/eLife.05025.001

eLife digest In 1953, an American man called Henry Molaison underwent surgery to remove the medial temporal lobes of his brain in an effort to cure him of severe epilepsy. After the surgery, his epilepsy was indeed improved, but he was left without the ability to form new memories. His case is now seen as one of the first demonstrations of the medial temporal lobes being involved in memory. Beneath the surface of each medial temporal lobe is a structure called the hippocampus, which is essential for the formation of new memories. However, memories are not stored permanently within the hippocampus: instead they are transferred ultimately to the neocortex, which is the outer layer of the brain. Some neuroscientists believe that the content of memories remains unchanged during this transfer, whereas others argue that memories are stripped of their context—that is, details of when and where they were acquired—before they reach the neocortex. In a brain imaging experiment, Ritchey et al. have now attempted to distinguish between these two possibilities. Volunteers were asked to memorize sentences linking an object to a room, such as ‘the apple is in the bedroom’, on two occasions 24 hr apart. Immediately after the second session, the volunteers were asked to complete memory tests while lying in the brain scanner. In one test the volunteer was shown a list of objects and asked to identify those objects they could recall seeing in either of the training sessions, and to identify objects they recognised as familiar, even if they could not specifically remember seeing these objects during training sessions. Analysis of the brain imaging data revealed that regions of the medial temporal lobes were more active when the volunteers recalled objects than when they recognised them as familiar. Moreover, for the ‘recall’ responses—in which the volunteers could still retrieve contextual information—the activity of the hippocampus depended on the age of the memories. The anterior (front) part of the hippocampus was active when subjects recalled either new memories or memories from 24 hr previously, whereas the posterior (rear) hippocampus was active only during the recall of new memories. In addition, patterns of activity observed in the anterior hippocampus could be used to determine which room was previously associated with the object. This suggests that contextual information is retained in the anterior hippocampus, but lost from the posterior hippocampus over time. Overall the results of Ritchey et al. highlight the fact that individual components of the medial temporal lobes, including hippocampal subregions, have distinct roles in the storage of memories, with these roles also changing over time. Moreover, the data lend support to the idea that contextual information may be lost from memories before they reach the neocortex. DOI: http://dx.doi.org/10.7554/eLife.05025.002