Shift from hippocampal to neocortical centered retrieval network with consolidation

Abstract

The standard model of system-level consolidation posits that the hippocampus is part of a retrieval network for recent memories. According to this theory, the memories are gradually transferred to neocortical circuits with consolidation, where the connections within this circuit grow stronger and reorganized so that redundant and/or contextual details may be lost. Thus, remote memories are based on neocortical networks and can be retrieved independently of the hippocampus. To test this model, we measured regional brain activity and connectivity during retrieval with functional magnetic resonance imaging. Subjects were trained on two sets of face-location association and were tested with two different delays, 15 min and 24 h including a whole night of sleep. We hypothesized that memory traces of the locations associated with specific faces will be linked through the hippocampus for the retrieval of recently learned association, but with consolidation, the activity and the functional connectivity between the neocortical areas will increase. We show that posterior hippocampal activity related to high-confidence retrieval decreased and neocortical activity increased with consolidation. Moreover, the connectivity between the hippocampus and the neocortical regions decreased and in turn, cortico-cortical connectivity between the representational areas increased. The results provide mechanistic support for a two-level process of the declarative memory system, involving initial representation of new associations in a network including the hippocampus and subsequent consolidation into a predominantly neocortical network.

Figure 1.

Experimental design. A, Experimental procedure. On day 1, the subjects were first familiarized with 60 faces (set A, remote). In the subsequent learning session, subjects underwent three cycles of encoding (passive viewing of the face being associated to 1 of the 6 locations, once each for all 60 face–location pairs) and retrieval (retrieval of the correct location to the cued face without any feedback), all in randomized order. On day 2 (24 h later), the subjects performed the same procedure of face familiarization and three cycles of encoding and retrieval, with a new set of 60 face stimuli (set B, recent). After a break of ∼15 min, the subjects underwent a final retrieval test (Test session) where all the associations of both remote and recent faces (sets A and B) were tested in randomized order. B, Timing of a trial at retrieval block: The face cue and the six location probes appeared on the screen for 4 s in which the subjects were instructed to respond to the correct location using a joystick controlled cursor. This was followed by a confidence rating on a scale of 1 = unsure to 5 = sure. The pink circle on the picture indicates the joystick controlled cursor and the green arrow indicates the movement of the cursor to either the selected location or the confidence rating (green arrow did not appear on screen).

Figure 2.

Brain areas that support retrieval of consolidated and unconsolidated memory traces. A, Increase of activity with consolidation was found in a distributed cortical and subcortical region. Significant clusters are superimposed on a template 3D brain. B, Decrease of activity with consolidation was observed bilaterally in the posterior tail of the hippocampus. Significant clusters are superimposed on a coronal slice and on an intensity projection map viewed from the top of the head. Activations are depicted at p < 0.001 voxel level uncorrected and cluster size that exceeds 100 voxels, showing the significant clusters. L, Left; R, right.

Figure 3.

Networks that support retrieval of unconsolidated and consolidated memory traces. A, Functional connectivity (PPI) with the posterior hippocampus. The connectivity to the early visual areas, extending to FFA, bilaterally and PPC bilaterally decreased with consolidation. B, Functional connectivity with FFA. Increase with consolidation was observed to the early visual areas bilaterally and the left PPC extending to the motor cortex (top). Decrease with consolidation was observed in the left anterior hippocampus, the left parahippocampal gyrus and the right inferior to middle temporal lobe (bottom). A and B (top) show significant clusters superimposed onto a 3D image viewed from the back side and on intensity projection maps. B (bottom) shows the representative sagittal and coronal slices showing the significant cluster. Blue shows decrease with consolidation and red shows increase with consolidation. The circles in the sagittal and axial intensity projection maps represent the approximate location of the seed regions. Activations are depicted at p < 0.001 uncorrected and cluster size that exceed 90 voxels for all panels, showing the significant clusters. R, Right.