How hippocampus and cortex contribute to recognition memory: revisiting the complementary learning systems model

Abstract

We describe how the Complementary Learning Systems neural network model of recognition memory (Norman and O'Reilly (2003) Psychol Rev 104:611-646) can shed light on current debates regarding hippocampal and cortical contributions to recognition memory. We review simulation results illustrating three critical differences in how (according to the model) hippocampus and cortex contribute to recognition memory, all of which derive from the hippocampus' use of pattern separated representations. Pattern separation makes the hippocampus especially well-suited for discriminating between studied items and related lures; it makes the hippocampus especially poorly suited for computing global match; and it imbues the hippocampal ROC curve with a Y-intercept > 0. We also describe a key boundary condition on these differences: When the average level of similarity between items in an experiment is very high, hippocampal pattern separation can fail, at which point the hippocampal model will start to behave like the cortical model. We describe the implications of these simulation results for extant debates over how to describe hippocampal versus cortical contributions and how to measure these contributions.

Figure 1

Illustration of the sharpening of hidden (perirhinal) layer activity patterns in a miniature version of the CLS cortical model. (a) shows the network prior to sharpening; perirhinal activity (more active = lighter color) is relatively undifferentiated. (b) shows the network after Hebbian learning and inhibitory competition produce sharpening; a subset of the units are strongly active, while the remainder are inhibited. Figure adapted from Norman and O’Reilly (2003).

Figure 2

Diagram of the CLS hippocampal network. The hippocampal network links input patterns in entorhinal cortex (EC) to relatively non-overlapping (pattern-separated) sets of units in region CA3. The dentate gyrus (DG) serves to facilitate pattern separation in region CA3. Recurrent connections in CA3 bind together all of the units involved in representing a particular EC pattern; the CA3 representation is linked back to EC via region CA1. Learning in the CA3 recurrent connections, and in projections linking EC to CA3 and CA3 to CA1, makes it possible to recall entire stored EC patterns based on partial cues. Figure reprinted from Norman and O’Reilly (2003).

Figure 3

Histogram of the memory signal generated by the CLS cortical network, given 20% average overlap between input patterns. Figure adapted from Norman and O’Reilly (2003).

Figure 4

Histogram of the memory signal generated by the CLS hippocampal network, given 20% average overlap between input patterns. Figure adapted from Norman and O’Reilly (2003).

Figure 5

Histogram of the memory signal generated by the CLS hippocampal network, given 40.5% average overlap between input patterns. Figure adapted from Norman and O’Reilly (2003).