When we have a new experience, our brains quickly form a long-term memory of that event. The process of memory formation uses two major brain structures, the hippocampus and the entorhinal cortex (EC), which are further subdivided into sub-regions or layers that are thought to perform specialized functions relevant for the different kinds of events in our lives.

Memory neuroscientists have believed that when we recall these memories, our brains use the same hippocampal and EC circuits that were used to form the original memory. In a recent study, we surprisingly found that recalling memories recruits a “detour” brain circuit that branches off from the original memory circuit.

While most regions of the hippocampus memory network have been well studied, the functional role of a small area called the subiculum has been mysterious. To investigate this, we developed a novel genetically engineered mouse line that enabled us to use light to turn off/on subiculum neurons . By modulating the activity of subiculum neurons during a fear-learning event, in which mice have to learn that a particular chamber is scary, we attempted to determine whether the subiculum neurons are important for fear learning, fear memory recall, or both.

Turning off subiculum neurons only during fear learning had no effect on their ability to form and recall this memory. However, in another group of mice, we observed that although these animals learned the fear experience normally, they couldn't recall this experience when subiculum neurons were turned off. This evidence indicated that the detour circuit involving the subiculum – that is hippocampal CA1 to subiculum to EC layer 5 – is necessary for memory retrieval but not memory formation. In a series of further experiments, we found that the direct circuit from hippocampal CA1 to EC layer 5 is necessary for memory formation but not memory retrieval. Quite unexpectedly, we had uncovered that the output of the memory brain region, the hippocampus (CA1), has distinct brain circuits for memory formation and retrieval.

The existence of distinct memory circuits raises an interesting question. Why would we need two brain circuits for memory formation versus retrieval? Why not use the same brain circuit for both functions? When thinking about our memories, we often have experiences that require memory updating/editing. A common example is a safe neighborhood that we are familiar with. If we have a traumatic experience in this neighborhood, our impression is “updated” to a potentially scary neighborhood. Of course, a similar phenomenon occurs for extremely positive experiences in our lives. We realized that two, essentially parallel circuits would make it easier and more efficient to edit/update a memory rather than one single circuit. In a set of experiments, we were able to find evidence in support of this suggestion, specifically that as the subiculum recall circuit is activated, simultaneous activation of the direct hippocampal CA1 to EC layer 5-memory formation circuit allows us to add new information to the recalled memory.

While rapid memory updating/editing is one reason to have distinct memory circuits for memory formation versus retrieval, we expect that future research will discover many other functions. For example, we found that in the early stages of Alzheimer’s disease, memory retrieval but not formation is impaired, suggesting that changes in the subiculum brain circuit may contribute to these symptoms.