The human hippocampus as a complementary indexing machine for episodic memory

Episodic memories make us who we are, yet how these memories are coded in the human brain remains an open question. We know that the human hippocampus is essential for turning unique experiences into memory episodes, such as a memory from the last concert you have attended. We also know that the human hippocampus has two types of neurons that could be critical for coding such episodic memories. These are Concept Cells, which store general semantic concepts (i.e. the artist, or the location where the concert was), and Episode Specific Neurons, which hold together all the elements that belong to an episode (i.e. seeing Taylor Swift performing in Edinburgh). Concept cells, which are well-known, represent a context invariant and highly abstract code for certain elements that matter to us, such as a certain place, or a certain person. In contrast, Episodic Specific Neurons, which have only recently been observed, represent a conjunctive code that uniquely identifies individual episodes. However, how these two types of neural codes are used by the hippocampus to code episodic memories is unclear. Specifically, we don’t know whether and how Episode Specific Neurons and Concept Cells interact with each other and with the neocortex. We don’t know where in the hippocampus they are, how they are allocated to code for a memory, and whether they are causally relevant for human episodic memory. The aim of MemoryIndex is to answer these questions via testing computationally motivated hypotheses in a multimodal project using a unique combination of human single neuron recordings, ultra-high field fMRI and invasive electrical stimulation. The results of this project will critically advance our understanding of the fundamental coding mechanisms that the human hippocampus uses for storing episodic memories. These findings could potentially pave the way for developing a neural prosthesis to restore memory function in patients.