Separate neurons for content and context enable flexible human memory

The human brain must be able to link memory content to the circumstances in which it occurs. Researchers in Bonn have now discovered how the human brain uses two different groups of neurons to store content and context separately. These nerve cell groups work together in a coordinated manner to form memories, rather than mixing signals in the activity of individual cells. The study results have now been published in the renowned journal Nature.

Humans have the remarkable ability to remember the same person or object in completely different situations. We can easily distinguish between dinner with a friend and a business meeting with the same friend. "We already know that deep in the memory centers of the brain, specific cells, called concept neurons, respond to this friend, regardless of the environment in which he appears," says Prof. Florian Mormann from the Clinic for Epileptology at the UKB, who is also a member of the Transdisciplinary Research Area (TRA) "Life & Health" at the University of Bonn. However, the brain must be able to combine this content with the context in order to form a useful memory. In rodents, individual neurons often mix these two pieces of information. "We asked ourselves: Does the human brain function fundamentally differently here? Does it map content and context separately to enable a more flexible memory? And how do these separate pieces of information connect when we need to remember specific content according to context?" says Dr. Marcel Bausch, working group leader at the Department of Epileptology and member of TRA "Life & Health" at the University of Bonn.

Watching the human brain in real time

To investigate this, the Bonn researchers used the electrical activity of individual neurons in the brains of people with drug-resistant epilepsy. For purely diagnostic purposes, electrodes were implanted in the hippocampus and surrounding brain regions of these patients - regions that are essential for memory. While their seizures were being recorded to determine whether they were eligible for surgery, they voluntarily participated in experiments on a laptop. They were shown pairs of images that they had to compare based on different questions. For example, they had to decide whether an object was "bigger" if the trial began with the question "Bigger?". "This allowed us to observe how the brain processes exactly the same image in different task contexts," says Mormann.

The research team analyzed more than 3,000 neurons and identified two largely separate groups of neurons: Content neurons fired in response to specific images (e.g., a biscuit, regardless of the task. Context neurons fired in response to specific task contexts (e.g., the question "Bigger?"), regardless of the image shown. In contrast to rodents, very few neurons encoded both simultaneously. "A key finding was that these two independent groups of neurons encoded content and context together and most reliably when the patients solved the task correctly," says Bausch.

The connections between them strengthened over the course of the experiment: the firing of a content neuron began to predict the activity of a context neuron a few tens of milliseconds later. "It seemed as if the 'biscuit' neuron was learning to stimulate the 'Bigger?' neuron" says Mormann. This happens in the sense of a gatekeeper for the flow of information, such that only the relevant context that was previously active is retrieved. This process, known as pattern completion, allows the brain to reconstruct the complete memory context from only partial information. "This division of labor probably explains the flexibility of human memory: the brain can reuse the same concept in countless new situations without needing a specialized neuron for each individual combination, by storing content and context in separate 'neural libraries'" says Bausch, and Mormann adds: "The ability of these neuronal groups to link spontaneously allows us to generalize information while preserving the specific details of individual events."

Although the study used specific questions as interactive contexts on a laptop, there are also other contexts that are passive, such as the room you are in. It remains to be determined whether these everyday background contexts are processed by the same neural mechanisms. In addition, the mechanisms must also be tested outside the clinical setting. An important next step will be to investigate whether deliberately disrupting the interaction between these neurons prevents a person from retrieving the correct memory in context or making the right decision.