Share on:
01/12/2026 13:12
How the brain organizes memories
Researchers from TU Berlin are involved in a Nature study that shows how the human brain stores content and contexts separately and specifically links them together
How does the human brain manage to reliably separate memories and yet connect them at the right moment? A new study, published in the renowned journal Nature, provides a fundamental answer. Johannes Niedieck, researcher at the Berlin Institute for the Foundations of Learning and Data (BIFOLD) at the TU Berlin, was also involved in the international research work led by the University of Bonn.
The study shows for the first time directly in humans that the brain stores memory content and its context in two largely separate groups of nerve cells. However, these specialized neurons work closely together and dynamically couple when a memory is formed or correctly recalled. This interaction enables the high flexibility of human episodic memory.
Why we don’t confuse experiences
Johannes Niedieck explains how this memory function works in everyday life using a personal example: “On Sunday I was at the Technology Museum with my daughter, on Monday we sang together in the kitchen. I can clearly remember both events separately – I don’t suddenly think that we sang in the Technology Museum.” At the same time, the individual elements of a memory can be quickly linked together: “When I think of my daughter, I immediately think of both the technology museum and the kitchen, even though I only experienced both situations once.” It is precisely this ability – to cleanly separate the contents and contexts of a memory on the one hand and to connect them reliably on the other – was the focus of the study that has now been published.
Watch human memory in real time
This mechanism was investigated under the direction of Dr. Marcel Bausch and Prof. Florian Mormann from the Clinic for Epileptology in Bonn used high-resolution recordings of individual nerve cells in patients with drug-resistant epilepsy. For diagnostic reasons, hair-thin electrodes were implanted in the hippocampus and adjacent brain regions, which play a central role in memory. During voluntary experiments, participants worked on tasks on laptops in which identical images had to be assessed in different task contexts.
The analysis of over 3,000 neurons revealed two clearly distinguishable groups: content neurons, which responded to specific images regardless of context ––, and context neurons, which responded to the respective task regardless of the image. Only very few nerve cells encoded both pieces of information at the same time.
What was crucial was that memories were formed most reliably when both groups of cells worked together in a coordinated manner. As the experiments progressed, the connections between them measurably strengthened. A single memory element could thus reactivate the entire context – a process called pattern completion.
Contribution from TU Berlin
Johannes Niedieck was already involved in planning the study, conducting the experiments with epilepsy patients and discussing and classifying the results during his doctorate at the University of Bonn. He continued this work while working in the Machine Learning group of Prof. Dr. Klaus-Robert Müller at the TU Berlin and BIFOLD. His interdisciplinary background at the interface of neuroscience and data-driven analysis helped to precisely evaluate and understandably describe the complex neuronal connections.
Perspectives for research and medicine
The results provide a central explanation for the flexibility of human memory: the brain can use the same content in ever new contexts without having to store every possible combination separately. Future studies should clarify whether everyday background contexts, such as places or moods, are processed according to the same principles and how disruptions to this neural interaction influence remembering.
The study “Distinct neuronal populations in the human brain combine content and context” is available online at Nature: https://www.nature.com/articles/s41586-025-09910-2
DOI: 10.1038/s41586-025-09910-2
We would be happy to provide you with further information:
Dr. Johannes Niediek
Department of Machine Learning
BIFOLD
Faculty IV Electronic Engineering and Computer Science
Email: johannes.niediek@tu-berlin.de
Images
Criteria of this press release:
Journalists
Information technology, Nutrition / healthcare / nursing
transregional, national
Research projects
German
