Scientists Show Cellular Membranes Play Direct Role in Memory Formation

Two scientists at Oak Ridge National Laboratory have spent decades studying the wrong problem—or so they thought. John Katsaras, a neutron scattering specialist, and Pat Collier, a cleanroom engineer, began by asking separate questions: one wanted to understand how soft materials might power brain-inspired computers; the other had spent forty years mapping the architecture of lipid membranes. When they decided to combine their work, they stumbled onto something neither had anticipated: the membranes that wrap our cells appear to be direct participants in how we form memories.

The finding emerged from deceptively simple experiments. Katsaras and Collier started by studying water droplets suspended in oil, a setup called a droplet interface bilayer that mimics the basic structure of a cell membrane. All biological membranes share the same fundamental design—a double layer of lipid molecules, each with a water-loving head and a water-repelling tail. When the team applied electrical stimulation to these artificial membranes, they noticed something unexpected in the electrical readings. The patterns looked like the kind of activity you'd see in a working brain.

That observation pulled them toward the actual membranes surrounding neurons, where memory and learning actually happen. What they discovered there was striking: these lipid bilayers don't just sit passively while ions flow through them. The membranes actively regulate that flow, and in doing so, they exhibit two electrical properties—memristance and memcapacitance—that are hallmarks of memory storage. In one region of a membrane, lipids can rearrange to form a memory resistor. In another region, the same membrane can function as a memory capacitor. The membrane, in other words, can remember.

The implications ripple outward in multiple directions. For neuromorphic computing—the field of building machines that process information the way brains do—this suggests a new class of materials that could be far more efficient and versatile than current approaches. The work also opens a window onto how existing drugs might work. Lithium, widely prescribed for bipolar disorder and under investigation for neurodegenerative diseases like Alzheimer's, interacts with these same membranes. Understanding those interactions at the molecular level could explain lithium's effects and point toward new treatments.

The research was only possible because of Oak Ridge's unusual combination of resources. The laboratory houses the Spallation Neutron Source, one of the world's most powerful tools for studying atomic-scale structure, and the Frontier supercomputer, the first machine to reach exascale computing power. Katsaras and Collier could observe how lipid molecules rearrange under electrical stress with unprecedented precision, then use computational power to model what they were seeing. Over five years, they assembled a growing body of evidence: in 2022, they showed that lipid bilayers could mimic long-term memory. More recently, they demonstrated that electrical and mechanical signals could restructure these membranes into distinct, stable states that persist over time.

The team has also begun exploring molecular machines called rotaxanes—tiny structures that change shape when exposed to light. Working with Louisiana State University, they found that light-triggered changes in these molecules can drive membranes to reorganize between memory and learning states. The next phase of work will use neutron scattering to watch how lithium atoms move through membranes and alter the flow of potassium ions, essentially observing the molecular machinery of memory formation in real time.

What started as two researchers asking whether they might find common ground has become something larger: evidence that the boundary between biology and materials science is more permeable than we thought. A membrane is not just a barrier. It is an active participant in thought itself.