30 October 2018 - Research

Understanding how we learn and remember means measuring and decoding the activity patterns of complex circuits in our brain. The lab of Fabian Kloosterman has developed a new method to measure and interpret brain activity in real-time. Their solution is a real game-changer for studies on memory and other complex brain processes.

At the very basis of the complex processes taking place in our brain, lies the signaling between neurons, resulting in specific neuronal activity patterns. Only by understanding the meaning of these patterns can scientists unravel their link to behavior and thought.

Decoding memories

The neuronal activity patterns in the hippocampus are of great interest to both scientists and clinicians, because this brain area is important to memory. We know that for every new experience, a walk in a new city for example, there is a unique pattern of neuronal activity that represents that experience—a brain spatial map if you like, created by designated place cells in the hippocampus. Where it gets really interesting, is that the same pattern comes into play for the memory of this walk, even when resting or sleeping.

Fk Kloosterman Fabian Mobile 300 340 C1

Detecting the neuronal activity patterns of these brief moments of so-called replay is very challenging, explains prof. Fabian Kloosterman:

“The algorithms to map the activity of a large number of neurons often have high computational demands and generally require a large amount of data. This is an issue, because to really link specific neuronal activity patterns to behavior, or the memory of behavior, we need to be able to manipulate neuronal circuits in real-time.”

Real-time measurements

“We want to be able to read out patterns of activity in real-time, which means decoding information from a large amount of neurons, but in an extremely short time,” explains Davide Ciliberti, first-author of the study. To make this technically feasible, the team devised a new system that detects spiking patterns online with minimal computational delays.

Fk Ciliberti Davide Mobile 300 320 C1

The software framework—aptly named Falcon, after the fastest moving creature on earth—was validated in an experimental setup with rats exploring a maze.

Ciliberti: “Each 10 milliseconds of data is translated into an estimate of where the animal is on the maze. If you would do this when the animal is moving, you can track exactly where the animal is, just by looking at the neuronal activity pattern in the hippocampus. When the animal is resting, instead of a real position you extract the remembered or imagined position on the maze.”

Previously, computing such estimates was done offline and retrospectively, but Falcon allows to move all of this into a real-time scenario.

“Our ready-to-use brain-computer interface is unique, as it enables us to selectively manipulate specific neuronal activity sequences at a millisecond timescale,” says Kloosterman. “It’s a big step forward towards a deeper understanding of how spontaneous and subconscious brain activity patterns mediate learning and long-term storage of new information. In the long run, this could help us understand and treat conditions triggered by specific memories, for example post-traumatic stress disorder.”

'Real-time classification of experience-related ensemble spiking patterns for closed-loop applications' by Ciliberti et al. has been published in eLife
View other articles: