Skip to main content

Can the brain keeps useful information while ignoring unuseful noise

Extreme close-up of neurons in the human brain
Published Mar 22, 2022

Imagine you're in a stadium reporting on a game. It's a loud environment which you don't want to transmit. But when you're interviewing a player, then you do want to transmit. Wouldn't it be great if you could automatically switch on or off the microphone based on the type of information, i.e. noise vs the interview?

Hi, Arvind Kumar! 

You and Luiz Tauffer have published a new paper: "Short-term synaptic plasticity makes neurons sensitive to the distribution of presynaptic population firing rates".

What problem does your research address?

"A brain is a noisy place, pretty much like a busy marketplace. In each tiny corner of the brain, one can find thousands of cells constantly sending signals to one another. This, of course, generates a lot of noise. A fundamental question that has always occupied neuroscientists' thoughts is: how cells in the brain can retain useful information while ignoring the surrounding noise?"
 
What is your solution to the problem?

"One traditional view of this problem assumes that the neurons' connections are modified to solve this noise-cancelling problem. Indeed, networks of cells can filter out environment noise when selected synapses are properly tuned. That is what we call learning. But this is obviously a slow process.

Noisy brain
The activity in the brain is like a highly noisy ocean. The traces of synaptic connection (red and blue traces) are taken from Silberberg Curr. Op. In Neuobiol. 2008, with permission.

We have now found the optimal combination of short-term-dynamics properties that automatically enhance informative signals whereas filtering out noisy fluctuations – without any learning. We were surprised that these theoretically identified properties are found in the cerebellum and hippocampus. While the cerebellum is crucial for fine motor control, the hippocampus plays an important role in spatial navigation and memory formation. This leads us to believe that the signal gain effects we have found are likely playing an important role in behaviour."

How can this discovery be used?

"Our results show that whether the information is encoded as a small change in the activity of many neurons (dense code) or a large change in the activity of few neurons (sparse code) depends on the short-term-dynamics of synapses. So right away, our results predict the nature of neural code used by different brain regions. This is of great importance when trying to decode brain activity, for example, to create brain-machine interfaces."
 
In what context can this be helpful to people in society?

"These results unravel a simple mechanism that the brain uses to solve a key problem – communication in the presence of high noise. Short-term dynamics of synapses are altered in brain disorders such as autism spectrum disorder. It is too early to say but building on our results, we can learn much more about brain diseases which essentially arise due to poor information processing.
 
What significance or use do you think these results will have in 5 or 10 years?

"The results have revealed a completely new consequence of biological synapses for neural coding – how information is encoded in the brain. Of course, proper information encoding is crucial for healthy brain function. So, a particular direction we are exploring is how short-term-dynamics are altered in brain diseases.
In parallel, we expect that colleagues working in the field of neuromorphic engineering will find our results very useful for building brain-inspired computing devices.

The technologies that interact with the brain are evolving fast and becoming ever more precise. At the point where networks become controllable at a single-cell scale, theoretical findings such as ours will provide the means to develop the neuro-algorithms that will lead the human experience in the future."

Link to the online version of the paper:
www.eneuro.org/content/early/2021/02/11/ENEURO.0297-20.2021

Contact:

Belongs to: School of Electrical Engineering and Computer Science
Last changed: Mar 22, 2022