The resilient brain: Losing neurons without missing a beat

Illustration by: Shira Lottem.



Dropping your phone into a bowl of soup is a pretty effective way to destroy it. But it doesn’t necessarily take complete submersion to render a phone, or any other piece of technology for that matter, utterly useless. By comparison, brains are incredibly resilient to damage and can instantly absorb the loss of neurons, without inconveniencing their owner in the least. Now, Christian Machens, head of the Theoretical Neuroscience Lab at the Champalimaud Centre for the Unknown in Lisbon, and his team, have come up with a mathematical model that can explain how brains succeed where machines fail.

“A single burnt transistor can cause your alarm clock to fail on the morning of a flight to the Caribbean, or your computer to shut down in the middle of an important presentation. This is the way with most man-made machines, they don’t work very well if you take parts away. Biological systems such as the brain, on the other hand, have the amazing ability to keep working even if you eliminate some important-looking bits”, says Machens. Tall and lean, he speaks with a humorous air and a very precise German accent about his work, recently published in the scientific journal eLife.

Machens’ model proposes that brains can compensate instantaneously for neural death, a capability that would surprise many researchers in the neuroscience community who are only familiar with compensation across long timescales. “It is well known that the brain is very plastic”, he explains. “People can often recover, at least to some extent, motor or cognitive functions they lost as a result of brain damage.”

This recovery happens through a process called plasticity, in which connections between neurons are rearranged, and it takes time, sometimes years. “But what we are proposing here is a way by which the brain can compensate for the loss of neurons without missing a single beat”, argues Machens. “You could be having what is called a ‘silent stroke’ while brushing your teeth and carry on as though nothing had happened. It could be literally years before the damage is discovered in some unrelated exam.”      

How is it possible that, even though many neurons die, nothing bad happens in terms of brain function? “To answer this question, you have to think about the way the brain represents information in the first place,” Machens replies, gesturing towards the window, where we see a seagull perched on the mast of a small boat. “This image is encoded in your brain across a flexible network of neurons. Each time you look at the same image, though it looks identical to you, it is actually represented by a different pattern of electrical activity in your brain. This is because information is encoded in the brain in a redundant manner. On the one hand this may seem less efficient, but in fact this is what makes the brain so robust against damage.”

The model Christian and his colleagues constructed shows how this redundancy is a key aspect of brain function, which, as a by-product, makes the brain immune to partial loss of neurons. “There is more than one possible neural representation that conveys the same mental object. In our model, we show that when some neurons are lost, the network just jumps to the next available set of neurons that are able to encode the same information. Moreover, we can predict exactly which neurons will become more or less active, depending on which neurons were eliminated.”

Obviously, at some point, the brain will be overwhelmed by neural loss. “We can’t just keep eliminating neurons forever”, Machens  points out. “Another prediction of the model is at what point the system will ‘give up’. Information can be redistributed across the network until the brain hits what we call ‘the recovery boundary’, which is where the system can no longer absorb loss and deficits emerge.”

Though this model was developed mathematically, it is more than just a theory. As Machens explains, there are several lines of evidence that support it. “A silent stroke is one example. Another example, which is even more direct, was published last year. In that study, the researchers discovered that silencing certain neurons triggers the immediate emergence of activity in other neurons within the same network to take their place.”

He pauses for a moment and then continues with a smile. “In fact, the frequent failure of ‘silencing studies’ is another piece of support in a sense. In these studies, researchers temporarily deactivate a set of neurons in the brain in order to gauge their importance for a certain function. We believe that many ‘failed’ silencing studies, in which researchers silence neurons but see no effect, might actually be cases of successful instantaneous compensation. There is potentially a lot of interesting data out there that is being overlooked.”

Currently, Machens’ team is working on finding more evidence to support the model, which carries implications that go well beyond the field of theoretical neuroscience. “One of the things that make this work so exciting is that it provides the neuroscience community with a new perspective on brain damage. Most studies on disorders that involve neural loss focus on the molecular level, but here we show that there are processes happening on the level of the network that are also important for understanding how the brain deals with neural death”, he concludes.




Liad Hollender works as a Science Writer at the Science Communication Office at Champalimaud Research



Edited by: Ana Gerschenfeld & Catarina Ramos(Science Communication office).; Illustration by: Shira Lottem.



Loading Likes...