From the moment we are born, and even before that, we interact with the world through movement. We move our lips to smile or talk. We reach out to touch. We move our eyes to see. We wiggle, we walk, we point, we dance. How does our brain remember this wide range of movements? How do you learn new things? How do calculations make it necessary for us to drink a glass of water without dropping, crushing or losing it?
Technion professor Jackie Schiller of Ruth School of Medicine and Bruce Rappaport and her team examined the brain at the level of a single neuron to shed light on this mystery. They found that computation occurs not only in the interaction between neurons (neurons), but within each individual neuron. It turns out that each of these cells is not a simple key, but a complex arithmetic machine. This discovery was recently published in Sciences Journal, promising changes not only in our understanding of how the brain works, but a better understanding of conditions ranging from Parkinson’s disease to autism. And if that’s not enough, these same findings are expected to advance machine learning, and provide inspiration for new architectures.
The brain’s primary motor cortex controls movement. In the field, researchers can pinpoint exactly which neurons fire at any given moment to produce the movement we see. Professor Schiller’s team was the first to get even closer, examining the activity not of the entire neuron as a single unit, but of its parts.
Each neuron has branching extensions called dendrites. These dendrites are in close contact with the ends (called axons) of other neurons, allowing communication between them. The signal travels from the dendrites to the cell body, and then forwards through the axon. The number and structure of dendrites varies greatly between neurons, as the crown of one tree differs from that of another tree.
The special neurons that Professor Schiller’s team focused on were the largest pyramidal neurons in the cerebral cortex. These cells, known to be highly involved in locomotion, have a large dendritic tree, with many branches, sub-branches and sub-branches. What the team discovered is that these branches don’t just forward information. Each subbranch performs a calculation on the information it receives and passes the result to the larger subbranch. The subsidiary performs a calculation on the information received from all its subsidiaries and passes it on. Furthermore, multiple tree branches can interact with each other to amplify their common computational product. The result is a complex computational process that is performed within each individual neuron. For the first time, Professor Schiller’s team has shown that a neuron is divided, and that its branches perform the calculations independently.
“We’re used to thinking of every neuron as some kind of whistle, whether it’s a horn or not,” explains Professor Schiller. “Instead, we’re looking at a piano. Its keys can be pressed simultaneously, or sequentially, to produce an infinite number of different melodies.” It is this complex symphony that plays in our brains that enables us to learn and perform an infinite number of different, complex and subtle movements.
Multiple neurodegenerative and neurodevelopmental disorders are likely to be associated with changes in the ability of neurons to process data. In Parkinson’s disease, the dendritic tree has been observed to undergo anatomical and physiological changes. In light of the new discoveries of the Technion team, we understand that as a result of these changes, the ability of neurons to perform parallel computations is reduced. In autism, it seems possible that the excitability of the dendritic branches is altered, resulting in the many effects associated with the condition. A new understanding of how neurons function is opening up new research avenues regarding these and other disorders, with the hope of alleviating them.
These same findings could serve as inspiration for the machine learning community. Deep neural networks, as their name suggests, attempt to create a program that learns and works somewhat similarly to the human brain. Although their progress is constantly making news, these networks are rudimentary compared to the living brain. A better understanding of how our brain works can help design more complex neural networks, enabling them to perform more complex tasks.
This study was led by two Ph.D. Prof. Schiller. Student candidates Yara Ater and Shay Ahfat, who contributed equally to the research. The team also included postdoctoral fellow Nate Cermak (now a neuroengineer) and a Ph. Student Hades Benesti and three collaborators: Professor Omri Barak, Yitzhak Schiller, and Alon Poleg Polsky.
The study was supported in part by the Israel Science Foundation, the Prince Funds, the Rappaport Foundation, and the Zuckerman Postdoctoral Fellowship.