How does the brain learn?

Summary: A new open source model of synaptic plasticity in the neocortex could promote understanding of how learning occurs in the brain.

Source: University of Montreal

Everyone knows that the human brain is extremely complex, but how exactly does it learn? Well, the answer may be much simpler than is commonly believed.

An international research team involving the Université de Montréal has made a major advance in accurately simulating synaptic changes in the neocortex believed to be the key to learning, opening the door to greater understanding of the brain.

The scientists’ study, featuring an open source model, was published on June 1 in Nature communications.

“This opens up a world of new directions for scientific inquiry into how we learn,” said Eilif Muller, IVADO Research Assistant Professor at UdeM and Canadian CIFAR AI Chair, who co-directed the study at the Blue Brain Project of the École Federal Polytechnic of Lausanne (EPFL), Switzerland.

Muller moved to Montreal in 2019 and is pursuing his research at the Biological Learning Architectures Laboratory, which he founded at the CHU Sainte-Justine Research Center in association with UdeM and Mila, the Quebec Artificial Intelligence Institute.

“Neurons are shaped like trees and the synapses are the leaves on their branches,” said Muller, senior co-author of the study.

“Previous approaches to modeling plasticity have ignored this tree structure, but we now have the computational tools to test the idea that synaptic interactions on branches play a critical role in driving learning in vivo,” he said.

“This has important implications for understanding the mechanisms of neurodevelopmental disorders such as autism and schizophrenia, but also for the development of new and powerful neuroscience-inspired artificial intelligence approaches.”

Muller collaborated with a group of scientists from the Blue Brain Project of EPFL, the University of Paris, the Hebrew University of Jerusalem, the Cajal Institute (Spain) and the Harvard Medical School to develop a model of synaptic plasticity in the neocortex. based on data-bound postsynaptic dynamics of calcium.

How does it work? It’s complicated, but ultimately, simpler than you might think.

The brain is made up of billions of neurons that communicate with each other forming trillions of synapses. These connection points between neurons are complex molecular machines that constantly change due to external stimuli and internal dynamics, a process commonly referred to as synaptic plasticity.

In the neocortex, a key area associated with learning high-level cognitive functions in mammals, pyramidal cells (PCs) account for 80-90% of neurons and are known to play an important role in learning. Despite their importance, the long-term dynamics of their synaptic changes have only been experimentally characterized among a few types of CP and have been shown to be diverse.

As a result, there has been only limited understanding of the complex neural circuits they form, especially across the stereotyped cortical layers, which determine how different regions of the neocortex interact.

The innovation of Muller and his colleagues was to use computational modeling to gain a more complete view of the dynamics of synaptic plasticity that govern learning in these neocortical circuits.

By comparing their results with the available experimental data, they showed in their study that their model of synaptic plasticity can capture the various plasticity dynamics of the different PCs that make up the neocortical microcircuit. And they did so using only one set of unified model parameters, indicating that the plasticity rules of the neocortex could be shared across pyramidal cell types and thus be predictable.

Verification of the generalization of the plasticity model on the type of connection from L4-PC to L2 / 3-PC. a 3D render of a representative pair of L4-PCs connected to L2 / 3-PCs in the in silico model. The inset shows an enlarged view of the synapses that mediate the connection (yellow spheres). b Evolution over time of EPSP amplitude simulated during a typical plasticity induction protocol (top left; one coupling shown on 100). The mean EPSP amplitudes (top right) are shown before (baseline; blue) and after (long-term; orange) the induction protocol. c Comparison of in silico and in vitro EPSP ratios for positive and negative times and with presynaptic NMDAR blocker MK801. Experimental data and simulations without MK801 on the left panel, with MK801 (in vitro) and γd = 0 (in silico) on the right panel. Welch’s bilateral t-test of unequal variances was ns for each protocol (p-value from negative to positive pacing time: 0.268, 0.209 MK801, 0.959 MK801; n = 100). Experimental data (in vitro) by Rodríguez-Moreno and Paulsen42. Population data reported as mean ± SEM. Credit: researchers

Most of these plasticity experiments were performed on rodent brain slices in vitro, where the calcium dynamics driving synaptic transmission and plasticity are significantly altered compared to learning in the intact brain in vivo. Importantly, the study predicts plasticity dynamics that are qualitatively different from the reference experiments performed in vitro.

If confirmed by future experiments, the implications for our understanding of plasticity and learning in the brain would be profound, believe Muller and his team.

“The interesting thing about this study is that this is further confirmation for scientists that we can overcome gaps in experimental knowledge by using a modeling approach when studying the brain,” said EPFL neuroscientist Henry Markram, founder and director of the Blue Brain Project.

“Also, the model is open source, available on the Zenodo platform,” he added.

“Here we have shared hundreds of plastic pyramid cell connections of different types. Not only is it the most widely validated model of plasticity to date, it also represents the most comprehensive prediction of the differences between plasticity observed in a petri dish and an intact brain.

See also

“This leap is made possible by our collaborative approach to team science. Furthermore, the community can go further and develop their own versions by modifying or adding to it: this is open science and will accelerate progress. “

About this research news on synaptic plasticity

Author: Press office
Source: University of Montreal
Contact: Press Office – University of Montreal
Image: The image is attributed to the researchers

Original research: Free access.
“A calcium-based plasticity model to predict long-term potentiation and depression in the neocortex” by Giuseppe Chindemi et al. Nature communications


Abstract

A calcium-based plasticity model to predict long-term potentiation and depression in the neocortex

Pyramidal cells (PCs) form the backbone of the layered structure of the neocortex, and the plasticity of their synapses is thought to underlie learning in the brain.

However, such long-term synaptic changes have only been experimentally characterized among a few types of PC, posing a significant barrier for the study of neocortical learning mechanisms.

Here we introduce a model of synaptic plasticity based on data-constrained post-synaptic calcium dynamics and show in a neocortical microcircuit model that a single set of parameters is sufficient to unify the available experimental results on long-term potentiation (LTP) and depression at long term (LTD) of connections to the PC.

In particular, we find that the different plasticity results between different PC types can be explained by cell type-specific synaptic physiology, cell morphology, and innervation patterns, without requiring type-specific plasticity.

By generalizing the model to extracellular calcium concentrations in vivo, we predict plasticity dynamics that are qualitatively different from those observed in vitro.

This work provides a first complete null model for LTP / LTD between neocortical PC types in vivo and an open framework for further development of cortical synaptic plasticity models.