Reacción a "Synapses in individual neurons do not follow a single strategy during learning as previously thought"

The press release is a good summary of the article. The results are obtained using a very precise technology (tracking neuronal activity in vivo in the brains of mice using fluorescent calcium indicators and 2-photon, or multi-photon, microscopy), which we also use in our lab, and which they have managed to adapt to the needs of their study in a brilliant way.

For years, scientists have known that some synapses get stronger while others get weaker when we learn, but we didn't know why some synapses change and others don't. In this paper, researchers show that neurons do not follow a single rule when learning, as previously thought. Each neuron can use several rules at once, depending on where in the cell the synapse is located.

This discovery changes what we knew about how the brain solves the so-called ‘credit allocation problem’, which is how the smallest components of our brain (such as synapses) know whether they are helping overall learning.

The study is descriptive, based on direct observations, and does not speculate on the interpretation of its results. It also provides a comprehensive analysis of the mechanism by which this process takes place in mammals, using very precise and novel technology. He notes different applications, but leaves that for future studies.

The most important limitation is that the study is carried out in animals, rodents, and not in humans. Knowing the vast differences between our brains and those of mice, especially in complex information processing such as that which occurs during learning, we might speculate that the same processes might be slightly different in humans. In fact, as a curiosity, studies have been carried out in which human cells have been introduced into the brain of a mouse (what we call a chimera) and different (enhanced) capabilities could be demonstrated in these animals (I could provide a reference, but I don't think it is relevant).

The results have important implications for the field of neuroscience. They certainly help us to better understand how the brain works and what goes wrong when it suffers from disease. It may also help us design better strategies to repair the brain when it is damaged. Finally, more in the technological than the biomedical realm, it could inspire new ways of creating artificial intelligence, with networks that use multiple rules like real neurons. The designs of artificial neural networks used in deep learning have always been inspired by how our brains work, and so this new breakthrough may provide new ideas on how to make this tool more powerful and efficient.