Reacción a "Published a detailed map of the connections between brain cells in mice"

The MICrONS project provides a dataset of unprecedented scale and resolution. This dataset combines functional recordings with the high-resolution anatomical structure of several mouse visual cortical areas. Thus, more than 70,000 excitatory neurons and their responses to videos of natural scenes spanning 1 mm³ of the visual cortex were functionally recorded. This work reveals the detailed structure of approximately 60,000 excitatory neurons and 500 million synapses, representing the largest neocortical structure-function study conducted to date.

This body of work lays many of the foundations for several principles of functional organization that, although assumed, were unproven and represented gaps in our understanding of the nervous system. In fact, the connectivity principles now revealed at the structural and functional levels appear to play a fundamental role in sensory processing and learning. Furthermore, it is noteworthy that these principles are shared by both biological and artificial systems, so they represent fundamental information when designing artificial networks based on artificial intelligence. Collectively, these findings undoubtedly represent a long-awaited giant step forward, and are merely the tip of the iceberg of what is to come in understanding how the brain works. In fact, the authors also demonstrate that artificial recurrent neural networks trained on a simple classification task develop connectivity patterns that mimic the connectivity rules revealed by biological data. In short, the system's ability to process information and store it in the form of memory is determined by the connectivity of the system itself.

Furthermore, the authors have generated a basic artificial model that not only predicts neuronal activity in the visual cortex, but also the functional properties of neurons and their anatomical characteristics. In my view, these findings are of paramount importance. They will have a major impact on the field of neurocomputing and will facilitate further discoveries about the functional organization of neural circuits by allowing the exploration of previously unconsidered questions.

Both this data collection and the openly available models can be very useful for exploring which specific circuit dysfunctions could lead to functional alterations compatible with known pathologies.