Scientists have discovered the mechanism by which the brain maximizes its ability to process information

How do billions of independent neurons in our brain interweave with each other and build a powerful system that can even defeat the most advanced artificial intelligence? All of these tiny interactions seem to have something to do with incredible computing power The mystery has always been a mystery In the past 20 years, more and more evidences support the theory that the brain's own steady-state mechanism allows it to automatically adjust to a critical point, so as to maximize the excitation and not fall into a phase change like disorder This critical hypothesis asserts that information processing capacity is maximized on this critical line However, a key prediction of this theory (criticality is indeed a set point, not just necessity) has never been verified Until recently, in a study published on neuron, a team from the University of Washington directly confirmed this long-term prediction in the brains of free-living animals Keith hengen, research correspondent and assistant professor of biology at the school of Arts and Sciences at the University of Washington, said: "when neurons join together, they actively look for critical areas Our new research validates most of the critical line theory and proves that criticality is a feature of the normal working brain network " The researchers determined that criticality is actively regulated, but the mechanism behind it is not simple "We were surprised to find that in our model, criticality is mainly caused by a group of inhibitory neurons, which are very good at regulating the organization of larger networks," hengen said Criticality is the only known computer system that optimizes information processing, such as memory, dynamic range, and the ability to encode and transmit complex patterns Theoretical physicists initially suggested that the brain might be critical, but neuroscientists did not respond consistently "There's a long history of solid theoretical research on criticality, and there are some interesting controversies that add interest," hengen said I think the argument comes from two sides First of all, most in vivo studies are largely descriptive, because these data sets are difficult to collect and analyze In any case, there is no direct evidence that the brain processes critical states Secondly, there are many arguments about the mathematical methods people use to measure criticality Recently, people no longer measure the simple power law that can be obtained from random noise, and began to study the so-called exponential relationship So far, it's the only true sign of criticality and the basis of all our measurements " "Our laboratory has made a very important contribution to solving the problem of criticality in the brain, because we study the resolution of a single neuron, and observe the dynamic changes of the critical area as a function of time for quite a long time," he said Zheng Yu Ma, Ph.D in physics at the school of Arts and science at the University of Washington, and Ralf wessel, Professor of physics, CO authored the new study The study used neural recording data collected by hengen at the University of Brandeis from free-acting mice Hengen later set up his own laboratory at the University of Washington, and he is collecting his own neuron records, which span months and come from hundreds of neurons "The time resolution is very high, which is an advantage," Ma said Moreover, they can keep records for nine consecutive days I'm still very surprised Few laboratories are able to keep records for nine consecutive days " With a few exceptions, the past neural recording time was usually 30 minutes, up to a few hours, which was used to limit the maximum value of critical experimental tests With Ma's breakthrough in computing power, hengen and his coauthors can combine a large number of single neuron data recorded by him to construct the activity model of the whole neural network Using the technique of tracking neuron activity for more than one week, researchers have demonstrated for the first time that the network dynamics in the visual cortex can be adjusted to a critical state steadily even in the whole light dark cycle Next, by blocking one eye's vision, the researchers found that the critical state was severely damaged, a day before the operation that affected the firing frequency of a single neuron worked After 24 hours, a critical state appeared again in the recording, at which time a single neuron was inhibited by visual deprivation "When there's a mismatch between what animals expect to see and what they actually see through their eyes, there's a problem with computing power," hengen said "This is consistent with the theoretical physical view that the critical region is independent of the discharge frequency," he said This is not just the total number of peaks in the network, because in the early stage of visual deprivation, the discharge frequency does not change at all, while the critical state is unbalanced " Now, researchers believe that the key to the brain may be the inhibitory neurons that exert and organize computational dynamics This finding is of great significance for motor learning and disease treatment Hengen pointed out that the brain's ability of self-organization around criticality is an active process In many serious human diseases, the disorder of homeostasis regulation increasingly involves serious human diseases, such as Alzheimer's disease, epilepsy, Leiter syndrome, autism and schizophrenia "This study shows that criticality is the ultimate goal of the brain network to achieve self balance," hengen said It's a good idea that the brain can adjust the features that appear to be in a state that physicists accurately predict Intuitively, evolution selects the gene that produces the optimal solution But time will tell We still have a lot of work to do "