Less is more, an efficient rule for brain neural connections and dynamic operation

The human brain has extremely powerful thinking and computing capabilities, but it only needs about 20W of ultra-low power, and its energy consumption is much lower than that of electronic computers

Recently, the research team (Dr.

Research shows that when the globally randomly connected network (RN) is reconnected to a more biologically realistic module network (MN), the network's operating consumption (distribution rate) and connection consumption are significantly reduced, and there is no standard in the dynamic mode The degree of avalanche (that is, criticality), which makes the network more effective in responding to external stimuli

Further analysis found that the increase in the density of the module during the reconnection process is the key to the overall property change: the increase in network topology correlation leads to the increase in dynamic correlation, making it easier for neurons to fire

The research clearly gives guidelines for the interaction of brain structure and dynamic properties to achieve a common efficiency optimization (rather than a trade-off between the two), and provides people with an understanding of the efficient operation principle of biological brains and high-performance brain-like computing devices.

###

Article information:

Less is more: Wiring-economical modular networks support self-sustained firing-economical neural avalanches for efficient processing