Human brain cells are far more powerful than scientists thought

A groundbreaking study published in the Proceedings of the National Academy of Sciences (PNAS) challenges previous notions about the capabilities of human brain cells. According to the research led by Hebrew University scientists, individual human neurons demonstrate far greater computational power than their counterparts in other mammals. This revelation could provide insight into the unique cognitive abilities of humans, such as language, imagination, mathematics, and invention.

To determine the computational prowess of individual brain cells, the researchers employed a novel method that blended sophisticated computer modeling with artificial intelligence. By gauging the difficulty of an advanced artificial neural network (ANN) replicating a biological neuron's response to incoming information, they found that neurons in the human cortex consistently outperformed those in other mammals.

The richly branched dendritic trees and distinctive electrical properties of human cortical neurons enable them to execute intricate computations on visual input, such as distinguishing between pictures of cats and dogs.

This newfound understanding of the human neuron's capabilities contradicts the prevailing belief that intelligence stems primarily from the sheer number of neurons and the connections between them. Instead, the study suggests that the complexity of individual neurons may have played a significant role in the evolution of human cognition.

Furthermore, the researchers developed a systematic framework for linking the physical structure of brain cells to their computational abilities, offering a promising approach for unraveling the mysteries of human thought, learning, and cognition.

The implications of this research extend beyond neuroscience and could also inform the future of artificial intelligence. Current machine learning systems rely on simplified artificial units. The new findings propose an alternative path: brain-inspired AI systems comprised of artificial units that possess computational depth and power, more akin to the capabilities of biological neurons.