Kijun Times

Andreas Nieder, a neuroscientist at the University of Tubingen, with researchers, produced an artificial neural network that can recognize the number of animals and things. The artificial network learned to recognize objects on

the basis of 1.2 million images which were categorized to one thousand categories. After finished this training, the network was able to categorize thousands of new images with a high proportion of success. Then, they learned the dot pattern of 0 to 30 points in the neural network and observed the dot pattern that this neural network was showing

Next, they proceed the training that the artificial neurons were trained to detect amounts. And they succeeded in making show the 'number sense' like a person. "Although the network was never trained to distinguish between numbers. But the network had developed a sense of numbers itself." says Nieder. The most fundamental mission of vision is to recognize visible objects. And this new study shows that how number neurons can develop itself from an artificial visual system that was trained to recognize visible objects. “Number sense does not seem

to depend on a specialized area of the brain, but rather on neural networks formed by vision. This now makes it possible to explain why even newborns and untrained wild animals have a number sense,” Nieder explains.