When we look at a tree branch moving in the wind, or contemplate the small details in a painting, thousands of nerve cells in our brains work together to transform what our eyes see into a clear and understandable image.
Hot Topics
Tropical city in China enjoys success thanks to coffee industry
Al-Nasr enters the Guinness World Records through the Gulf region
But scientists wondered: Are there fixed limits to the amount of information the brain can process, such that adding more neurons after a certain point is no longer beneficial? Or does increasing the number of cells mean that the brain can always handle more information?
For a long time, mathematical models suggested a limit that neural networks could not exceed. But a recent study , published in the journal Science Advances, found that this limit is not as absolute as previously thought, and that neural networks in mammalian brains can increase their ability to represent information as their number of cells increases.
“This issue had to be resolved based on real data,” said Hideaki Shimazaki, associate professor at the Graduate School of Informatics at Kyoto University in Japan, in an exclusive statement to Al Jazeera Net. “In recent years, technologies capable of recording the activity of a thousand or more neurons at once have made it possible to observe how information increases directly as the number of cells involved in the process increases.”
For thirty years, the prevailing assumption in many previous studies has tended toward pessimism; that nerve cells are not completely isolated computer chips, but rather interconnected biological units immersed in a chemical environment that makes them susceptible to mutual influence.
When a cell experiences random fluctuations or noise in its activity, this noise is automatically transmitted to neighboring cells, and with the accumulation of this noise, increasing the number of neurons that accumulate information is not beneficial.
To test this, the team relied on a study of the primary visual cortex in mammals, specifically by reanalyzing detailed experimental records of approximately 20,000 neurons in the brains of five mice. The mouse visual cortex is a well-established biological model for understanding the general mathematical principles of image processing in mammals. "If you present the same image to the animal several times, the nerve cell in its visual cortex does not respond in the same way each time. This variation from attempt to attempt is what we call 'neural noise'. When the image changes slightly, the average activity of the nerve cells also changes; this change is the signal that the brain uses to distinguish between the two images."
Mathematical analysis by the research team revealed that the shared oscillations between cells do not coalesce into a random mass that blocks the visual pathway, as previously thought. Hideaki adds, "The results clearly show that all five mice fall into the 'continuous information growth' category; the shared noise slows down the growth of information, but it doesn't stop it as long as the observed rules continue to function."
Tags:
science
