UW Research Provides Basic Structure of Neural Networks in Mammalian Brain
Two researchers at the University of Wyoming decided to pick each other’s brains, so to speak. Specifically, they looked at the importance of the frontal cortex, the part of the brain used in decision making, expressive language, and voluntary movement.
And the two scientists learned that a recurrent neural network structure, or RNN, is responsible for these functions.
“This RNN receives inputs from emotional regions of the brain and sends outputs to the motor cortex, the part of the brain responsible for voluntary movement,” said Qian-Quan Sun, professor of zoology and physiology at UW. “In the field of artificial intelligence, computer scientists have designed various artificial neural networks, including RNNs, that effectively solve problems, such as language translation and object recognition, by simulating the neural network in the brain of mammals.
“This article provides a basic structure of neural networks in the mammalian brain. This basic structure will guide us in the study of behavioral strategy,” continues Sun. “Once more details have been acquired, we can translate it into an artificial neural network, using it to solve real world problems.”
Sun, director of the Wyoming Sensory Biology Center of Biomedical Research Excellence at UW, is the lead author of an article titled “A Long-Range Recurrent Neuronal Network Linking the Emotion Regions with Somatic Motor Cortex” which was published today. hui (Tuesday) in Cell Reports. The open access journal publishes peer-reviewed articles across the life science spectrum that report new biological knowledge.
The first author of the article is Yihan Wang, holder of a doctorate. studying in the doctoral program in neuroscience at UW, Beijing, China. The research was funded by grants from the National Institutes of Health.
Artificial RNNs are important deep learning algorithms that are commonly used for ordinal or temporal lobe problems, such as language translation, natural language processing, speech recognition, and captioning. images, explains Sun. An RNN recognizes the sequential characteristics of data and uses models to predict the next likely scenario. RNNs are built into popular apps like Siri, Google Voice Search, and Google Translate.
The biggest surprise is that RNNs not only exist in our brains, but are built with a much more delicate function and, yet, very efficient in processing sequential inputs. In general, cortical neurons are spatially reciprocal and intertwine. However, Wang’s data not only showed that RNN exists in the most important part of the brain – the frontal cortex – but in addition, this network is less complex than we thought and mostly unidirectional. This is a big surprise to us, because it tells us that this network may be in charge of unique functions compared to others. “
Qian-Quan Sun, Professor of Zoology and Physiology at UW
Sun and Wang analyzed the brains of mice for laboratory research. Different strains of genetically engineered mice have provided both with the ability to tag specific types of neurons with fluorescent proteins that follow brain connections – and to monitor the activities of specific neurons with inherently fluorescent markers.
The research has many real-world implications, according to Sun.
“First, now that we know this important building block, the work will help further decipher how our brains make decisions,” he says. “Second, it will help uncover other similar RNNs in other parts of the brain. It will help researchers use computer simulations to predict how our brain encodes short-term memory and how it can be used. , specifically for this study, it will help us understand how emotions, such as fear and anxiety, regulate our movements. “
The content and research approach used by Sun and Wang is expected to have very broad interests among artificial intelligence researchers, biologists, computer modelers and neuroscientists, Sun said.
“The accurate connection map can also help us understand the cause of neurological and psychiatric disorders where there are problems with regulating emotions or voluntary movements,” Sun said. “However, before this discovery can have wider applications, there are a lot of details – such as how the local inhibitory network refined the RNN and how different components underlie specific emotional states – that have yet to be worked out. understood.”
Wang’s goal is to sort out these details in his thesis work, Sun says.
Wang, Y., et al. (2021) A long-range recurrent neural network connecting emotional regions to the somatic motor cortex. Cell reports. doi.org/10.1016/j.celrep.2021.109733.