How neuroscience and artificial intelligence can help each other
Over the past decades, brain research has revealed a lot about the physical structure of the brain and how the nervous system perceives and processes information. But much remains to be discovered.
The brain is not a machine
Machine learning systems are better than humans at finding complex and subtle patterns in huge amounts of data.
Artificial neural networks
Artificial neural networks, the most common approach to machine learning, are interconnected networks of digital processors that take input data, process measurements of that input data, and generate output data.
If you want a machine learning system to display text “ This is a cow”, when she is shown a photo of a cow, first you need to provide her with a huge number of different photos of cows. If you show this system a picture of a cat, it will only know that it is definitely not a cow, but it will not be able to tell what it really is. Since the brain and machine learning systems use fundamentally different algorithms, each of them is superior to each other in some aspects. For example, the brain is more effective at making decisions in unfamiliar situations or rapidly changing conditions. Beyond the discoveries of how the brain works, it is still unclear what processes in the brain can work well as machine learning algorithms. One way to find out is to focus on research in both directions, from brain science to artificial intelligence.