A brief history of Neuroscience and Artificial Intelligence

How understanding the human brain can advance AI

As humans, our intelligence is what distinguishes us from other species. We study and try to understand how we think; “That is, how a mere handful of matter can perceive, understand, predict, and manipulate a world far larger and more complicated than itself .” [1].

Human intelligence is a combination of cognitive processes rather than an individual process or entity. These Cognitive processes include perception, memory, reasoning, problem-solving, and learning [2].

Meanwhile, Artificial Intelligence is the study and analysis of computational agents that act intelligently. An intelligent agent learns from experience, takes actions that are appropriate for its goals, and is flexible to change. Thus, simulating the cognitive processes of humans.

So in a sense, Artificial Intelligence attempts to understand and build intelligent entities. For that, we need to further understand human intelligence, more specifically Neuroscience.

What is Neuroscience?

Neuroscience is the study of the nervous system, its development, structure, and functions. Particularly the source of consciousness, the brain.

How Neuroscience came to be?

In 1861, while Paul Broca was studying speech deficit in brain-damaged patients, he demonstrated the existence of localized areas of the brain. Different areas of the brain control specific cognitive functions. The brain was known to consist of neurons, but neurons couldn’t be observed at the time. Not until 1873, with a microscope Camillo Golgi was able to see the outline of an individual neuron, using a staining technique he developed. This technique was a breakthrough in neuroscience.

Subsequently, Santiago Ramón y Cajal used Golgi’s technique to demonstrate the “neuron theory”, stating that the nervous system consists of individual nerve cells or neurons.

Understanding Neurons

With such breakthroughs in neuroscience, we now have an idea of the mapping between areas of the brain and the parts of the body that they receive sensory input from or control [1]. To illustrate, imagine a part of your brain, now imagine this part has a path between it and your eyes. Through this path, the visual signals travel to give you the necessary information to see.

There has been inventions and development such as electroencephalography (EEG) by Hans Berger and functional magnetic resonance imaging (fMRI) by Seiji Ogawa. These inventions allow neuroscientists to measure and get detailed images of brain activity, enabling the mapping of cognitive functions.

Yet, there is no full understanding of how cognitive functions actually work, or how other localized areas can take over the functions of one damaged area.

Also, we don’t know how individual memory is stored. One theory suggests that bits and pieces of the memory are distributed across millions or even billions of neurons. A more scientifically credible theory suggests that far fewer neurons are involved, each neuron would contain a part of the memory [3].

In other words, “a collection of simple cells can lead to thought, action, and consciousness” [1].

Human Brains and AI

As humans, we perform some tasks subconsciously. When walking we usually don’t have to consciously think about putting one foot forward and the other foot back. For AI, this simple task can require a lot of computation. This is known as the Moravec’s paradox, stating that AI can perform reasoning and logical tasks such as maths with ease, while sensorimotor tasks are much harder to achieve. As humans we achieve sensorimotor tasks subconsciously, but perhaps at the time it was hard to a program or a write a set of rules for AI to achieve sensorimotor tasks [4].

Nonetheless, in recent years AI can perform unconscious sensorimotor human skills, such as vision and moving about. There’s a much better understanding of the human brain now. Respectively, the way AI researchers and scientists understand and approach these tasks are different than perhaps how Moravec would.

Conclusion

Human brains are truly fascinating, with the ability to learn, link one memory to the other, imagine a scene, and replay previous memories. And in a sense, understanding how are these tasks are being done by the brain will allow us to build intelligent systems.

“We get this virtuous cycle between neuroscience, cognitive science, and AI because we are trying to study the same thing, which is intelligence”. — Jess Hamrick, Research Scientist at DeepMind