Watching Neurons Learn
(Inside Science) -- Researchers in Pittsburgh, using a brain-computer interface, have shown why learning something similar to what you already know--a repertoire of previous knowledge--makes learning new things easier. Learning unfamiliar ideas or behavior is more difficult.
While that sounds self-evident, the researchers have actually watched it happen in animal brains to learn how it works.
“It makes perfect sense, but now we can see it in terms of the brain itself instead of just in terms of looking at someone’s behavior,” said Aaron Batista, assistant professor of bioengineering at the University of Pittsburgh. The team also included researchers from Carnegie Mellon University, in Pittsburgh, and Stephen Ryu, a consulting engineer at Stanford University and a neurosurgeon at the Palo Alto Medical Foundation in California.
“We see the neuromechanics of it. The entire field of neuroscience tries to reduce behavior to the activity of neurons, and we’ve been able to do that to learning in ways that are novel,” Batista said.
The findings were published in Nature. The study has applications for people who have survived a stroke, or developed other forms of brain damage.
The work also shows the restrictions on learning.
Byron Yu, an assistant professor of electrical and computer engineering and biomedical engineering at Carnegie Mellon said considerable research had gone into the notion that anything is learnable, but this study shows there are limits to that hypothesis.
In the eight-month study, researchers placed electrodes under the skulls of monkeys to monitor the activity of the brain’s neurons. They linked the electrodes to a device similar to one used to assist quadriplegics and amputees. They mapped the activity in the neurons and matched them to the motion of the cursors.
The signals moved a cursor on the screen.
The monkeys sit with their arms at the side. They have learned, through operant conditioning, that they can move the cursor by thinking about it just as if they were moving their arms and hands to do it. Once they have learned that, the pattern is changed and the monkeys have to learn new thoughts to move the cursor. It is as if, Batista said, they had grown a second spinal cord and were relearning how to move the cursor in a new way. If the pattern matched a previous method, they did it more easily.
If there was no earlier established pattern, it took them longer.
The work has practical implications for people with various brain disorders or injuries associated with abnormal brain activity patterns, such as strokes, Yu said.
Because the patient can plot the brain activity on the screen he or she can adjust behavior to show normal activity patterns again, something like watching a real-time readout of a car’s gas mileage to increase fuel efficiency. They would modify what they are doing based on what they see on the screen.
Importantly, said John D.E. Gabrieli, a cognitive neuroscientist at MIT, it shows why a person cannot learn or would have difficult time learning something.
“We have to learn things that are hard to learn all the time,” Gabrieli said. “And those situations where you have to learn a novel, very difficult skill, would not be represented...by the way these neurons were fired.”
There would be other forms of interaction. Other mechanisms in the brain, in other places, such as the ganglia or cerebellum, would kick in to help you learn novelty, he said. That would produce different neuron patterns.
“It’s going to be harder. It’s not in the brain’s repertoire. One example would be in learning a language,” he said.
If you know Spanish, learning Italian would be fairly easy because of the similarities in vocabulary and structure. Learning Mandarin Chinese would be hard because of the lack of similarities.
“You would be starting almost from scratch," Gabrieli said. ”You need another strategy.”
The technology might also someday lead to new ways to teach, but Batista said society would be better served by sticking to diseases first.