Skills are easier to learn if associated to skills we have
Learning new skills and acquiring knowledge is essential to the growth and development of modern society, but there is still uncertainty regarding the biological mechanisms behind this occurrence.
Recent findings by researchers at the joint Carnegie Mellon and University of Pittsburgh Center for the Neural Basis of Cognition (CNBC) suggest that it is easier to learn new skills if people can associate them with skills they already have.
The researchers include Carnegie Mellon’s Byron Yu, an assistant professor of electrical and computer engineering and biomedical engineering; Steven Chase, an assistant professor of biomedical engineering with a courtesy appointment in electrical and computer engineering at Carnegie Mellon and researcher at the CNBC; as well as University of Pittsburgh researcher Patrick Sadtler, a bioengineering Ph.D. candidate.
The team is also comprised of Carnegie Mellon’s Matthew Golub, Stanford University’s Stephen Ryu, and University of Pittsburgh’s Aaron Batista, Elizabeth Tyler-Kabara, and Kristin Quick.
The team used brain computer interfaces (BCIs) to translate neural activity into something more easily defined, Chase explained.
“BCIs take neural activity and map it directly to something else. It’s often a robotic arm, but in this case it was a computer cursor,” Chase said. Animal subjects were trained to move a computer cursor to targets on a screen while the researchers monitored their motor cortex, the section of the brain responsible for voluntary movements.
The movement of the cursor to the target required the subjects to produce specific patterns of neural activity. If the subjects were able to efficiently move the cursor, it signaled that they had produced the correct neural activity pattern.
“We designed mapping between neuron activity and cursor movement that would either take existing patterns in neural activity and use them in new ways to drive the cursor in different directions or generate new patterns of neural activity that had never been used before,” Chase said. “The whole goal of the experiment was to try to make a prediction about what sorts of things are learnable and what sorts of things are not.”
The research showed that certain neural activity patterns were easier than others for subjects to learn. Subjects generally had an easier time learning patterns that were comprised of pre-existing patterns. Patterns consisting of totally new patterns were much harder for subjects to learn.
“You can change existing patterns and map them to cursor movement,” explained Chase. “What you can’t do is generate new patterns — or at least you can’t do it very quickly.”
The possible applications of this research are extensive, including improved motor learning rehabilitation and more efficient, functional BCIs. Chase explained that motor learning is a topic that necessitates further study.
“Motor learning is one of those big open questions in neuroscience research; we really don’t understand what it is and how it occurs,” Chase said.
“We normally don’t think about motor control very much because we’re extremely good at it, but when it goes wrong it’s devastating.”
Spinal cord injury, stroke, and Parkinson’s disease are all widespread afflictions associated with faulty motor control. “Stroke is an instance where we hope that we will be able to learn to overcome whatever deficits there are. It’s a question of learning, and if we knew the right way to derive the learning, we could help people recover from strokes better,” Chase said.
He further commented on the ability of interdisciplinary collaboration to prompt innovation. “This was a true collaborative project that brought together techniques from a bunch of different fields, including neurobiology, engineering, and machine learning,” Chase said.
“The end goal of all of this research is to really understand motor learning at a fundamental level and take that information to perform stroke rehabilitation more effectively and design better BCIs for those who can’t communicate.”
The research was funded by the National Institutes of Health, the National Science Foundation, and the Burroughs Wellcome Fund. The paper was published in the Aug. 28 issue of Nature.