We know that prior knowledge is very, very important for learning, also for skills. Learning a new skill is easier when it is related to an ability we already have. For example, a trained pianist can learn a new melody easier than learning how to hit a tennis serve. Scientists from the Center for the Neural Basis of Cognition (CNBC) — a joint program between Carnegie Mellon University and the University of Pittsburgh — have discovered a fundamental constraint in the brain that may explain why this happens. This research doesn’t tell that many new things from a cognitive perspective on learning, but does shed some new insights on the technical reasons behind the importance of prior knowledge.
From the press release (bold by me):
Published as the cover story in the August 28, 2014, issue of Nature, they found for the first time that there are limitations on how adaptable the brain is during learning and that these restrictions are a key determinant for whether a new skill will be easy or difficult to learn. Understanding the ways in which the brain’s activity can be “flexed” during learning could eventually be used to develop better treatments for stroke and other brain injuries.
Lead author Patrick T. Sadtler, a Ph.D. candidate in Pitt’s Department of Bioengineering, compared the study’s findings to cooking.
“Suppose you have flour, sugar, baking soda, eggs, salt and milk. You can combine them to make different items — bread, pancakes and cookies — but it would be difficult to make hamburger patties with the existing ingredients,” Sadtler said. “We found that the brain works in a similar way during learning. We found that subjects were able to more readily recombine familiar activity patterns in new ways relative to creating entirely novel patterns.”
For the study, the research team trained animals to use a brain-computer interface (BCI), similar to ones that have shown recent promise in clinical trials for assisting quadriplegics and amputees.
“This evolving technology is a powerful tool for brain research,” said Daofen Chen, program director at the National Institute of Neurological Disorders and Stroke (NINDS), part of the National Institutes of Health (NIH), which supported this research. “It helps scientists study the dynamics of brain circuits that may explain the neural basis of learning.”
The researchers recorded neural activity in the subject’s motor cortex and directed the recordings into a computer, which translated the activity into movement of a cursor on the computer screen. This technique allowed the team to specify the activity patterns that would move the cursor. The test subjects’ goal was to move the cursor to targets on the screen, which required them to generate the patterns of neural activity that the experimenters had requested. If the subjects could move the cursor well, that meant that they had learned to generate the neural activity pattern that the researchers had specified.
The results showed that the subjects learned to generate some neural activity patterns more easily than others, since they only sometimes achieved accurate cursor movements. The harder-to-learn patterns were different from any of the pre-existing patterns, whereas the easier-to-learn patterns were combinations of pre-existing brain patterns. Because the existing brain patterns likely reflect how the neurons are interconnected, the results suggest that the connectivity among neurons shapes learning.
“We wanted to study how the brain changes its activity when you learn, and also how its activity cannot change. Cognitive flexibility has a limit — and we wanted to find out what that limit looks like in terms of neurons,” said Aaron P. Batista, assistant professor of bioengineering at Pitt.
Byron M. Yu, assistant professor of electrical and computer engineering and biomedical engineering at Carnegie Mellon, believes this work demonstrates the utility of BCI for basic scientific studies that will eventually impact people’s lives.
“These findings could be the basis for novel rehabilitation procedures for the many neural disorders that are characterized by improper neural activity,” Yu said. “Restoring function might require a person to generate a new pattern of neural activity. We could use techniques similar to what were used in this study to coach patients to generate proper neural activity.”
Abstract of the research:
Learning, whether motor, sensory or cognitive, requires networks of neurons to generate new activity patterns. As some behaviours are easier to learn than others, we asked if some neural activity patterns are easier to generate than others. Here we investigate whether an existing network constrains the patterns that a subset of its neurons is capable of exhibiting, and if so, what principles define this constraint. We employed a closed-loop intracortical brain–computer interface learning paradigm in which Rhesus macaques (Macaca mulatta) controlled a computer cursor by modulating neural activity patterns in the primary motor cortex. Using the brain–computer interface paradigm, we could specify and alter how neural activity mapped to cursor velocity. At the start of each session, we observed the characteristic activity patterns of the recorded neural population. The activity of a neural population can be represented in a high-dimensional space (termed the neural space), wherein each dimension corresponds to the activity of one neuron. These characteristic activity patterns comprise a low-dimensional subspace (termed the intrinsic manifold) within the neural space. The intrinsic manifold presumably reflects constraints imposed by the underlying neural circuitry. Here we show that the animals could readily learn to proficiently control the cursor using neural activity patterns that were within the intrinsic manifold. However, animals were less able to learn to proficiently control the cursor using activity patterns that were outside of the intrinsic manifold. These results suggest that the existing structure of a network can shape learning. On a timescale of hours, it seems to be difficult to learn to generate neural activity patterns that are not consistent with the existing network structure. These findings offer a network-level explanation for the observation that we are more readily able to learn new skills when they are related to the skills that we already possess.