View Full Version : The cerebellum and trial-and-error learning

05-05-18, 17:00
A team of researchers at Johns Hopkins University has unearthed exciting new details about how cerebellar Purkinje cells (link is external) master complex movements through a trial-and-error learning process in which each error that occurs during practice helps the cerebellum improve fine-tuned motor coordination. This paper, “Encoding of Error and Learning to Correct That Error by the Purkinje Cells of the Cerebellum (link is external),” was published online April 16 in the journal Nature Neuroscience.

The most significant aspect of this study is that senior author Reza Shadmehr (link is external), professor of biomedical engineering and neuroscience at the Johns Hopkins University School of Medicine, and colleagues have pinpointed how Purkinje cells work. This research adds fresh insight and empirical evidence to support landmark papers from the 20th century that speculated about what the mysterious Purkinje cells were actually doing such as "The Cerebellum as a Neuronal Machine (link is external)," (John Eccles et al., 1967), "A Theory of Cerebellar Cortex (link is external)" (David Marr, 1969), and "A Theory of Cerebellar Function (link is external)" (James S. Albus, 1971).

A May 3, 2018 press release about the cutting-edge Purkinje cell research being conducted at Johns Hopkins Medicine, “Decoding the Brain’s Learning Machine (link is external),” describes the cerebellum as a “learning machine” within the mammalian brain. Their most recent study on monkeys shows that Purkinje cells make predictions and learn how to master complex tasks by constantly correcting little mistakes. With practice, errors decrease and precision increases. Over time, this results in what I call "superfluidity (link is external)."

As an athletic example of Purkinje-based cerebellar mastery, the statement says: “When learning to shoot a basketball, people usually miss many times before getting one shot through the hoop. As the arm moves, the cerebellum makes predictions about the consequences of the action. When the prediction does not match reality — that is, the ball misses the hoop — the cerebellum receives feedback from the eyes and the arm to learn from the error, fine-tuning factors such as aim, force and release to make a basket. This trial-to-trial learning from error produces gradual improvements in performance” The same process of trial-and-error cerebellar learning occurs in every sport and helps to explain why practice makes perfect. Notably, having an "eye for the ball" is also directly linked to the cerebellum and our vestibulo-ocular reflex (link is external) (VOR).

Purkinje Cells Master Trial-and-Error Learning Via “Simple” and “Complex” Spikes
According to the latest research from Johns Hopkins, Purkinje cells communicate via two types of electrical signals called “simple spikes” and “complex spikes.” Simple spikes reflect predictions that Purkinje cells are making about optimal movements. Complex spikes reflect information sent back to a cluster of Purkinje cells to let them know if there was an error in the prediction or if the timing and speed of a movement were accurate. For each unique motor skill, specialized Purkinje cells all receive the same error message and appear to work in concert when making appropriate motor corrections based on a specific mistake.

"You can think of the simple spikes as the 'student' that makes a prediction and the complex spikes as the 'teacher' that provides feedback," Shadmehr said in a statement.

According to Shadmehr, "One of the advantages of the cerebellum's architecture is that it protects memories.” Once the muscle memory required to do something that is primarily cerebellar (like riding a bicycle) is hammered and forged into your Purkinje cells, the simple and complex spikes associated with coordinating this motor skill become hardwired.

Touch-Typing Relies on Purkinje-Based Cerebellar Learning
Mastering the skill of “touch-typing” without looking at the keyboard is a perfect everyday example of Purkinje-cell based trial-and-error learning. The next time you're at a keyboard, place the tips of your left and right index fingers on the little ridges found on the “F” and “J” keys respectively. Putting your hands in this ‘home position' guides your cerebellum to automatically know where all of your fingers are located in relation to specific keys. This creates predictability via cerebellar proprioception.

Can you type "the quick red fox jumped over the lazy brown dog" without looking at the keys? If so, how many times in one minute can you type these 10 words? This sentence has all the letters of the alphabet and is a gold standard for testing someone's touch-typing ability. With practice, your fingers will implicitly learn where all the keys are located and you can master typing upwards of 40 words per minute (WPM), which is considered an "average" speed.

Based on the latest research from Johns Hopkins on trial-and-error learning, one could speculate that the more you practice correcting errors while typing "the quick red fox jumped over the lazy brown dog" that clusters of Purkinje cells are being fine-tuned to master this skill via simple and complex spikes.

Interestingly, once the implicit ability of your cerebellum to predict where the appropriate letters are located without looking down, most touch-typists eventually lose their cerebral knowledge and declarative memory of being able to say where each letter is located on a QWERTY keyboard. Anecdotally, I can corroborate this phenomenon. As someone who has been touch-typing since high school, in order to describe the two keys with braille-like ridges, I had to look down at the keyboard to verify that "F" and "J" are where I intuitively place the tips of my index fingers every time I begin to type.

If someone asked you where the home keys (A, S, D, and F for the left hand and J, K, L, and semicolon for the right hand) are located on a QWERTY keyboard, would you know the answer? (Some trivia: The name of this standardized keyboard layout comes from the sequence of six letters on the top, upper left row.)

Below is a 2-minute video by researchers at Vanderbilt University that illustrates how we can learn to type automatically without any explicit knowledge of where specific letter keys are located:

Damage to the structure or functional connectivity of the cerebellum can profoundly impact someone’s ability to coordinate and execute motor movements fluidly. Diseases affecting the cerebellum typically result in various forms of ataxia and dysmetria. One of the potential applications of the latest Johns Hopkins research on Purkinje cells and the cerebellar process of trial-and-error learning could be the development of more refined methods for diagnosing cerebellar abnormalities.
Another paper, "Locomotor Activity Modulates Associative Learning in Mouse Cerebellum(link is external) (https://www.nature.com/articles/s41593-018-0129-x)," was also published April 16 in Nature Neuroscience. This studyfound that mice who run faster on a treadmill learn implicit-memory tasks more quickly than their slower running counterparts. For more see, "Why Does Running Faster Speed Up Learning in the Cerebellum? (https://www.psychologytoday.com/us/blog/the-athletes-way/201804/why-does-running-faster-speed-learning-in-the-cerebellum?amp)"



Very interesting, makes me think about how the larger cerebellum in Homo sapiens possibly helped them become the dominant hominid species. They probably learned better from trial-and-error learning than neanderthals.