Brain Networks: Blog 2: Skill-Learning Networks
by Brandon Ascencio UA ‘23, Brandy Eggan, and Jim Stellar
In a prior blog on the default mode network, we talked about how this network comes online when you do not have a task and are engaged in mind-wandering. Remember, brain networks establish themselves not only to help the execution of something of which you are aware, but also as an unconscious process – maybe a bit like in the processor in your computer that is continually working even if you are simply reading this blog. In this blog we look at active skill-learning where research has shown that the brain again utilizes an interconnected network of various brain areas. It appears that his kind of network communication is critical for mastery of a task. More generally, we also believe it is important that the communication occur between brain areas that serve conscious cognitive functions and lower brain areas that serve unconscious skill-learning – the kind of integrated learning that happens during experiential education activities like an internship.
To back up a bit, in any basic anatomy or neuroscience course you first learn that there is a brain region associated with a behavioral function, e.g. a specific place on the motor strip and the contraction of the rectus femoris muscle (large muscle on the front of thigh) that it causes extension of your knee as in kicking a soccer ball. If you were to take an electrode and with very precise aim insert it into the brain at that exact spot, turning on this electrode would cause that kicking motion. Sounds simple, right? Unfortunately, as both a soccer player and a young aspiring neuroscientist, I (BA) have come to realize that things are not as easy to understand as one may think.
For one thing, the movement of an individual soccer kick occurs in the context of the game, which is also played as a team. So too the motor movement leg center discussed above communicates with other brain regions. We might call this “brain team” the skill-learning network. And this brain network learns with real-world practice. If practice were not required, I (BA) would be able to stand in front of the goal on a penalty shot, spin around five times with my eyes closed, kick, and the ball would dart right through the goalie’s hands. Too bad that is not the case. Practice is an essential piece of the skill-learning network.
To go deeper on this analogy, a kick is actually composed of many various elements. You have postural balance (which is coordinated by the cerebellum as well as premotor areas); visual cues based on where your opponents and teammates are (occipital cortex, particularly the dorsal movement pathway); the planning of the exact desired trajectory and orientation of the body (supplementary motor cortex); the curve of the ball which requires the recruitment of many accessory muscles in the leg (back to motor cortex); the sound of a defender running up behind you (auditory cortex); the patterned movement you have unconsciously ingrained in your brain during practice – left foot, right foot, left foot, drop the right hip, transfer the weight, use your power, aim with the shoulders, follow through (basal ganglia) – and those are just a few.
Below is an image from a recent review paper on athletic and musical skill-learning that shows brain activity (fMRI scan) during motor skill execution. We use this figure to show that just like with the default mode network, one sees activity across multiple areas of the brain. They include the major frontal motor cortical areas (right side of the figure), but also include sensory areas of the brain (left side). Also not shown is the involvement of many deeper brain regions below the cortical surface. The entire network is critical. Remember the one electrode we just discussed that is placed on the motor cortex to cause leg extension? That is clearly not what is going on in the brain network during skill-learning or execution. We would need thousands of electrodes to recreate the brain-wide network actually behind this seemingly simple task as the activity image below reveals.
It is not surprising that this figure shows major changes in the activity of frontal brain areas, and in practiced skilled athletes or musicians show anatomical changes as well. What may be more surprising is to see the activation of sensory areas as part of the network. In addition, there are other complexities at a subregion level. For example, when you focus on just one of the frontal areas (e.g. the electrode discussion above) and attach a function to it (e.g. leg movement), you will find a difficulty in simply saying that “it does this.” Rather it is more of a network itself and it may be better to say “it does this when activated by this other region but not when activated by that other region” – the kind of complexity that a network brings.
Let’s delve a little deeper into the dissection of the skill-learning network that is marked by a gradual decrease in reaction time and errors as the skill is attained. This is a field in itself and there have been many studies of the effects and adaptations of the skill networks and. An example below is from one study involving mice and the influence of visual spatial cues on the motor skill learning of three specific non-motor areas: the hippocampus (memory), the anterior cingulate cortex (error detection, etc.), and the primary visual cortex (basic perception). Mice were used not only because these brain areas parallel humans functioning, but also because powerful techniques of brain manipulation can be used in mice. The key technique here (as explained in the paper (and as shown below) is optogenetics.
Here certain brain areas were genetically modified to express proteins on the surface of neurons that can be activated by experimenter-delivered light (shown as blue above). In this study, the optogenetic technique was used to suppress the regional neural activity in the moment. Importantly, this regional optogenetic suppression can be done while the animal is learning a skill. Aside from its key theoretical basis, this experimental work represents the kind of network function dissection that is needed. We simply must get more precisely at how these networks operate dynamically in a way that overall fMRI brain activities studies cannot.
What did they find? Over the course of a three-week operant skill learning experiment, the occasional suppression of the primary visual cortex reduced mouse task performance accuracy on those trials but did not change the rate of learning. The occasional suppression of the anterior cingulate cortex produced elevated error rates on the task at the moment and these showed up a bit in all trials, particularly in later trials of the 3 week training session when higher skill levels were attained by the mouse. Finally, occasional suppression of the hippocampus seemed to only modestly reduce performance and did so at the same level regardless of the attainment of skill through cumulative practice. These three regions showed a parallel involvement of non-motor areas in a network in the mouse, which can lead to future discoveries in regards to human skill learning. We thought it was important that the motor skill-learning in the study above took some time to develop over the three weeks of the study. We thought it was important that the brain areas were only intermittently suppressed in their activity as it gives a different view from when the brain areas are destroyed as in many previous studies in neuroscience history.
Looking ahead, we think this kind of research shows how each network (skill-learning, default-mode) might be studied in animals to gain insights into how they might work in humans, which then potentially could be verified with non-invasive techniques of functional magnetic imaging as used in the first study. Finally, related to this blog, we think that this kind of research could provide insights into how brain networks of skill-learning might operate in a broader learning context, like when a pre-medically oriented college student (like BA) goes on an internship and immerses himself in a medical environment or that of a research lab, and he learns hands-on skills to accompany his academic learning – just like he learned to kick a soccer ball and bend its trajectory.
