The Science of Everything Podcast - Episode 133: Motor Control

Episode Date: January 1, 2023

A journey through the complex network of regions controlling the human motor system, beginning with the spinal cord and its central pattern generators, and working up through the primary motor cortex, the premotor cortex, the posterior parietal cortex, the cerebellum, and the basal ganglia. I discuss the computational roles of each part of the motor control hierarchy, focusing on what functions are performed and what information is represented in each unique brain region. Overall I emphasise the complex interaction between top-down and bottom-up feedback in controlling muscle movement and executing complex motor tasks. Recommended pre-listening is Episode 132: The Muscular System, and Episode 38: Neurons and Synapses.   If you enjoyed the podcast please consider supporting the show by making a PayPal donation or becoming a Patreon supporter. https://www.patreon.com/jamesfodor https://www.paypal.me/ScienceofEverything

you're listening to The Science of Everything podcast, episode 133, Motor Control. I'm your host, James Fodor. In this episode, we're going to continue from where we left off in the previous episode, which is the prerequisite listening, by the way, episode 132, the muscular system. We're going to pick up from where we left off there and talk about how muscles and the contraction thereof are controlled by the central nervous system. So we're going to talk about some of the computational difficulties with control controlling the motor system and some of the solutions that have been devised by evolution to resolve these.

We're going to go through and talk about the spinal cord and how it contains motor programs that produce patterns of muscle contractions to produce more complicated motions. We're going to talk about locomotion in particular and how that is mediated by the spinal cord and parts of the midbrain and brain stem. Then we're going to talk about the role of the cortex, including the parietal cortex, the primary motor cortex and the pre-motor cortex and look at some of the different ways in which aspects of motor control

and motion representation are encoded by neurons there. And we'll also discuss the cerebellum, which is the bit of the brain at the back of the skull in its role in performing specific types of computations for fine tuning and giving feedback to motor control. And we'll conclude by looking at briefly at the basal ganglia and its role in action selection. So I already mentioned that the prerequisite episode for this is 132 on the muscular system.

Also, it would be good to have listened to an episode or two on the nervous system. So episode 38 on neurons and synapses would be a good example to get a bit of background for what we're going to talk about today. So there's quite a bit to get through. Let's jump right in and start talking about some of the computational factors or problems and difficulties that need to be overcome with respect to motor control. So motor control is concerned with how our nervous. nervous system controls the contraction of muscles, including the timing of contraction, which muscles

contract and how intensely they contract, in order to produce all of the motions that we need to survive. So this includes locomotion, which is walking, includes moving the hands and the digits and the arms, moving face muscles, all of that sort of stuff. The focus here will be on control of skeletal muscles, so that's what we also focused on in the previous episode. The nervous system also controls the activation of smooth muscles, for example, that are involved in digestion of food, but we're not really going to talk about that too much here. Naively, we might think that motor control is a relatively simple problem, right? I mean, effectively all we have to do is get a series of instructions about which muscles to

contract and when to contract them, and then you just sort of do that, right? You just sort of play it out, like a marionette puppet or something that executes the instructions as required. But it turns out that it's actually much, much more complicated than that. So let's talk about some of the things that make motor control difficult. First, motor control is not a series of static problems to solve. Like you can just replay the same motions over and over again, like a marionette puppet or something like that. Motor control is highly complex and dynamic problem.

Some of the things that make it difficult include the non-linearity of the problem. So there are many, many different limbs. There's several hundred muscles in the human body. And all of them are attached to bones, which then pull limbs and other parts of the body around. and it's not just an issue of each muscle is connected to a new limb because if you think about your arm, for example, you've got your upper arm, which is then attached to your lower arm, which is then attached to a series of bones in your hand and then moving to the finger, there's several different bones in your finger. So if I want to move the tip of my pinky to a particular location, there's many degrees of freedom that I can use to do that. I can bend my finger, I can bend my hand, I can move my shoulder, I can move my whole arm about my shoulder, or I can move my lower arm, bend that at the elbow, and all of those allow me to sort of move the same parts of my body, that is ultimately my fingers and the rest of my hand. And the thing is, they're not independent of each other. Those movements are not independent of each other. If I say bend my elbow, that moves my whole hand. And so if I want to think about where I want to position my fingers, I have to think about how all of these motions interact with each other. And that becomes a very highly non-legged.

linear problem. So it's not just an issue of specifying where I want each of the joints, because there's highly complicated interactions between them, which require very complex calculations. So that's one problem, the non-linearity. I kind of mentioned another, which is redundancy. So the fact that there's many degrees of freedom, many joints that can lead to the same position, but as a result of bending my joints in a different way, that means that there are many possibilities and many sort of variables that need to be considered. So it's not just like there's one way to achieve one outcome. There's many ways that all need to be considered. And of course, nervous system has to consider things like the amount of energy or the time that it's going to

take in order to execute a particular motion. So that has to be factored in. Another big difficulty is inherent delays in the system. And a good way to think about this is if you've ever tried to get the temperature right in a shower, which has the hot and the cold taps, right? That's not too difficult if the temperature responds very quickly, if I turn on the hot tap and I get pretty quick feedback. It's relatively easy to... Where it becomes difficult is if there's a significant delay between when I change one of the taps and when I feel a change in the temperature. The longer the delay, the harder it is to get the balance right, because essentially you don't know what the effect has been of the action that you've just taken. You have to wait and see

and try to judge what the effect is and then how much more you need to adjust it. So delays in feedback make the problem much more difficult to solve. And there are significant delays in feedback of motor control because there's time taken from the nerve signals to be sent down to the actuator muscles and then time for the actual movement to occur. And then signals need to be sent back. So that all leads to delays which significantly slows the overall process and therefore makes it more difficult to process that feedback. A final problem is that the nervous system is inherently noisy. So if we try to move a contractor muscle a certain way or move a limb in a certain direction, you know, we generally get the outcome we expect, but we make

mistakes. Sometimes you don't realize how heavy an object is or move your arms slightly too far and you accidentally knock something. Like we do that actually quite frequently, accidentally knock something or step too far or stub a toe or whatever. Part of the reason for that is, because of the noisiness of the nervous system and we just kind of make mistakes and don't get things right all the time. And that's more, that's exacerbated if you sort of look at it at a fine, grained level. And that makes it even more difficult to resolve these sorts of problems when there's this inherent noise and you don't know exactly where the limbs are and you don't know exactly how much force you've exerted on this muscle and so forth because of the noise inherent there. So it's these

problems of non-linearity, redundancy, feedback delays and inherent noise that make the problem of motor control very difficult. Now, to break down the problems that need to be resolved, by the motor system, or the motor control system, we can decompose actions into at least six steps. So the first thing that we need to do is determine the position of the target object in the environment. I mean, it could be a person or it could be an animal or a physical object, whatever it is, that whatever we want to interact with or even a location. We have to determine the position of that in the environment. So we use perception to do that, like vision or whatever.

The second step is that we need to determine the current position, for example, of our hand. let's say we're going to grasp an object. We need to determine the position of our hand relative to that object. So we have to situate our hand relative to the object in a kind of internal model of the environment. For best accuracy, we should do that by combining sensory inputs, like a visual image of where my hand is, with a proprioceptive inputs. So proporeoception is the kind of sensory process that allows us to detect where our body is in space and the relative positioning of our limbs and so forth. So ideally we use vision and proprioception for that to give a more accurate idea of exactly where our limbs are, because neither of them

is sort of perfect, and ideally we use both to increase the accuracy. So, once I've determined the position of the target and the position of my hand, the third step is to compute the spatial relationship between the two of them by transforming the sensory information into a common frame of reference. So I could have representations of the target object and a representation of my hand, but they need to be in the same frame of reference or in the same like coordinate system effectively, in order to represent the distance and spatial relationship between them. So that's a further set of transformations that need to happen. The fourth step is to form a motor plan that specifies the direction and distance

from the current position that the hand is in to the target location. So I need to know where the hand is, but then I need to know where it's going to go and the specific pattern of motor commands that I need to get there. So step four is once I've worked out that direction and distance that I need to move, I need to actually compute the motor commands needed to get there. So that's step five. So one is like what's the path and five is like how are you going to actually move along that path? What series of contractions is going to be performed? Step six is to adapt motion as necessary due to changing circumstances and changing environments, which of course we need to do.

If I'm walking along and something moves out in front of me or I realize that there's a step that I didn't see or something like that, I have to adapt rapidly and dynamically to the changing environment. And that has to be something that involves repeated feedback and processing of changes in my appropriate reception, vision, auditory system, and so forth, that is constantly taken in. Hopefully you get the sense here that motor control is actually an extremely complicated and difficult problem. It's not a simple matter of just performing like by rote a series of instructions or something like that. And the difficulty of this problem is why it requires so many different brain regions and so many sort of specialized forms of computation and circuits for

processing and integrating different types of information. So now that we're we've discussed some of the computational factors or problems that need to be resolved through motor control, let's now begin talking through the system as a whole. And before we sort of go through all of the different components, well, we won't go through all of them, but we'll go through a number of them, let me set the stage by talking about the motor control system in general terms, and then we'll kind of step through and go through a bit more detail. So we'll start at kind of the lower end and work upwards. So in the previous episode, we talked about skeletal muscles and the motor neurons that innovate them. And we talked about how motor neurons have their cell bodies as part of the spinal

cord and that they innovate the skeletal muscles coming from there. And we talked about how the action potentials produced by the motor neurons determined the activation and strength and timing of the contraction of the muscles. What we didn't talk about is what determines the rate at which action potentials are fired by those motor neurons. And that's where we'll pick up today and talk about the spinal cord. So the spinal cord actually carries out quite a lot of complicated computations relating to motor control at the sort of low level close to the actual execution. So begin by talking about that and the various circuits that operate at the level of the spinal column. We'll then move up to the cortex, which is the outermost region of the brain, the kind of wrinkly

part of the outer layer of the brain, and talk about some of the higher level control. If you think of it as the spinal cord specifies which specific muscles will contract, and the patterns of activation that leads to that, the cortex, it's like a higher level of command. It's sort of like, if you think of organization that has different levels of management, the higher you go up, the management hierarchy, the broader the scope is that they cover, but also the more kind of general and abstract are the instructions that they give, whereas as you go further down, the instructions get more sort of precise and specific and focused on one particular area, or in this case, a particular set of muscles.

So as we move from the spinal cord up to the cortex, we see that the cortex is involved in or is largely responsible for motor planning. So deciding what actions are going to be taken and the pathways that we need to get there and also incorporating information from the senses in order to ensure that we know like a position of the target object and the position of our body relative to that and so forth. So it's that kind of higher level of planning that's more coordinated by the. the premotor cortex, the posterior parietal cortex, and then the primary motor cortex. Then there's the cerebellum, which is, again, that little sort of striated or lined part that looks like a mini brain that kind of sits at the back and underneath the cortex. So the cerebellum is a highly specialized structure. We'll talk more about that later.

It is largely involved in sort of fine-tuning motions and ensuring that they occur in a sort of precise and fluid manner, as well as also integrating information from like sensory feedback. It doesn't produce motion directly, but it sort of like improves its efficiency and efficacy. Finally, there's the basal ganglia. The basal ganglia does many things, and we'll only touch on it briefly here. The basal ganglia is a series of subcortical structure, so they're like, sort of in the middle of the brain, so to speak, although they're spread out. It's not a single structure. They're sort of spread out at different parts in kind of the middle of the brain. And it performs many functions. The function that we're going to focus on here is its role in

action selection. So the motor cortex is kind of in involved in planning out motion controls and giving sort of general instructions to the brainstem and then further down to the spinal column, the basal ganglia can be thought of as choosing which of multiple possible actions will actually be executed. So it's in some sense that's the highest level. That's the general hierarchy we're looking at. So think of it as starting kind of at the primary motor cortex and then moving down through the premotor cortex and the posterior parietal cortex, which sort of help with processing and integration. Then we move down through the brainstem

and then down to the spinal column with the cerebellum providing support and feedback to improve the quality of the motion and provide feedback from, integrate information from sensory feedback as well. And Basil Gagnol, are kind of sitting over the whole thing helping with action selection. So having set the general stage,

let's go through the different aspects in more detail, starting with the spinal cord. Now there are many tracts of myelinated axons that make up the spinal cord. The one that's of most relevance to us is the corticospinal tract. So this is a white matter motor pathway that starts at the cerebral cortex and then terminates on the lower motor neurons. So those are just motor neurons that have their bodies in the spinal cord itself, as well as interneurons. So an interneuron is a neuron that is neither like a sensory neuron, nor is it a motor neuron. So basically, it doesn't innovate a muscle directly, nor does it

receive direct input from like our skin or from a muscle tendon or something. them like that. So it's neither sensory nor motor, so it connects them together. So it's an interneuron. Interneurons, you can think of them as being associated with like processing information. Instead of directly receiving sensory input or directly producing motor activity, it sort of connects those together and helps with processing. So the cortical spinal tract projects from the cortex down to motor neurons and also interneurons. And these motor neurons are the ones that actually directly control the skeletal muscles of the limbs and the trunk. So there are different types of lower motor neurons. The alpha motor neurons are the ones that we talked about mostly in the previous episode.

So they innovate skeletal muscle fibers. So they're the most common type of muscle fiber. And they're the ones that perform the actual contraction. But the other type of motor neurons that I'm going to mention here are gamma motor neurons. So these innovate what are called intrafusal muscle fibers. So these are a specialized type of muscle fiber, which make up a structure called a muscle spindle. So their job, these intrafusal muscle fibers and the spindles that go along with them, is to measure how either relaxed or taunt a muscle.

is part of the pro preceptive system. Remember I said the proprioception is the way that our body determines the relative position and status of our limbs. So these gamma motor neurons innovate, so they provide input to these intrafusal muscle fibers, which are part of the pro preceptive system. And so the muscle spindle's role is to provide signals which are sensitive to the stretch of the muscle. So when the muscle is stretched, it will send information about that back to the central nervous system. In order for that system to work properly, it's necessary for the intrafusal muscle fibers to regularly contract, because basically the intrafusal muscle fibers ensure that the spindle is taunt, and it being taunt is what allows it to detect basically changes in the stretch,

because if it's like a taunt rope or rubber band, then it can detect when there's changes. But if it's, relaxed and loose, then there's a change, you actually can't detect that, right? You can only detect changes in stretch when it's taunt. And so the spindle needs to be kept taunt, and that's kept taunt by firing of the intrafusal muscle fibres. And these gamma motor neurons regularly fire to keep the intrafusal muscle fibers contracted, and thereby keeping the muscle spindle taunt, so that it can provide regular and useful feedback about any stretching that occurs in the muscles. Now, these two different types of motor neurons and the muscles that they innovate is very important because, because as I mentioned at the outset, motor control is not just about sending out signals of what to do.

It's also about receiving feedback and knowing how to adjust in the light of that feedback. And so there's pro-perceptive signals that are received from the intra-fusal muscle fibers of the muscle spindle, as well as other aspects of the pro-receptive system, are very important for the motor system, as we'll talk about a bit more later. So, anyway, we were talking about the cortic spinal tracks, so that's all these white matter motor pathways that project onto these motor neurons and some interneurons as well at different parts of the spinal cord. As I mentioned, some of the neurons from the cortex that project down the cortic spinal tract

synaps directly with lower motor neurons, which allow them to have more direct control over muscle contraction, but there are also others that synaps with interneurons, which have a more sort of regulatory and integrative role with respect to motor control. Proper receptive input from muscle spindles and other parts of the properceptive system also provide feedback input to the spinal interneurons. And so it's not just a top-down process, it's also there's a feedback element to it where the interneurons get input from, like the motor cortex,

as well as from proprioception and other sources, and that's integrated in the interneurons before sort of generating a signal at the actual output of the lower motor neurons, which innovate the skeletal muscles directly. So the important point here is that the spinal cord is not just like a highway. way. It doesn't just transmit information from the brain down to the muscles. It's actually a

computational device. There's a lot of integration of information, both information from like different parts of the cortex and the brainstem as well. There's neurons that come from there, which we'll talk about in a moment. But it integrates information from different parts of the cortex and the brain in order to determine how often to fire action potentials to innovate a particular muscle. in addition, it also computes and integrates feedback, sensory feedback from proprioception, which also has an effect on the timing and magnitude of muscle contractions. There's more that the spinal cord does as well, which we'll get to in a moment as we now move to talk about locomotion. So we're still kind of thinking at the level of the spinal cord, but locomotion is a particularly important function that the spinal cord serves,

or more particularly like the motor neurons in the spinal cord. In order to understand how the signals for locomotion are generated and how that's controlled, we need to introduce the concept of central pattern generators. So a central pattern generator is a self-organizing biological neural circuit. So it's a bunch of neurons connected to each other in a particular way, which is able to self-organize its output. So it is to generate particular types of rhythmic outputs facilitating locomotion. So if you think about locomotion, that's a rhythmic activity. You know, you swing one leg and then you place it down and then you swing the other leg.

you place it down and you go on and on, right? It's a, it's a sort of a rhythmic cyclical process. So these central patent generators are able to produce rhythmic outputs that play a role in facilitating locomotion in the absence of any rhythmic input. So they do require input, but the input doesn't have to be rhythmic. So they can generate this kind of cyclical or rhythmic activity that's needed for locomotion just all by themselves. And they do that because of the way that their architecture is set up. So basically, they consist of a network of tightly coupled excitatory neurons which are connected to each other with inhibitory interneurons so basically so think of it as if there's two excitatory interneurons one connects to let's say

a flexor motor neuron and the other an extensor again the difference between flexor and extensor imagine if you straighten your arm and if you bend your elbow to move your lower forearm towards yourself that's flexion if you now move it back that's extension so flexor and extension motor neurons will just be motor neurons that innovate either flexor or extensile muscles respectively. That's not super important for this example here, but just makes it a bit more concrete. So imagine you have one excitatory engine neuron which synapses with a flexor motor neuron, and then you have another one which synapses with an extensal motor neuron. Now imagine that those two excitatory injureons inhibit each other. So what this means is when the flexor interneuron activates, it produces contraction of the flexor muscle

that's connected to the flexor motor neuron. And then it is, it is a flexor inner neuron. And then it inhibits the extensor motor neuron. But then sometime later, the reverse happens, right? The excitatory interneuron corresponding to the extensor fires, and that it inhibits the excitatory neuron that is attached to the flexor. And so they go one and then the other. When one is active, it inhibits the other, but then after a time that reverses, and then the other one inhibits the other. So this is how the rhythmic output is generated, even without any rhythmic input. You just need a sort of a baseline of input that gives some baseline activity. to the system and then you sort of naturally just get this rhythmic output. So this is really cool.

And it turns out that it's been experimentally verified using various animals that the basic pattern of like stepping, not all of the detail, but the basic pattern of stepping can be produced by a spinal cord without needing any descending commands from either the cortex or feedback sensory input. Now this doesn't mean that we can walk without needing any input from the cortex or sensory feedback, like you can't, but the basic pattern of stepping can be produced by just the spinal chord itself without needing any external input. And so in terms of how this is done, I've been talking about the rhythm generator, which is part of the central pattern generator, that generates the rhythmic component. But in addition, you also need the more specific pattern of muscle contractions

that leads to the actual walking behavior. And this series of muscle contractions is produced by networks of interneurons called patterning networks. So basically the idea is you just, you just generate your rhythmic motion using the rhythm generator, and then downstream from that, like as part of the later parts of the neural network, those simple rhythmic motions are converted into more complex patterns of activity, which then occur in a rhythmic manner. That's all facilitated by a complicated network of interneurons, which then can connect up ultimately with the motor neurons that innovate the relevant muscles. Now, although these central pattern generators can activate and produce like simple stepping

motions without any external input, either from the cortex or from sensory inputs. In order to walk properly, we do need that input. And that's provided, for example, by the proprioceptive system. And muscle spindles, which I talked about before, which provide signals for how much stretch has occurred at a given muscle at a given location. Muscle spindles provide critical feedback about basically where the system is in the process of generating locomotion. So if you think about walking, there is a step action and a swing action. And those need to be alternated between the two different limbs. So step swing, step swing, alternating one limb with the other, right? In order to tell that the limb is sort of ready and at the right stage for moving to the next stage, whether it's

stepping or swinging, the muscle spindles can provide that sensory feedback saying, oh yes, we have the right level of stretch for the next stage to go ahead. And alternatively, if that hasn't happened, if we've tripped or, you know, taken a long step for whatever reason, or moving faster, or something else has happened, then there'll be a signal that says, oh, hang on, we need to adjust for this, and then that is integrated via the intern neurons and adjusts the gate as necessary. So again, this is what makes it complicated, because you can't just rely on that central pattern generator to just do the same thing every time. That provides a kind of a baseline, but that needs to be then adjusted dynamically with

pro-preoceptive feedback from the various joints. But there's also feedback that comes from the sensory system, right? So based on what I see and the path in front of me and, you know, things that I hear in my environment, I may also adjust my gate. And so there's continual integration of all of these sorts of signals at the level of the spinal cord. We now move on to talk about the primary motor cortex. So we're moving up in the hierarchy from talking about the spinal cord and a little bit about the brain stem. We're now moving up to talk about the cortex.

The primary motor cortex is the primary site of the initiation of voluntary motor actions. So this is where motor actions are sort of most directly initiated from. Higher areas that we will talk about like premotor and posterior parietal cortex as well as the cerebellum are involved in motor planning and regulation and feedback, but not so much the direct initiation of motion. The primary motor cortex is located kind of right along the top of the middle of the brain across both hemispheres, and there is a contralateral representation, which means that the left side of the body is controlled by the right side of the brain and vice versa. The primary motor cortex contains pyramidal neurons in layer five.

There's a number of layers to the cortex, and in layer five, there are large pyramidal neurons that send out long axons down the corticospinal tract that synapse either directly with lower motor neurons that then directly synapse with the muscles themselves, or some of these pyramidal neurons from the primary motor cortex also go down the corticospinal tract and synapse with interneurons in the spinal column, which remember we talked about how those are more related to integrating information with sensory pro preception as well as top down signals. So there's both types of neurons in the primary motor cortex that synaps directly with motor neurons as well as those that synaps with interneurons in the spinal cord. One very interesting

feature of the primary motor cortex is that it contains a somatotopic ordering of body parts. And what this means is that there is a spatial relationship between regions of the cortex and parts of the body that they control or innovate. So the way it works, is that towards the most lateral part of the cortex, so kind of the most to the side, we have a representation of the tongue, and then the face from the jaw, lips, nose, and up to the eyes. And then above that, there's a representation of the thumb and the fingers, and then the hand, wrist, elbow, and up to the trunk, and then down the leg, and to the ankle and toes, going right up to the very middle of the brain. And then there's a mirror representation of the other side of the body over the opposite side of the brain.

So this was quite surprising when it was first discovered, although subsequently we've discovered many of these sort of spatiotopic maps in the brain. I think I would have mentioned in the context of the visual system, there's a spatiotopic mapping between receptive fields in visual space on the retina and regions in the primary visual cortex in which those are represented. There's tonal topic mapping of different frequencies in the audio cortex, and so this is this type of mapping between regions of like signal space or representation space and physical layout of the cortex is fairly common. And one way that these are sometimes described is in

terms of a homunculus, which means like a little person. This is sometimes called the motor homunculus. One thing about the motor homunculus though is that although the spatial relationship between regions of the body and regions of the cortex that control those corresponding muscles is preserved, the relative size of the body is not preserved. So the representation of the face and the hands is extremely large relative to their actual size in the body, which makes sense because we have much smaller and intricate muscles in the hands and the face that we need to use for eating and making facial expressions and talking and so forth, and then for the hands, for manipulating objects. Obviously, that's very important. We don't need to have such fine grain

control over the arms and legs. And so the chorus, responding regions that control them in the cortex are much smaller. So that's one difference. Another thing about the motor homunculus is that it's not quite as precise or as perfect as it's sometimes portrayed in diagrams or in some intro textbooks. There's actually a lot of overlap between different regions of the body and kind of fuzziness at the boundaries. And there are some regions like the fingers that are represented more than once. So it's a little bit less precise than is sometimes portrayed. But nevertheless, there is a general. relationship between regions of the primary motor cortex and the regions of the body that they

innovate. Electrical stimulation of cells in the primary motor cortex elicits motor activity. So if you basically zap people's brains in a particular part of the motor cortex, you will elicit motor activity in the corresponding part of the body. But usually it's not like a single muscle will contract at the same time each time you stimulate a particular cell. Normally, what you get is a more complex motor behavior in a general body region. like you might have an arm movement or a hand motion or something like that.

It varies a little bit for each person but the overall architecture is broadly the same. And this is not surprising right because the primary motor cortex is relatively high up in the hierarchy. It's not at the very top but it's relatively high up. The primary motor cortex directly innovates cells in the spinal cord and it's those motor neurons in the spinal cord that themselves directly innovate the muscles. So the primary motor cortex doesn't directly control muscles, it indirectly control muscles, it indirectly control some through motor neurons in the spinal cord as well as those interneurons. There's also some cells in the primary motor cortex that project other parts of the brain,

including various motor regions in the brain stem and the medulla and other lower parts of the brain. I haven't talked about those in too much detail, but we'll just say that there are other regions that control various aspects of locomotion, posture, control of the eyes and so forth that are located in the brain stem, which are these sort of lower parts of the brain below the cortex. And there are some cells in the primary motor cortex that project there as well, which then in turn project lower on into the spinal cord. So it's not like the primary motor cortex directly controls everything, but it is sort of the primary site of motor initiation where the actual sort of intention, I guess conscious intention or sort of subconscious intention to

enact a motor task is most directly executed. And then those signals are then passed down through the chain to directly activate the muscles in the appropriate way. Neurons in the primary motor cortex encode a variety of different aspects of movement, so it's not just that they encode for kind of different parts of the body, as we talked about with a motor homunculus, but there are also neurons that code for the direction of motion, the force of motion, and the timing of the contraction of particular muscles. So there have been various experimental studies that have found

that the rate of firing of particular neurons correlates pretty strong, with the force applied by particular muscles. So basically that will eventually translate into the rate at which the motor neurons are firing, which then results in a faster rate of contraction and faster rate of contraction by the muscles, and activation of a larger number of motor units. Because remember, each muscle contains many different motor units within it, which are bunches of cells that are kind of activated together, and the amount of force that is generated by the muscle is determined,

both by the rate at which individual motor units are activated, but also the number of motor units that are recruited. And so both of those can be affected by, and therefore are encoded by cells in the primary motor cortex. I also mentioned that the direction of motion is coded by population codes in the primary motor cortex. There are individual neurons that are selective for motion in particular directions.

So this is, like, direction is going to depend on the muscle in question, but if you're talking about like reaching in different directions, for example, or extending a leg out in a particular direction or whatever it be, there'll be particular neurons in the corresponding region of the primary motor cortex that preferentially activate when motion in that direction is elicited or is required. And the way that the overall system works is that it's not any single cell that determines what the direction of motion is. It's the overall population coding.

So you kind of like take each of the neurons in that region, and then you kind of average their directions, or their preferred direction of firing, so some will prefer to activate motion in like the forward direction, some backwards, some left or right and so forth. You kind of average all of those out, but weighted according to how active each neuron is. So for example, if the overall population is to code for forward motion, what that will mean is that the forward motion neurons are most active, and then like the kind of diagonally left and diagonally right neurons are like fairly active, but a bit less active, the side to side neurons

are not too active and then the backwards neurons are like the least active. So there's kind of noise to it and it's not perfect, but this population coding also provides redundancy, right, and ensures that if there's damage to a small number of neurons that code for forward motion, then that is not going to eliminate the possibility of forward motion, that there is an ability to be robust to damage or to other neural noise. So this method of population coding is very common throughout the nervous system and is found in the motor cortex in particular. So to work out the direction of motion, you have to kind of do a weighted average of the preferred directions of all of the active neurons.

The primary motor cortex receives input from the pre-motor cortex and posterior parietal cortex, which we'll talk about in a moment. It also receives a feedback and input from the cerebellum. So remember, that's the little brain kind of behind the cortex, which is responsible for motor tasks. So the primary motor cortex receives input from the cerebellum. It also receives input from higher brain regions like the frontal cortex and from the basal ganglary. Remember those are the subcortical structures that are involved in sort of decision making. So again, the primary motor cortex doesn't

like decide what motion to do. It's more like it executes the broad strokes. It's in charge of executing the broad strokes of the plan that was decided on. So you can think of as kind of like middle management. The primary motor cortex or different parts of the primary motor cortex are kind of in charge of different body regions. They don't directly execute the motion themselves. That's the motor neurons in the spinal cord that do that. But they are in charge of kind of like recruiting the right, recruiting the help in a sense to get their job done, recruiting the right combination of spinal neurons and interneurons in order to elicit the desired motion. But the primary motor cortex doesn't decide on exactly what motion will be performed. That is

when I say decided. I mean, like it doesn't compute that directly. Those computations are performed by higher areas that we'll talk about. I mean, there are some computations in the primary motor cortex, but that's not its main role. Its main role is like direct recruitment of the specific populations of spinal neurons that are required. One final point to mention is that cortical motor maps, not just in the primary motor cortex, but in other motor cortices as well, they are adaptable following injury or also following motor learning. So, you know, this humunculus that I talked about, this relationship between regions of the cortex and areas of the body that they innovate or that they control. This is not fixed, so it changes over time. One particularly interesting application of this is following the loss of a limb.

So what tends to happen is that if a particular region of the cortex doesn't receive input for a long time and or it also doesn't require activation, because say that limb doesn't exist and so you can't do anything with it, that area of the cortex will gradually be taken over and used for other. things. So if you lost a finger, for example, then you would tend to see that region of the cortex would be taken over and start to be used in representing surrounding fingers. This sort of research has been done on monkeys, for example. And so there is not one fixed representation, but it's dynamic, and it responds to experience, and it responds to motor learning, and it responds to injury. This also means that following even very significant injuries, often, especially for younger people,

it's a bit easier for younger people, but even people who are quite old as well, often there can be substantial regain a function following therapy and basically just learning to use like the new limb or to kind of work around the particular deficit. So there is a very large capacity for adaption, plasticity and relearning. It's not like that these maps are all fixed in place. Okay, so that concludes what I have to say about the primary motor cortex. Now let's move up the hierarchy a bit further and talk about the pre-motor cortex. So the pre-motor cortex is an area in the cortex lying just in front of the motor cortex. So it's part of the frontal lobe technically, but just like at the back of the frontal lobe just in front of the primary motor

cortex. And it's responsible for tasks, including the preparation of movement, spatial guidance of reaching and other types of movements, and integrating sensory input with motor goals and motor tasks. So while the primary motor cortex has more direct control over execution of motor tasks, the premotor cortex is more involved with planning and preparation relative to an organism's goals. And therefore, the output of the neurons in the premotor cortex, as well as the kind of representations or the information encoded by those neurons, is highly dependent on the context in which the action is performed, the relationship between like the stimulus and the desired response and rules that may have been learned. Whereas in the primary motor cortex, the outputs and the commands

are most closely related to the specific details of the movement, and they're less influenced by the behavioral context. This always has to be understood relatively, so it's not like that there's no effective context on the primary motor cortex, or that there's like no neurons that code more directly for motor execution in the pre-motor cortex. These claims should all be understood relatively, and not absolutely, because the brain is not that finely engineered and it's, you know, quite messy. But we can understand relatively what the specialization of the different regions is.

So I think a loose analogy is to think of the pre-motor cortex as a higher level of management. It's maybe not at the very top level of management. You might think of that as the prefrontal cortex plus kind of like the basal ganglia. Those two are involved in basically deciding which motor task will be executed or which general goal the organism will pursue. So they're the kind of like chief executives, if you like. At the pre-motor cortex, we're kind of like the next level down. We're like high-level managers, but not the chief executive.

So the pre-motor cortex is responsible for integrating sensory, information with high-level goals that have already been selected and determining kind of the sort of preparing movement to determine the best way in order to carry out those goals. So that's all a bit abstract. Let me give some examples of specific experimental findings or specific types of information that have been found to be encoded in the premotor cortex to give you an idea of the sort of computations that are performed here. So some premotor neurons have been found to encode for particular

voluntary behaviors that are based on decision rules that have been learned by the organism. Many of these sorts of studies that I'm going to be talking about have been done in monkeys, by the way, because they have a good visual system, they can manipulate objects with their hands like humans, and they're quite intelligent, so much of this was primate research. As I said, so some premonore neurons code for particular voluntary behavior. So an example would be if you show an image to a monkey and you train it to learn which images kind of match with the sample image in which don't, and then you teach it to, like, if the image matches, perform this motor task, like reach over to the left, and then if

it doesn't match, you reach over to the right. So basically, there's a behavioral task that needs to be performed that's contingent on whether two images match or not. Now, the important point here is that the behavioral task is not dependent just on what the image is. That's a more sort of lower level of processing. The high level of processing is that, irrespective of what the images are, It's just whether they match or not that determines whether you reach left or reach right. Even though I could present lots of different pairs of images, it's just if they match, then do this. If not, then do something else, right? And what's been found is that there are neurons in the premotor cortex that code for whether the images match and whether the corresponding motor action will be performed.

So there are motor neurons that code for whether there's rule-dependent activity that's going to be executed. So this is an indication that the neurons there are integrating information from the kind of executive function about the knowledge of the rule, plus from the sensory cortex based on looking at the images, plus motor activity based on like, okay, this is the motor task that we need to perform now. So that shows the kind of level of abstraction that these, at least some of these neurons are coding for. Some other neurons that are found in the premotor cortex have received a lot of attention in the popular minors. media and science outreach. And these are called mirror neurons. So mirror neurons are neurons that are active, so they fire action potentials, both when an animal performs a particular action, as well as when the animal observes that same action being performed by someone else. So it could be reaching for an object or consuming food or something like that, right? So the neuron

fires when the animal does that thing and when the animal observes that thing being done by someone else. These neurons are said to be mirroring what the behavior is in someone else, as though you're self-acting when you observe someone else performing that action. It is not known what the purpose of mirror neurons is. There are many different hypotheses about what they might be useful for, ranging from learning by observation, all the way through to forming the basis for empathy or self-awareness, consciousness itself. This is still hotly debated in the literature. And in my personal view, there's been a lot of unsubstantiated claims that have been made, especially in the popular media, about the significance of mirror neurons. It's not even

clear whether mirror neurons form a specified like subpopulation within the cortex, or whether it's just sort of a byproduct of the fact that these neurons are performing certain types of computations or encoding certain types of information that they happen to respond in this particular way. Because remember, the neurons don't come with labels about like, this is what I respond to. the only way we can figure out what they're responsive to, what their receptive fields or conditions are of activation, is by providing stimuli and seeing when they are active. The problem with that approach is that we only know what they are active, what they respond to, or what they don't respond to, out of the set of stimuli that we've tried.

And of course, there's basically a infinite or near infinite range of stimuli that we could try, and we only ever try a small subset of those. So the point is it's not clear if mirror neurons are really mirroring a behavior, or if that's like a byproduct that we've found, but really they're sort of doing something else, and that's just like one manifestation of that behavior. So it is very interesting that these are found in the premotor cortex, and clearly there is some sort of more complex processing going on

that may relate to like observational learning or theory of mind possibly, but those are all quite speculative. And so be wary of some of the claims made in the popular press about mirror neurons. By the way, there are mirror neurons found in other parts of the brain as well. It's not just the premotor cortex, but that is one region where they have been discovered. So let's talk about some of the other types of, information or actions that are encoded by neurons in the premotor cortex. So there are some neurons

in the ventral pre-motor cortex that have been found to be active selectively to one particular type of grip. It's not which muscles are involved, because it's the same muscles, not in the precise exact combination, obviously. And it's not like the motor task itself, which is grip this object, but it's rather the type of grip that's performed. And so again, we see that there's a higher level of abstraction here that's not at the level of like particular muscles, but it's also not quite at the level of abstraction is like just grip this thing. It's like perform this type of grip. Another type of encoding that's been found is that there are neurons which are selective

only when movements are performed in a specific order. So they'll teach the monkey a series of motor tasks like push, turn, pull, and then they'll train it on different tasks that require different order of those basic motions. So push, turn, pull, pull, push, turn, push, pull. turn and so forth. And what they found is that there are some neurons which activate only when those movements are performed in a specific order. And so this is kind of a bit like the neurons that are selective to a particular grip, except this time it's a sequence of motion. So it's encoding for a high level of abstraction than like which muscles necessarily, but it's a lower level

abstraction. It's a bit more contextual than what the overall task is. It's sort of a higher level planning type of information that's needed, which will then, it passes down to the primary motor cortex which will then execute that motion in particular with particular muscles. Finally, there are populations of neurons in the premotor cortex that code for information about possible actions. And this is one that I find particularly interesting. So basically the idea here is that there are two possible actions which are visually represented to the monkey by like two different colors or dots on the screen, like red and a, like say a red

circle and a blue circle. And you have both of these presented on the screen and the monkey knows from training that that means that there's like two things it can do, like reach left or reach right or whatever. Now what we find is that there are neurons in the premodic cortex, which when you present this sort of either-or stimulus, it's like blue-red, it could be either, there are cells that are active that represent both of those possibilities. So some cells are kind of representing the red possibility and some are representing the blue possibility. And then what happens is now let's say the blue circle disappears and it's only the red circle that's visible. And that's a way

to tell the monkey, oh, it's actually red. Like previously it could have been red. or blue and therefore left or right, but now it's going to be right. So that's a way to tell the monkey that this is the motion that you're going to need to perform. And critically, the monkey knows that it doesn't need to perform the motion yet, because it will be given a go signal later on, but it knows what motion it will be performing. So when that happens, what you see is that the neurons that were kind of preparing for either type of motion, so like there were some that preparing for the red and some that were preparing for the blue, some left, some right. Now only say the red, the red corresponding neurons, let's say the left, let's say red corresponds to the left. So only the left reaching

neurons are active now, but they're more active. So they increase in activity. And then the blue corresponding neurons, so the ones that encode for like reaching right, those stop responding. And this is interesting because this is before any motion has even occurred. This is just like preparing for motion. Previously the monkey wasn't sure whether it needed to reach right or left. And now that it's been shown that it will need to reach left, but doesn't need to yet, but it will need to, then the neurons coding for left motion increase in action. activity, whereas those that were previously a little bit active, coding for a possible right motion, those now stop being active. And then the final stages, after the go signal is given, then

the left-reach neurons become even more active. So there's like three stages of activity. There's the possible motion where they're a little bit active, then there's the prepare-for motion, which is they're more active, and then there's the act now, which is the most highly active. And then we also see that the reach-right neurons are a little bit active when it's maybe reach-right, but then once we receive the motor cue that says, oh, we actually won't need to reach right, we'll be reaching left instead, then those neurons stop being active. So this shows that there are neurons in the premodered cortex that not just code for actual motion that's anticipated, but code for potential motions as well.

And then integrate that information based on sensory feedback. And so that the output of the neurons increases when the monkey knows based on sensory information, that it will need to perform this type of motor action, but not this other one. Okay, let's move now to talk briefly about the posterior parietal cortex. So this is technically not part of the motor cortex. It's more of a sensory integration cortex, but it's highly relevant for motor control. So if you recall, the primary motor cortex lies kind of along the middle of the brain, from the side right through the middle and then down the other side, along kind of the center.

And specifically, the primary motor cortex lies just in front of a kind of valley, if you like in the center, across kind of the middle of the brain, which is called the central sulcus. So the primary motor cortex is just in front of that. In front of the primary motor cortex is the premotor cortex. Now, just to the back of the central sulcus is the primary somatosensory cortex. So that's the region of the body that receives sensory input from touch and related senses of different parts of the body. We haven't really talked so much about that here because we're focusing on motor control. But then just to the back of the primary somatosensory cortex, is the posterior parietal cortex. And so that's an association area, which is highly involved

in coding for motor functions and motor behaviors as well. So let's talk a little bit about what that does. So its particular focus, as I sort of mentioned, is integrating sensory information and linking it to motor actions. It does not contain a topographic mapping like the primary motor cortex does. And by the way, as far as I know, the pre-motor cortex doesn't contain one of those homunculus mappings either. But that's because the premotor cortex and the posterior parietal cortices are involved in representing more sort of complex, higher level and more abstract aspects of motion. The posterior parietal cortex particularly focused on integrating sensory information with motor tasks. Not to say that the premotor cortex doesn't do that either, but that's more of a focus

of the parietal cortex. So one aspect of what the posterior parietal cortex would focus on is encoding information relating to planning for navigating obstacles in locomotion, because obviously that requires a lot of quite specific sensory feedback from like vision or maybe the auditory system as well. And so that's sort of processed and integrated with the motor tasks in the posterior parietal cortex. So overall, we can say that the posterior parietal cortex is involved in motor planning and motor intentions, representing the state of the body and integrating these with sensory inputs. So the premotor cortex also is involved in sort of planning and representing. any motor intentions. But what the posterior pridal cortex more focuses on is representing the

current state of the body and integrating that with sensory inputs. So it's a bit more sensory focused, a bit more sort of pro-perceptive focused, a bit more about body representation focused, as opposed to the premotor cortex was a bit more kind of planning and representing motor task focused. So as an example of some of the specific information that's known to be encoded by some neurons in the posterior parietal cortex. Some visual neurons in the parietal cortex have been found to have receptive fields that expand when a tool is grasped. So I should emphasize these are visual neurons, so they're activating in response to visual input, not in response to motor control directly, but they're receptive fields, so that's effectively the range of the visual field that they're

responsive to expands when the tool is grasped. So the basic idea would be this particular neuron in the posterior parietal cortex responds to input from like the monkey's hand. So when the monkey kind of observes this part of its visual field, then this particular set of neurons is active in the posterior parietal cortex. However, when the monkey grasps a tool, now these visual neurons are active, not just kind of in the region where the monkey can see its hand, but also where it can see the tool. And interestingly, and this is the coolest part, if the monkey is simply passively holding the tool without kind of any intention or need to use it, then,

the receptive field shrinks back again. So it only sort of is receptive to input from the hand. So it's only when the monkey is using the tool or planning to use the tool that the receptive field increases in size, indicating that there's a representation here that's integrating sensory and motor task or motor goal information, which is quite interesting. There are other neurons in the posterior parietal cortex that have been found to have receptive fields, again, so these are visual neurons, but they have receptive fields that are specific to the relationship between where the monkey is focused on in their gaze and what type of motion it's performing. So, for example, these would be neurons that are sensitive to motion to the right of the gaze or motion above the gaze

or grasping in the direction that you're looking at as opposed to like to the left of where you're looking at. So things like that. So different regions of the posterior parietal cortex have receptive fields that are responsive to different combinations of like visual perception and motor tasks. So these are the sort of neurons that have been found in the posterior parietal cortex that really integrate like motor tasks or higher level motor goals with particularly visual, but also select somatosensory and other types of representations. So that's how the posterior parietal cortex is kind of differentiated, at least to some extent, from the premotor cortex. It's more about sensory integration. Okay, so we've talked about the primary motor cortex, the premotor and the

posterior parietal cortex. There are other motors as well, like the supplementary motor cortex that I haven't even talked about, but I've decided to try to keep it a bit simple. But just bear in mind that there are other important regions as well, but I've focused on, I think, the three that are most important. Now, we're nearly at the end, but we're going to conclude by talking about two remaining regions that are very important for motor control, the cerebellum, and then the basal ganglia. So we've kind of, at this point, reached the top of the hierarchy. I mean, the premotor cortex and the posterior parietal cortex, you can think of them as kind of like upper level management. The very highest level, like the chief executives, think of that as the frontal cortex plus the

ganglia. Those are involved in decision-making and determining which motor task will be performed or what the overall goal is. I'll talk a bit more about the basal ganglia in a moment. But there's another player in the story, which is very important, but it's kind of hard to fit in this sort of hierarchical description that I've been giving. I mean, the hierarchy is an approximation at best anyway. But the cerebellum kind of sits outside of that, because the cerebellum, being this sort of bulging, striated structure that sits kind of at the back and below the rest of the cortex, is very unusual in many ways, and it's known to be very tightly and significantly involved in motor control, particularly of fine motor tasks, like complex motion.

Damage to the cerebellum produces disorders in fine motor control and movement, as well as equilibrium posture and motor learning in humans. So it's known to be very important for a wide range of motor tasks. That doesn't mean that the cerebellum is only involved in motor tasks. There's increasing kind of interest in the role that the cerebellum may have to play in other types of, say, sensory perception and representation of semantic information as well. I've seen some studies on that. So that's a little bit controversial. So you shouldn't think of the cerebellum as like only involved in motor function, although that's mostly the respect in which it's been studied, but it is known to perform very important motor functions. But the thing about the cerebellum is that, as I said, it doesn't

really fit into this hierarchy very well. It seems to provide like additional kind of processing support and learning support. The cerebellum doesn't directly instigate motions in the way that the primary motor cortex does. It doesn't directly control the motor neurons like the spinal cord does. And it's not really involved in motor planning in the same way that say the premotor cortex is, although it may do like little bits of those things, but primarily it seems to be about providing like feedback and also learning on the basis of that feedback for fine and complex motor tasks. The cerebellum has a very intricate and regular structure, which is somewhat unusual for the brain. Well, I guess it depends on the region, but that's been quite well studied.

I won't try to explain that in detail here, that the structure of the cerebellum is not our main focus here, but I will just mention that the surface of the cerebellum is covered in finely spaced parallel grooves, which is different to the cortex, which has these sort of irregular convolutions and folds, whereas the cerebellum has as quite regular and as kind of smaller grooves. And the whole surface of the cerebellum is actually a continuous layer of tissue that's very finely folded. So it's kind of like an accordion, it's kind of squashed up, but it's actually one single layer. Again, a bit like the cortex, but it's more sort of finely folded and more regularly folded. Within the kind of outer layer, so the cortex of the cerebellum, there are several types of neurons

that are connected in a highly regular repeating structure across basically the heart. whole surface of the of the cerebellum. Two of the most important of these are Pekingi cells and granule cells. The cerebellum is functionally divided into three main regions, or the main part of is divided into three main regions. I won't go through the names of them or try to describe exactly how it's split because it's a little bit complicated, but very loosely you can think of just kind of like the sides, the middle, and then a region in between. So those are like three regions. Because remember, most aspects of the brain are bilaterally symmetric. So So it's the left and the right side perform the same function, and then the center bit performs

kind of the same function, and then there's a region in between the sides in the middle on either side, which also performs a similar function. And these three different regions receive input from and project input out to different brain regions and are involved in performing somewhat different tasks. So the lateral parts of the further to the side parts of the cerebellum receive input from the frontal, parietal and temporal cortices. So this is sometimes called the association cortex. So across a lot of the cortex, but not so much like the motor cortex or the somatosensory cortex, and not so much the primary visual cortex, which is at the very back of the brain. So it receives input from this very wide area

of the association cortex and projects largely to the motor, premotor, and the prefrontal cortex. And this part of the cerebellum, perhaps unsurprisingly, is going to be involved in integrating executive plans and sensory information with motor tasks and motor goals. The medial or middle part of the cerebellum receives most of its input from somatosensory receptors in like the trunk, so in the body, as well as directly from the auditory and visual receptors, so like the inner ear and the eye, and it projects out to really a wide range of brain regions, including the motor cortex and various interneurons in the spine. Finally, the remaining parts of the cerebellum in between the middle and the edge receive input

from the somatosensory cortex and the motor cortex and projects out to back to the motor cortex and as well as parts of the parietal and prefrontal cortex. And it also has projections out to interneurons in the spine and some other parts of nuclei in the brainstem. So there's quite a lot of different projections and connections here. It's not like it's a one region to one region thing. But just bear in mind that there is some sort of structural specialization of different parts of the cerebellum. I'm not going to go into that too much here. It's not that relevant. But it it's something to bear in mind that it's not like the whole thing is a homogenous structure. But basically, there is input coming from a wide range of different parts of the cortex,

as well as somatosensory, auditory, and visual information. And the output is projected to a wide range of areas of the brain, particularly motor and pre-frontal and premotorocortices, as well as there are direct projections to nuclei in the brainstem, which we talked about are also involved in various aspects of motor control, though we haven't gone into too much detail about that here. And there are also projections directly to interneurons in the brainstems in the brainstems, the spinal cord. So all sorts of levels of the kind of motor hierarchy are projected to by the

cerebellum and it receives input from many parts of it as well. This kind of makes sense if you think about how it has a sort of a supportive and integrative and feedback learning role in motor control. So it kind of sits a little bit outside of and to the side, if you like, of the motor hierarchy. Now let me just talk a little bit about the microcircuitary of the cerebellum because this has been studied in quite a lot of detail. And I mentioned that there is such regularity across most of the cortex of the cerebellum that it's thought that there is a canonical circuit that may perform some kind of canonical computation that is doing the same type of computation to its inputs, regardless of where those inputs come from. Although it's not known whether that's actually true

or what exactly that computation might be. But it does look like that the cerebellum is kind of specialized for like massively parallel standardized computation of some sort. The basic structure is that the input comes sort of from from the bottom through cells called mossy fibers, which connect up to other cells called granule cells. The granular cells sit in kind of their own layer. Those granule cells then project outwards across sort of the surface in these parallel fibers, which are these axons that all kind of run along the surface or near the surface of the cortex. So these are the outputs of the granular cells. These are parallel fibers. The parallel fibres then synaps with the dendrites of the Pekingi cells.

So you remember I mentioned the Pekingi cells before. These Pekingi cells have a very unique morphology. So they essentially look like trees, that they have a very highly branched and widely branched set of dendrites, which allows them to essentially pick up information from a wide range of these parallel fibers. And these Pekingi cells then project their outputs down to these deep nuclei, which are sort of backed down further sort of inwards towards the center of the cerebellum. So basically the path is sort of from mossy fibers up to the granule cells, which sit in a sort of regular layer. They project out parallel to the surface in all of these parallel fibers,

which are a way of sort of spreading their information across a wide area. The parallel fibers connect up with the Pekingi cells, which are these highly branched neurons, which then allows a lot of integration of information from many parallel fibers and many Pekingi cells, and then they project back downwards again to the deep nuclei, which form the output of the cerebellum. But the point is that this sort of set up with this fairly direct feed-forward structure, but with some sort of feedback mechanisms within it, as well as how widespread this is across the cerebellum, indicates that there is a very specific computation that's being performed here.

There's a particular way the information is being processed. We don't know exactly what that is. However, it appears that there's some type of important learning that's being performed here. The basic idea that at least many researchers have put forward is that what the cerebellum is doing, at least with respect to motor tasks, with the input that it gets from the cortex, it computes a representation of where the body is and then what movements will be necessary in order to produce a particular type of motion. So it has an internal like mental representation of the body and the positioning of the limbs.

And then it computes a prediction as to where, how that will change when I move like my limb this way or when I step forward or whatever. So it sort of simulates internally what the relative positioning of the body will be after a certain motion it is performed. And then the key thing is what it does is then it receives sensory feedback from somatosensory receptors, the visual system, and then other parts of the cortex as well. It receives feedback and then it kind of updates. It corrects on the basis of, okay, here's where I expected the body to be, here's where it

actually is, and then that leads to an error signal which allows me to kind of update and then correct, not just update my representation, but also correct the motion signals that produce that in the first place. So I'm like, oh, I thought I'd be here, but I was actually over here. so I need to update the commands that I was giving, and then I send that sort of update signals back to the cortex to update it appropriately. And this is why the cerebellum is involved in fine motor functions and like more complex motor tasks that require a bit more of this sort of fine tuning and feedback. So this is why when we see patients with deficits or like lesions in the cerebellum, that if you ask them to perform a motor task, like for example, just moving a limb up and down repeatedly, a normal person will be able to do that very regularly and in a sort of smooth way,

Whereas someone with damage to the cerebellum, they can still move the limb. So it's not like that they are unable to move, but it's just that the movement is erratic and irregular, and they can't just sort of do a smooth up and down thing. Or if you ask them to track, like, you know, point to this here and then move your finger to your nose and touch your nose. A simple thing like that. But that actually requires a lot of complex computations about, you know,

the smooth path that your finger's going to take as it moves to your nose. Someone with damage to the cerebellum will be able to perform that motion, but they won't be able to perform it in one smooth go. their finger will kind of move this way and that way and it will track you regularly before they finally manage to get to the nose. It seems to be it's because they're kind of lacking this fine, precise feedback mechanism from the cerebellum that kind of updates their mental model on the basis of sensory and somatosensory inputs. So that's a little speculative. We don't know exactly that that's what the cerebellum is doing, but it seems to be doing something like that. And it's this

sort of regular circuitry that allows it to make these sort of calculations based on feedback and the errors that are generated between the expectation of where my body is and where it actually is, then it sends those back to the cortex to update and fine-tune the motion. There is direct experimental evidence that lesions to the cerebellum prevent motor learning, particularly motor learning that involves hand-eye coordination. So one type of study that's been done is that they fit people with these prism glasses that kind of deflect all of the input to one side. So basically, if you are wearing these glasses, something that looks in front of you

is actually like 10 degrees off to the left or something like that. And people can learn to adjust to this so that they can perform the motor action success. So it's like, oh, I just need to make a mental adjustment that, you know, if I want to reach this way, I have to actually sort of do 10 degrees to the right or whatever. Not even necessarily consciously. We're just sort of our brain allows us to do that. The cerebellum seems to be important for doing that. And then what happens if you remove the prisms, then people are overcompensating.

So like they make the adjustment even though it's not needed. And so it takes them a little while to learn to then unadjust back to the normal case. So that's what normally happens. But if you put these prism glasses on someone who has damage to the cerebellum, they do not learn to adjust. So this is strong evidence that the cerebellum is required, at least in normal people, for performing this sort of adjustment and motor learning in order to adjust behavior in response to sensory feedback. Okay, and just before we finish up, let's talk very briefly about the basal ganglia. So I've mentioned the basal ganglia a few times.

We will be seeing this again in future neuroscience-based episodes because it is quite complicated and performs many tasks. I won't really say very much about it here. I just wanted to mention it at the end. The basal ganglia, or basal nuclei, they're also called, is a group of subcortical nuclei. A nuclei is just a bunch of cell bodies sort of clumped together, and subcortical just means it's below the cortex. So they're kind of like in the middle of the brain, under the cortex.

The basal ganglia, unlike the cerebellum, for example, is not a single structure. It's a bunch of things that are kind of connected functionally. Now, experimental studies have shown that the basal ganglia is very, or technically it's plural. I sometimes say that it's singular. The basal ganglia are very important for inhibiting motor control, actually. Not directly causing it to happen, but like inhibiting certain motor performances.

So what the basal ganglia seems to do in regulating what action we perform is to suppress a range of possible behaviors and then lift that suppression of the one that we sort of choose to do or that we make a decision to perform. And this is ultimately influenced by signals from many parts of the brain, including the executive function areas in the prefrontal cortex. So the basal ganglia is involved in kind of integrating information from a wide range sources from the brain and then kind of incorporating that to make a final decision about which inhibition is lifted to allow this behavior to occur. I'm not entirely sure why it works in this sort of inhibitory. I think one of the reasons is that motions are often prepared before they're enacted.

We kind of talked about that in the premarotech cortex, for example, how there are neurons that are active even before the motion actually occurs, but that it's kind of been prepared and ready. Like the computations are done, the activation is ready to go, it just hasn't been activated yet, so to speak. I assume that's done in order to save time and allow us to respond more quickly. But I'm not 100% sure if it's known why it works this way. But it appears that the basal ganglia kind of fits into this because it's like the different possible paths will be computed and prepared, or at least a number of the salient ones will be prepared beforehand.

And then when we sort of make a final decision to go or to perform this action, then the basal ganglia will kind of lift the inhibitory input that was holding that. holding that ready and allow that to be performed. So the hypothesis is that the basal ganglia is acting like a selection mechanism by lifting the inhibition. If there's insufficient activity in the basal ganglia, that can lead to hyperkinetic conditions, hyperkinetic conditions such as Huntington's disease, which are characterized by uncoordinated involuntary body movements, which is an indication of a lack of suppression of motions that we sort of ultimately don't want to perform. There's much else to say about the basal ganglia, but we'll leave it there for the moment, because there's already been more than enough

material in this episode. And let's now take a step back and summarize what we've sort of discussed and give an overall view of the motor system. So control of skeletal muscles and therefore performing of voluntary motor functions such as reaching, chewing, walking and things like that. So we're not talking about breathing here. We're talking about skeletal muscles and conscious motor control. This is all controlled ultimately, like most directly, by motor neurons, which have their cell bodies at various locations in the spinal cord. And those project to a particular motor pool, a set of neurons that are a set of muscle cells that are activated together and are controlled by that particular motor neuron. However, to perform any motion that's actually useful, what we need

is a combination of muscle contractions in the right timing and in the right manner in order for, in order to actualize the motion that we want. And that is controlled by the spinal cord. There are a range of neurons within the spinal cord, including the motor neurons directly, but also interneurons that integrate information from the somatosensory system, particularly the proprioception, which is a system that detects like the tension in muscles and the load and the positioning of different muscles and limbs of the body. And so there are a number of interneurons in the spinal cord that integrate that information and allow for finer control of the particular populations of muscles and the contractions of different groups of muscles to occur in the right

timing in the right order to actually carry out that motion. So the spinal cord and particularly the corticospinal tract where a lot of this goes. It's not just a conduit of signals from the brain downwards. It's actually an active participant of processing of what is the ordering and timing of the contractions that are needed. This is particularly evident in the case of locomotion. We talked about the central pattern generators, which are these circuits of cells that generate this rhythmic outputs. And so you'll have, you know, like the step and the swing motion for walking, for example, where you first activate the flexor neuron and then the extensor neuron and then the flexor in an alternating pattern. And that's mediated by these rhythmic circuits as well as then

patterning circuits, which provide the more specific and precise outputs of motor neuron activity than necessary. And this is dislocated in the spinal cord and requires feedback input from the proper receptive system and the, we talked about the muscle spindle, for example, which detects stretch in muscle. So it requires that to keep everything coordinated and aligned. so that muscles contract at the right time. I also mentioned that there are a number of nuclei at different parts of the lower parts of the brain, including the brain stem and the midbrain,

that are involved in many different specific aspects of motor control such as locomotion, direction of gaze, and posture. We didn't talk too much about those, but bear in mind that they are there. Then moving up to the high level, we talked about the primary motor cortex, which contains that mapping of neurons

that control different, regions of the brain, but where the size of the regions is proportional not to the actual physical size of that body region, but to the amount of precise control that's needed. So dramatic over-representation of the fingers and the hand and also the face and particularly the lips. We talked about how the primary motor cortex is involved in the direct activation or instigation of a motor action, and that there are cells there that encode for detailed aspects of movement like the direction and force and timing of contraction. Moving up further in the hierarchy, we talked about the both the pre-motor cortex and the posterior parietal cortex. So the pre-motor cortex was more involved

in planning of and preparation of motion and integration of motor goals and rules that have been learned with the particular activities that, like muscle activities that will be needed, before sort of passing that information onto the primary motor cortex for that to be processed. And that the posterior parietal cortex is more involved in integrating sensory input with motor planning. At the very top of the hierarchy, we looked at the basal ganglia and I've also mentioned mentioned the prefrontal cortex which are involved in ultimately selecting which actions or which behaviors we will perform. Supporting all of this is the cerebellum, which receives input from both the spinal cord and

across the cortex and then projects back to both of these regions, and essentially seems to be involved in regulating and providing feedback for fine motor control and also supporting motor learning tasks by performing some kind of canonical computation across its highly distributed set of cells across the cortex. But we don't know exactly what the function is, but it is clearly important in the facilitating fine motor control, sensory feedback and motor learning. So to put it all together, you can think of the motor control system as like a hierarchy, although it's only approximate. It's not like that in order to go from the top to the bottom, you have to go through all of the levels. There are kind

of direct connections, kind of everywhere, there are sideways connections, there are back connections. And the sort of overarching principle behind all of these is that motor control is very complex. It's not just passing on a series of simple instructions, but each layer in the hierarchy has to perform a series of very complex computations to integrate information from motor goals through somatosensory and other types of sensory input like the visual system, combining internal representations of the location of different body parts with the desired locations and working out a series of martial contractions that will actually get you from where you are to where you want to be. And all of this requires a great deal of computational power, which is then kind of separated

out in a hierarchical fashion so that different regions specialize in different types of computational tasks. So that brings us to an end of this episode. I hope you found it interesting. If you did, maybe consider supporting the show by spreading the word. You can post a review on iTunes or Spotify or the aggregator of your choice. You can recommend the podcast to a friend. You can also like the podcast Facebook page that we have, Science of Everything podcast on Facebook. You can go there and receive as well updates about how the show is doing. If you'd like to send me an email, you can do so at FODs12 at gm.com. That's FODDS1.com. If you'd like to to financially support the podcast, you can become a Patreon supporter. I'm very grateful to the

donors that I have there. It helps me to stay regular and to devote some extra time and funds to the show as well. And if you'd like to make a one-off donation, you can do so via PayPal, just to my email address. That works as well. Very grateful to those who've made some quite generous, actually, donations in the past. And I'll give a final shout out as well. So I mentioned this last time, but I'll mention it again. I'm in the process of trying to transfer some of my podcasts over to YouTube and in doing so I want to add some visual component to that in the format of still images. As part of that process, I am looking for people to help me. So I've already got a few people who have put their hands up. There is payment available for this, not a huge amount,

but at least it's something. And it doesn't require a whole lot in the way of skills, more just general familiarity with sort of computers and how to Google things, as well as an interest in science and a willingness to learn. So if you're interested in joining the team to help bring the past catalogue of episodes to YouTube, feel free to email me at FosthwaGmail.com and we can discuss the details there. So thank you once again for listening. Take care and I'll talk to you next time.