See discussions, stats, and author profiles for this publication at: [Link]
net/publication/341458604
Motor control and learning
Preprint · May 2020
DOI: 10.13140/RG.2.2.10298.21445
CITATIONS READS
0 1,200
1 author:
Jie Zhang
Ford Motor Company
25 PUBLICATIONS 1,882 CITATIONS
SEE PROFILE
All content following this page was uploaded by Jie Zhang on 18 May 2020.
The user has requested enhancement of the downloaded file.
Motor control and learning
Jie Zhang
May 2020
Abstract
In this article, I will present a theory about human motor control and learning.
1
Human movement is a complex phenomenon. It is a function of the interaction of many
elements, for example, the human brain, the human body, and the environment. It
involves not only a dynamic system but also a complex system. Therefore, the study of
human movement involves many aspects, such as neural, physical, behavioral, and
environmental. This makes the study of human movement extremely complicated and
difficult.
To study such a complex system, the first challenge is how to break it down into several
less complex subsystems, so we can easily find some clues about how this system
works. To this end, I divided the human motor control system into three subsystems:
senses, brain, and body. The sensory system is responsible for collecting information
from the environment and sending it to the brain system. The brain system processes
information selected by the attention system. Based on the results of the processing,
the brain can generate an activation signal so that the body system can take action.
Please note that the article presented here covers only voluntary movements.
Involuntary movements, such as those produced by reflexes and the autonomic nervous
system, are not considered.
Human brain-body interface
To begin with, I will introduce a new concept: the human brain-body interface. Imagine
that the human nervous system is successfully separated between the brain system and
the body system. The first part contains the central nervous system and a part of the
peripheral nervous system, which transmits stimuli from sensory receptors to the brain.
The second part contains the spinal cord and a part of the peripheral nervous system,
which transmits impulses from the central nervous system to the body. The area where
these two parts are separated is called the brain-body interface. Therefore, the interface
contains all output connections from the brain system; or we can also say that the
interface contains all input connections to the body system.
Through this division, the brain-body interface functions as input ports for the body
system. The most important feature of the brain-body interface is that each input port
controls only one motor unit. A motor unit is the smallest functional unit in the motor
system. It consists of a single neuron that innervates a group of skeletal muscles. The
neuron receives signals from the brain and stimulates all muscle fibers in the motor unit,
which means that when the input port is activated, only a simple movement from the
corresponding motor unit is generated.
By introducing the concept of the brain-body interface, the analysis of human motor
control can be greatly simplified. No need to consider how the body system works, we
can focus our attention on the brain system. The idea is that there is no need to know
2
the details of a system in order to control it. This is a very important understanding of
motor control analysis.
Generally speaking, to control the movement of the human body, it is not necessary to
know how the system is constructed. For example, we do not need to understand how
the human muscle system works to control it; just like, we do not need to know how to
make a car in order to drive it. What we need to learn and figure out is the relationship
between the input and output of the system. Therefore, we can skip the details of the
mechanism of complex body systems and simply use a "black box" instead.
A simple analogy of the brain-body interface concept is to imagine the input connection
of the body system as the piano keyboard and the output connection of the brain
system as the hand of a pianist. In order to play, the player does not have to know
exactly how to make the piano. What she needs to figure out is what kind of tone will
sound when a key is pressed by her finger.
The brain-body interface is likely to be mainly located in the primary motor cortex area.
Each input port of the primary motor cortex is connected to a motor unit. Activating the
input port of the primary motor cortex will produce corresponding basic/simple body
movements. Therefore, a complex body movement involves stimulating many such
motor units together.
Declarative memory and Procedural memory
The human brain is an information processing system. It contains two subsystems. The
first is called declarative working memory which is a conscious subsystem. It is
responsible for processing declarative (conscious) memories. The other is called
procedural working memory, which is an unconscious subsystem. It is responsible for
processing procedural (unconscious) memory. Procedural working memory is a slave
system of the declarative working memory. It executes the motor commands sent by the
master system.(Zhang, 2005)
When human information is encoded, stored, and retrieved later, it becomes a human
memory. There are two types of long-term memory in humans: declarative (explicit)
memory and procedural (nondeclarative, implicit) memory. The former can be
consciously retrieved, while the latter is not directly available to our consciousness at
the time of retrieval. So the key to judging whether a memory is a declarative memory is
to see if it can be consciously perceived or not in a memory recollection.
In human motor control, the activation signal is generated by the procedural working
memory and sent to the input ports of the body system. When this process is saved in
memory, procedural memory is formed. Since the procedural memory cannot be directly
retrieved by the conscious brain, it needs to be modified. In this case, the procedural
3
memory must be combined with declarative memory to form a binding pair. Therefore,
the procedural memory can be consciously retrieved by retrieving its declarative
partner. (Zhang, 2019)
Initialization of the human brain
At birth, both declarative and procedural long-term memories are blank. Information
collected through the sensory system cannot be processed by the brain. The control
relationship between the brain and the body has not yet been established. In fact, at
birth, the brain does not even know the presence of the body. Therefore, in the first
weeks or even months after birth, the brain must go through an initialization process.
Two tasks are required during the brain initialization process. First, build a basic
memory database about its environment to gain processing power. Another task is to
create a basic brain-body communication mechanism by establishing declarative-
procedural pairs for all motor units. Thus, voluntary body control becomes available.
These two tasks can be accomplished through two key brain functions: information
recording (creating memory), and memory binding. (Zhang, 2020)
To better understand the initialization of the human brain, we can use the piano analogy
again. Think of the piano like a human body system and the piano player as a human
brain system. Imagine a person who knows nothing about music and piano sits in front
of the piano for the first time. Her finger accidentally touches a piano key and make a
sound. In this process, if she can record her finger movement and the sound made by
the piano, she can establish the relationship between the key and its sound. Similarly,
when other piano keys go through this process, the relationship between each piano
key and its corresponding sound will also be established. Since then, she has the ability
to play the sound she wants to hear.
Here, we compare
1. Piano keys versus the input ports of the body system;
2. The action of the player’s finger tapping a piano key versus generating
stimulation signals to the input port of the body system;
3. The sound produced by the piano versus feedback of body movement.
Motor programs
After the brain is initialized, the brain is able to voluntarily perform basic body
movements through declarative-procedural pairs. However, it is still a challenge when
doing a complex body movement, because the brain must consciously make every
basic body movement involved. As a result, the movement is not only not smooth, but
4
also requests conscious support throughout the process. For example, suppose a
movement M involving three basic body movements M1, M2 and M3
M=M1+M2+M3
In order to perform the movement M, the brain must perform each basic body
movement one by one, namely M1, M2, and M3. However, the good thing is that the
new memory is about to begin to build. During this first trial, the following information will
be saved as the procedural memory: when to generate a stimulation signal, and on
which input ports. In addition, feedback on the movement will be saved as declarative
memory. After enough repetitions, the newly formed procedural memory (M1-M2-M3
memory chain) will be enhanced enough to be retrieved. At the same time, a binding will
also be formed between the new procedural memory and the new declarative memory
(the feedback regarding the movement M). From then on, the movement M can be
executed only by thinking about the movement M.
The procedural memory generated above is often referred to as a motor program.
Obviously, the term is borrowed from the computer field. Unlike computer programs
created by computer programmers, motor programs are created based on our own
experience. It is generated based on the record of activities occurring in the procedural
working memory and is the accumulation of a large number of trials. Therefore, the
motor program is likely to be expressed in a statistical, distributed, and fuzzy manner.
Similarly, if we use the piano metaphor, the relationship between the motor program
and the procedural working memory is like the relationship between the sheet music
and the piano player. The motor program tells the procedural working memory when to
stimulate which input port and the sheet music tells the player when to hit which piano
key.
The above example also tells us that the input ports in the brain-body interface can be
considered as building blocks for motor modules. Since each control port is responsible
for controlling a basic body movement, two or more motor units can be combined into
an integral motor unit to produce a complex motor movement.
As a result, these later formed motor modules will become a higher level of input ports
for the human body system. In the same thinking, motor units, motor modules, and short
motor programs can also be combined to form a more complex motor program.
In my opinion, the brain-body interface is located in the primary motor cortex. The motor
modules (the integral motor units) are likely located in the premotor cortex and the
supplementary motor area. Detailed parameters about when to stimulate are likely
stored in the cerebellum.
5
In the above discussion, we generally assume that the human motor system is a non-
dynamic and non-complex system. This is not true in reality. In fact, the human motor
system is not only a dynamic but also a complex system. All subsystems can potentially
change over time, such as brain maturation, brain aging, muscle plasticity, fatigue,
changes in health status and environmental changes, etc. For such a constantly
changing and complex system, it is impossible to create an optimum motor program
suitable for all similar situations. Therefore, for most movements performed in daily
activities, adjustments, or corrections based on the sensory feedback are often required
during a movement.
In addition, as a dynamic system, improvements in any subsystem will contribute to
better control. For example, strengthen the muscle system and take a good rest. Also,
try to avoid complex system situations, or at least make it less complex. For example,
when playing basketball, try to shoot as close to the basket as possible.
Motor learning
Following a movement attempt, four sources of information are stored in memory:
1. Movement preparation: the Initial conditions (limbs, body, and the environment),
the task goal, the movement procedure, and planning, etc.
2. Movement execution: the activities in the procedural working memory during the
movement (information about which input port of the body system is stimulated
and at what time).
3. The sensory consequence of the movement (information about how the
movement is felt, looked, and sounded).
4. The outcome of the movement (including knowledge of results).
In the above list, only the memory recorded from source 2 belongs to the procedural
memory. All the rest is declarative memory.
Therefore, when we say motor learning, we should not simply interpret it as procedural
memory learning, because a motor program must be bound with declarative memory
before it becomes retrievable. A motor skill usually includes three basic elements:
declarative memory, procedural memory, and link. Creating a retrievable declarative
memory requires repetition. Creating a motor program (procedural memory) requires
repetition. Binding the above two together also requires repetition. That is why people
say repetition is the mother of learning.
However, creating a skilled motor program for each task is very time-consuming. More
often, when faced with a new skill, we usually borrow skills from other similar tasks or
create a motor program by combining several short programs using associative links.
6
The disadvantage is that we cannot perform the task as quickly and expertly as using a
skilled motor program.
Suppose a person has mastered the following skills:
• When you see the letter L, click the left button;
• When you see the letter R, click the right button.
Mastering skills means:
• Created declarative memories of the letter L and the letter R;
• Created procedural memories of “click the left button” and “click the right button”;
• Form the binding pair of the letter L and “click the left button” and the binding pair
of the letter R and “click the right button”.
As a result, when he thinks of L, the left button is automatically clicked; when he thinks
of R, the right button is automatically clicked. He can perform the task very quickly.
Now, suppose to give him a new task:
• When you see green, click the left button;
• When you see yellow, click the right button.
Although this is a novel skill for him, he should have no problem doing it slowly. What he
can do is to create an associative link between green and L, and an associative link
between yellow and R. The creation of associative links can only be done between
declarative memories, and only requires one or several attempts. Therefore, when he
sees green, he can use the associative link to find L and using L to retrieve the motor
program (click the left button). However, using associative links is a slow process. If he
wants to do it quickly, a direct binding must be formed between green and the motor
program. That requires repeated practice.
In the example above, completing the task by creating an associative link between
green color and letter L belongs to declarative memory learning. It should be noted that
in the case of learning to quickly complete the task, he has all the declarative memories
and procedural memories used to perform the task. The only thing missing is the correct
binding. Thus, the above challenge is neither declarative memory learning nor
procedural memory learning, but creating new bindings.
Note: This paper is a supplement to my earlier paper “Towards a comprehensive
model of human memory” (Zhang, 2016).
7
References
Zhang, Jie (2005): Continual-activation theory of dreaming, Dynamical Psychology.
Zhang, Jie (2016): Towards a comprehensive model of human memory, DOI:
10.13140/RG.2.1.2103.9606
Zhang, Jie (2019): A solution to the binding problem, DOI:
10.13140/RG.2.2.27258.34249
Zhang, Jie (2020): Initialization of the human brain, DOI:
10.13140/RG.2.2.34759.44965/1
View publication stats