Communication Breakdowns

A crash course in the pieces we need in place to talk about training

Jul 06, 2026

Article voiceover

0:00

-31:58

Breaking my rule here of trying to stay focused on practical application. I have wanted to try and communicate this idea for a while, but it is very nuanced and complicated. I was spurred to write this after listening to some debates at the Sports Movement Skill Conference 2026, and reflecting on some previous discussion/comments I have exchanged with people.

Recent Discussions…

An average number thrown around for the time to get a blackbelt in Brazilian jiu jitsu is 10 years. However, sometimes I’m not sure if that is because Brazilian Jiu Jitsu is complex in the way we like to talk about it, or if we are poor at teaching it.

There has been a lot of spirited conversation the past few years in the Brazilian Jiu Jitsu / grappling community about training. Online discussion is often polarized, and at best filled with misunderstanding. In most conversations I have read and listened to, I still feel unsatisfied. Ignoring the worst aspects of online arguments, it’s been frustrating how heated or shallow some conversations seem to be. This isn’t terribly surprising though. There is a thread absent from these discussions that sets them up for pointless friction.

So what is missing from these conversations?

It starts with what we think we mean when we say the word “skill”!

What Even Is the Skill of Grappling?

Taken as a whole, what kind of skill is grappling? Motor Control theory provides us with a set of useful ideas and concepts that we can apply regardless of our starting beliefs. There are a few broad categories we can use. The first is the idea of a discrete vs a continuous skill.

• A Discrete skill is something with a clear beginning and end. A classic example of a discrete skill is dart throwing.
• A Continuous skill is something that doesn’t exactly have a clear beginning or end, like swimming.

A second set of categories we can use to identify skills is Closed vs Open.

• A Closed skill occurs in a predictable and stable environment, again like dart throwing.
• An Open skill has a changing or unpredictable environment that the performer must constantly attend and adapt to, just like a grappling match!

An important thing to note here is that the activity of grappling is very far from being a closed and discrete skill. But when we look at how it is taught, it is divided cleanly into discrete and closed pieces. Why do we think this might be the case? Where does the idea of breaking a Continuous and Open skill like grappling into Discrete and Closed pieces come from?

One other concept we need to cover here is the idea of Transfer. Transfer is the positive or negative change in performance based on previous experience with a task/skill. This is the result we are looking for when we talk about learning. As coaches we probably all have an interest in designing training that promotes transfer, but how do we determine what types of things will transfer positively?

With the idea of Practice Specificity!

Practice specificity is the idea that “what you learn depends largely on what you practice” — Lee & Schmidt, Motor Learning and Performance 7th edition (This is THE Information Processing Textbook by the way!). The implication here is that there is sensory feedback that is part of performance. This sensory feedback is a key ingredient to the learning of a skill. If that sensory feedback is diminished or completely absent, we may promote transfer in the wrong place. We may be spending training time practicing for a different thing. This may mean that just because something looks similar to a skill, it might not actually be the same as that skill. This can lead to a very inefficient use of practice time.

There is a recent example of this from Cooper, Daniel et al 2024, in which they did a metanalysis of law enforcement marksmanship. Officers in low stakes practice tasks performed well, but when faced with real world pressure those accuracy gains didn’t transfer well. Their accuracy went down when they needed to use their marksmanship skills for real. When the sensory feedback of real-world pressure was added to training, the accuracy score in real world deployment increased! As a rule: you get better at doing the thing, by doing the thing, but we need to be careful about what we think counts as doing the thing (Thanks again Scott Sievewright!). Our starting assumptions shape what we think qualifies as doing the thing.

So, the way we classify the skill of grappling can inform how we think that transfer happens. But how do we break skills into trainable parts that count as “doing the thing”?

Taking Skills Apart

This is where we can run into some issues. As a final concept from motor learning, we have 2 big tools we can use to teach different kinds of skills: Decomposition/Segmentation and Simplification.

• Decomposition/Segmentation is when we break a skill up into discrete pieces, practice them separately, and then put them back together hoping they will fit.
• In Simplification we reduce the level of complexity or difficulty of a skill, or aspects of a skill, so that the sensory feedback required for the specificity of transfer is maintained. We then scale up the complexity and difficulty as needed to get to regular performance of the skill.

The tools we use to build or take something apart will determine the structure of that thing. In the case of an automobile this method is straightforward. All the pieces of a car were made by humans to fit together in a certain way and fulfill a certain function. All those pieces match tools they were designed to fit with. I can take the car apart, decompose it, with my wrench and put the car back together and still get a working car (I personally cannot do this; I can barely jumpstart a battery). Pieces of the car are specifically designed to have a single function as well. The brakes of the car are made to stop the car, nothing else. The relationship between a part and a function is 1:1. This relationship gets weird with movement and biology though. Take our hands for example, they don’t have a single function! They can do so many different things in different contexts! Where do I draw the line for what something is for when it comes to my body? My body doesn’t split into pieces and functions as cleanly as the car does. What tools do we have to analyze movement, and what types of parts are we looking for to break that movement into?

When we decompose a skill (especially and open and continuous skill) we treat the body and grappling like a car, so we are assuming a few things:

• Where the seams are in the skill we are decomposing. That we are using the tools appropriate for how the “car” is put together. A wrench rather than a cutting torch.
• What sensory feedback is or isn’t necessary for learning. No matter how well we think we put the car together, it is the fuel that is not a part of the car that makes it go. Our tools may not let us realize that.
• That the decomposed parts will reassemble cleanly into the full skill. Do the parts we are breaking a skill into have the same relationship the parts of a car do?

This is the typical “steps” approach, we have the athlete repeat the steps of a knee cut several times on their partner, and then have them positional spar where they are trying to perform the knee cut pass.

From the simplification side, we can try to make parts of a skill easier so that sensory feedback is still present. In this way we can make sure our practice activities stay true to the specificity of transfer principle. The parts we divide the activity into may better reflect the whole. More sensory feedback remains present in the practice task. So the athlete never really has to piece anything back together. We just start making things more complex until we get to the “real” thing.

Using the knee cut example: We would divide that knee cut into a series of tasks the athlete can interact with. For the first task, the athlete’s only job is to keep their leg in between the guarded player’s legs (you can’t knee cut without your leg going between your opponent’s at some point). This is very simple! Once they get exposed to that task we can start adding complexity. Can they pull their leg out and put it back in again? Can they drop their knee to the mat and keep it there? In this format the athlete always has some level of sensory feedback closer to what they will have to deal with when performing for real. We don’t run the risk of accidentally removing something that may help promote learning. Crucially, we can now pinpoint pieces of the interaction to specifically practice them, in a manner close to how they would actually be performed.

However, with decomposition, the “Knee Cut” also exists as a “technique” we can point to and name. This technique is a large movement containing many different things. Yet the label of knee cut obscures all of those component pieces. With simplification, the way we set up the training tasks makes the “Knee Cut” not a singular thing! Instead, it’s a consequence of a consequence of a consequence all nested inside each other. A leg split – leg pin – body pin – leg extraction. I need to put my leg in between my opponent’s leg and be able to keep it there, I then need to hold my opponent’s leg still while holding their body still to some degree, and then to finish the pass I need to pull my leg out from between their legs while still pinning my partner to the mat. The tasks become focused on the functional aspects of the movement, rather than just the movement. This can be a mouthful to talk about but is important for designing training.

At the start of training this seems simple. But coaches don’t just set one activity and leave. Where things get complicated is in the decisions both approaches allow you to make moving forward. Let’s say I have an athlete that is struggling to learn that knee cut movement.

Decomposing: I may provide a different version of the knee cut for them to practice. I could provide a different focus for them as a step in that drill. But I’m repeating a similar process that didn’t seem to work the first time. If the decomposition is not fitting with the athlete, is the answer to decompose things “better”? What I don’t have is a systematic way to diagnose and adjust this problem in a live interaction.The live interaction is where the athlete is encountering the problem! Positional sparring just makes the problem more apparent. The tool of decomposition does not let us easily retain the sensory feedback that is necessary for transfer. We are just hoping the athlete will eventually figure it out. Most of the time they do, but what if we could target things in a more intentional, focused, and efficient manner?

Simplification: We can design training around a set of tasks that allow knee cuts to happen. Keep a leg in between your opponent’s legs, get your knee to the mat and keep it there, etc. We could start at the “end” of the knee cut and practice maintaining the position. The athlete gets significantly more time practicing the actual thing they may be struggling with, while hopefully limiting them to only the necessary frustration inherent in learning.

Both these diagnostic processes will hopefully get us to the same place over time; both methods clearly “work”. But we are taking apart the “car” with a different set of tools. One of these toolsets lets us intentionally take apart the car in a way that we can easily put back together and drive!

Why Should a Coach Care About Any of This?

Your understanding of how to classify and organize a skill into training is determined by your philosophy of coaching and skill acquisition. What you do in training is a direct consequence of this. Everyone has a philosophy that they follow. Even if they can’t articulate it. How we decide to “cut up” the activity of grappling is informed by that philosophy. Where you decide to locate what a skill is, and how you classify the skill is the foundation. If you are biased towards using decomposition you may not be able to see important functional pieces of a movement. If you are biased towards simplification you need to have a systematic understanding of the tasks that are important for a skill or movement.

How you understand what skill is and how to build it gives you access to tools like simplification or decomposition. Those tools give you different abilities to take apart the sport of grappling into trainable pieces. Those tools also give you a different perspective on the process of learning, when transfer is taking place and what indicates it.

But where did these initial biases come from? Why does one tool or the other seem to feel like “common sense”? There is a split here, but it’s challenging to articulate! The split is based on how you define “skill”, before you even begin classifying it using any of our categories! This is the split that seems to cause so much trouble for the grappling community.

As a simplified example of where this split occurs we can provide a little thought experiment. This example comes from Louise Barrett’s “Beyond the Brain”:

“When was the last time you danced? …Now, as you think of yourself dancing, think also about where the dance is located – Is the dance inside you? Is it some kind of state that you’re in? Or is the dance something simply that happens to you? Makes even less sense, now, doesn’t it? Because dances just aren’t like that. And that is precisely the point: A dance is something we do, not a thing we possess, or a state we occupy.”

When we do grappling, where is the activity of grappling? It’s in the interaction between the two people who are doing the activity (although depending on who you are training with it can sometimes feel like the other person is doing grappling, and you are just receiving the grappling). The grappling cannot occur without that interaction. Even with all the technical knowledge in the world you cannot grapple sitting around and thinking about it by yourself. As the great Roger Gracie said:

“You can drill for a month, after that month you should have an understanding of how the mechanic works, but you still have no clue how to apply the movement against a resisting opponent, you will never ever know how until you apply with a fully resisting opponent” (emphasis added). https://www.youtube.com/clip/UgkxUrAGohlLWgcMJfRm1keDga61fHO4Jokr

So what was actually “learned” during the drilling?

For me, this is a question of where we assume the skill of grappling is located, where it “starts”

If you start with an assumption that grappling or skill in general starts primarily in your head: you may opt for practice activities that ignore or de-emphasize the interaction. You “drill” mechanics outside of the interaction that are somehow going to fit into the context of the interaction. You can “drill for a month”.

If you start from an assumption that grappling or skill starts primarily outside the head, and inside the interaction of 2 people and a surface to grapple on: you may opt for practice activities that keep the relationship intact. You practice small, focused, pieces of the interaction so you can always “know” how things work against a resisting opponent.

For the purposes of running training, you will start from one of these points.

So roughly from each assumption, practice could flow like this:

• Grappling primarily in your head → The instructor decomposes and teaches you a movement or skill, the athlete rehearses the skill, and now that the athlete has “learned” the skill then they go and apply it to the game
• Grappling primarily outside the head → The instructor creates simplified interactions for an athlete to practice a skill, the athlete explores the relationship of that skill to the interaction, the athlete learns to maintain or modify that relationship in the future

Each starting assumption is going to provide you with different tools for organizing practice. Each starting assumption influences the way you deliver information, the way you expect to see learning happen, and the way you look at the skills necessary for training.

While it can be tempting to sidestep all of this, claiming things are just “Another tool in the toolbox”, please reflect on this. A tool does not exist in a vacuum waiting to be plucked at random.

You still have access to the full toolbox no matter your starting point but:

Your set of assumptions (consciously arrived at or not) determine the tool you reach for.
Your set of assumptions determines what you are trying to do with that tool.
Your set of assumptions influences what tool you reach for next.

If you’re only thinking about collecting tools, you have not solved the whole problem.

You need some critical thinking to determine why and how you pick your tools. However, we don’t need to panic, this is actually a reflection of a conversation that has been going on for a very long time. The grappling community is not on its own trying to work these things out, all we need to do is take a look back. These assumptions actually have a long and fascinating history!

Bodies Moving Through Places in Time

There is a long rich history of trying to describe coordination and control of movement. This history has also always been filled with metaphors! Our body is a chariot driven by our soul! Wait no, it’s a series of aqueducts of life force to our muscles. Wait, maybe it’s various animal spirits flowing from our brain to our muscles. Hold on, maybe we are clocks? No, No, we are definitely a computer! We run stored software on the hardware of our bodies, just don’t ask who wrote the software or where it’s stored.

We have always used metaphors for movement and our bodies. Those metaphors bloom out into our culture and language. How many times have you heard someone ask for a moment to “process” something? What exactly are they processing and how? Is it “just” a figure of speech, or is it a hint at a starting assumption for what we think the skill of thinking is?

This gets us to an idea from Thomas Kuhn’s “The Structure of Scientific Revolutions”. There are 2 essential concepts we need. One is the idea of a Paradigm. The current accepted framing and explanation of something is called the “Dominant Paradigm”. Over time, as new questions are posed and new evidence is gathered, that paradigm can be weakened. Over the course of a generation science can potentially have a “Paradigm Shift”, where a new set of explanations takes the dominant role in science and culture.

The second idea we need is the one of “Incommensurability”. This is the idea that 2 competing paradigms will have aspects that almost directly contradict each other. There are ideas between paradigms that cannot readily mix. Think about the conflict between Heliocentric and Geocentric theories. They can’t exist together! These ideas are ones that can cause lots of friction between groups. Since the ideas don’t easily mix, each paradigm struggles to understand things from the other’s point of view. It can be the source of incredibly heated conflict.

Skipping over a ton of history, let’s visit the 80’s where there are 2 “paradigms” trying to make sense of things in Human Motor Control.

• The Motor Systems Approach — where something like “Information Processing” would fall - roughly a bias towards decomposition as a starting point
• The Action Systems Approach — where something like “Ecological Approach” would fall - roughly a bias towards simplification as a starting point

In the 80’s there was a conference about these two approaches, trying to discuss which approach was more useful, or if they could be combined. This conference resulted in an edited volume labeled “The Motor Action Controversy”. The contributors, editors, and referees are a who’s who of giants in the field of human motor control. We have inherited this demarcation, and the split remains today.

You can go extremely deep and broad on this debate, from the rigorous empirical science all the way to abstract theories of mind and reality. It’s very humbling to see how hard people have been working on these ideas.

The debates are interesting, the theories are interesting. But what I find most amusing is that our sport of Grappling and the drama over training the past few years are echoes of this history. We are doing a pale imitation of the same types of arguments and discussions that came before us. The Kuhnian concepts of paradigm and incommensurability explain exactly why: we have two sets of assumptions that don’t easily mix, and most of us don’t even know we’ve chosen a side. If you are not aware that you are working from one Paradigm, people coming from another may come off as extremely confusing!

The thing to remember is: Both sides are roughly trying to explain the same thing, they just disagree on the routes we should take to get there.

What We All Agree Needs to Be Explained

If you crack open and read a textbook on motor control (I have, a lot!) you often see a common layout. There will be a chapter that talks about the differences in motor control theories. For people who aren’t aware, “information processing” and “The Ecological Approach” / “Dynamic Systems Approach” along with “The Constraints Led Approach” are mentioned in all of these textbooks. The common talking point in Grappling that these things are niche ideas is flatly not true. Both these ideas have a rich set of empirical data that justify their use.

Another thing that is present in the literature is a bunch of theory agnostic ideas. These are phenomena we have observed in motor learning, like the specificity of transfer, that any motor control theory will need an explanation for. This presents a problem for coaches:

If we want to have discussions about training and we both don’t agree on what we are trying to explain, how can we decide what the best way to explain it is?

This is why it is so important to have some understanding of theory when forming your coaching practice. It gives you a way to talk about and investigate your beliefs versus what you are doing.

I want to point out that relying on received wisdom and tradition is not necessarily bad. It is entirely normal and practical; we cannot go about changing our entire set of beliefs every day. But you may run into situations where the assumptions baked into your ideas (having possibly been received uncritically) can limit your ability to coach. If your theory of coaching doesn’t let you diagnose a problem properly, you will struggle to fix it. If you aren’t sure what you think and why you think it, evaluating new ideas is going to be impossible. If you want to be a person who takes coaching seriously, this is a mandatory piece of work.

So, what are some things motor control theories agree that need to be explained? Michael Spittle’s “Motor Learning and Skill Acquisition 2nd edition” contains a nice list of things that each motor control theory needs to be able to tell a story about. I want to give a selection here that is relevant to the conversation:

• Skill Acquisition – Both theories need a story about how we learn and improve at a skill over time, and how we can measure that. I can go from whitebelt to blackbelt
• Variability – Different people doing the same type of movement look different. The same person repeating the same movement will look slightly different every time. Why and how does this happen? The knee cut you drilled looks different to the knee cut you actually performed
• Degrees of Freedom – Of all the ways I could move my joints and muscles to complete a task, what guides the selection of any given movement? My body ends up moving somewhere!
• Motor Equivalence – How many ways can a similar action be performed? I can flip a light switch with my foot, my nose, and my hand. When does one get used versus the others if they all achieve the same goal?

I want to stress, these are not “eco ideas”, these are not “IP ideas”. These are things we empirically observe in motor control that the theories are trying to explain. Michael Spittle provides a quick look at how these are explained by each theory:

From an Information Processing Approach

• Skill Acquisition – We create a motor program in our head. Over time we refine this motor program with sensory feedback, building a schema that lets us respond to a wider variety of movements in a similar class.
• Variability – The motor program represents a class of movements. We send the same commands each time; the signals perform differently based on the class of movement and regulatory conditions present.
• Degrees of Freedom – We develop a program that defines the relevant and irrelevant pieces of our body to move
• Motor Equivalence – The parameters of the program are modified based on regulatory conditions present in the task

From an Ecological Approach

• Skill Acquisition – Our body uses sensory feedback to organize towards “attractor” states. Sort of like a habit, but with way more math. This continuously happens over time as we become more sensitive to sensory feedback.
• Variability – The movement pattern organizes based on the constraints present in the interaction between the 2 athletes and the surface they are grappling on. The movement is “stored in” the interaction, not inside anyone’s head.
• Degrees of Freedom – Based on our unique body and how it interacts with the task at hand, movement organizes based on the physical reality of the demands of the task.
• Motor Equivalence – Our body, the thing we are trying to do, and the place we are trying to do it determine how we functionally achieve something, not a stored set of parameters.

Each approach is trying to explain the same types of things. We are trying to teach motor skills that are adaptable, variable, and persist over time. One set of explanations may seem better or less confusing to you, based on your starting assumption of what skill is!

Based on your theoretical viewpoint, the whole structure of training is going to be rooted in these explanations.

This gets us to the place where one side can see “Knee Cut” as the whole skill, while the other sees “Controlling distance between two bodies” as a skill, with a “Knee Cut” as a way that skill is expressed.

These are very different things! This is why sometimes the language in these discussions can get so confusing. What’s at stake is that the language we use to talk about the activity directly informs the way we break that activity apart to practice it.

Broadly it comes down to this:

We have one side claiming the language of traditional BJJ technique is too big of a unit of analysis. That It doesn’t allow us to accurately analyze the sport and all the gray areas between the big landmarks of technique we are trained to look for.

We have another side saying the unit of analysis is correct, and the gray area is solved by “details” that are to be shared when an instructor deems necessary.

But when it comes to designing training we start from 2 different paths:

For Skill Classification:

A “Knee Cut” is a single unit, and we provide “details” on how to achieve it based on decomposition. You have a series of explicit steps proceeding from an assumed start point.

A “Knee Cut” is a collection of different skills nested inside of each other. You have a series of tasks that need to be achieved to a degree at different levels of distance from your partner. The starting point is defined by the needs of the athlete/student.

From an student/athlete perspective:

I am going to be given information up front on “how to” do something correctly.

I am going to be provided appropriately difficult problems to be solved from a position.

From a Coaching/Instructor Perspective:

I am looking for the details I provided to be implemented by a student to measure transfer.

I am looking for focus and interaction with the tasks I set up, not necessarily “success”. That focus and intention lead to outcomes over time I can use to measure transfer.

I hope this makes it clearer why the vocabulary we see growing seems to be a constant point of contention. Why saying things like “It’s just positional sparring” completely misses the point. We are approaching from completely different sets of starting assumptions that are informing the way we look at training.

It is an argument not just about training, but about how we as a community want to talk about and analyze our sport.

The language we use leads us down very different paths when we are talking about designing training, providing feedback, giving interventions, and assessing learning. It plays a role in how we build culture in our training rooms. An athlete focused on trying to hit techniques as a marker of progress versus an athlete looking to fulfill progressively more complicated tasks will develop a different relationship with navigating failure. We are talking about different worldviews, different paradigms.

Specific Transfer of What, Exactly?

Finally, I want to revisit that idea of specificity of transfer. We know certain elements of training need to be similar to the performance environment to promote learning. What we still don’t have a great understanding of is what parts, exactly, need to be similar, and how similar do they need to be?

From the Information Processing side, we have ideas like “Identical Elements” and “Identical Processing”, where we think the movements or cognitive load need to be “close enough”. The problem is this is very hard to put into practice. In decomposition, we often think of difficulty as increasing the number of steps, or the details contained inside of the steps. This could be increasing difficulty in the wrong direction. A knee cut in a drill and a knee cut “for real” might look the same, but things are happening in both cases for entirely different reasons. Does it “just” need to look the same? Or is there something else we need?

On the Ecological approach side we have the idea of “Representative Learning Design”. This is a framework that helps us keep sensory information relevant to the skill we are trying to practice intact. An example Rob Gray gave at SMSC 26 concerns baseball. If we want to have a batter practice hitting a pitch, we wouldn’t have them hitting off of a tee. We would ideally want a pitcher throwing a ball to them. The sensory feedback, like the movement of the pitcher, is available to the batter to help them adapt and learn how to swing. If we used a pitching machine, that information from the body movement of the pitcher wouldn’t be available. This is where skill analysis rears its head again. Where does the skill of batting start? If the skill starts when the ball is in the air you could default to using a pitching machine. If you think the skill starts when the pitcher begins to move you may default to using a real pitcher. Depending on your viewpoint, practice will look different.

There is an interesting study on balancing done in Serrien et al 2017 related to this idea of things “looking” similar when they aren’t. Would practice in one kind of balancing task transfer to another balancing task? The intuition is that since both tasks looked like they required “balancing” they are related, and we should see some transfer. Yet, there was low transfer. The two balancing tasks didn’t have very much in common even though we would both say they required balancing skills. The similarity between two tasks is not so easy to eyeball. It is up to the coach and their knowledge to be able to identify this in training.

So think to yourself: when does a knee cut “start”? Is there a discrete moment? Depending on your viewpoint, you may say yes or no, and you would have a rationale for doing so (hopefully). How you view that start point informs how you try to practice it.

What Do We Do Next?

So, is grappling complex or have we made it complex? It’s a little of both, but how are we locating and identifying that complexity: in the complex continuous interaction of 2 bodies, or in the discrete decomposed snapshots of movements we call techniques? Are these tools we inherited still the right tools for the job, or are they creating problems we never intended?

The goal isn’t to wipe the slate clean or completely change your mind. It is to recognize that every drill we run or cue we provide to an athlete carries:

• An assumption of how grappling works
• An assumption of where skill exists and how to learn it
• An assumption of why our decision is the best one to make at that moment

When you run practice, what are you practicing? What are you giving information about? How much information is necessary or made redundant by what you do next?

The questions posed in the 80’s still seem to be alive and well today. People have dedicated entire careers to having these debates. We might never get a complete explanation; things are never so simple. Yet we aren’t completely shooting in the dark! We have some powerful concepts and science to orient ourselves practically. We don’t need to re-invent the wheel by ignoring all this work. We can use it to better investigate our own assumptions and run better training.

As coaches we should foster curiosity and learning in our own coaching practice, just like we try to encourage that for our athletes.

As the sport grows, and as science continues, will you be prepared? Will you be ready to grow your skills or evaluate new and old tools? Or will your current assumptions keep you stuck.

Grappling with Skill

Discussion about this post

Ready for more?