The Difficulty Of Decision Trees in HEMA (Swordfighting)
Or, A Look At Event-Based Prospective Memory
There are a great many things that make learning sword fighting difficult. For the purposes of this article, I’ll be covering mostly one of those things, event-based prospective memory (EBPM).
We’ve all been there before. You know what you need to do. Your fingers clasp then loosen around the grip of your weapon. The heart races a little and you feel the sweat rolling down your neck. The sounds around you are distorted and amplified in your mask, and your vision has that familiar gauzed look, patterned by the mask you wear for protection. Your opponent circles to your left and you follow, deliberately placing one foot after another.
Left hand on the pommel, right hand at the crossguard, you rehearse in your mind the thing you have to do. Cut, step, bind, wind high, thrust, step, keep moving, win the point. Suddenly — it’s always suddenly! — a mass of black and flashing steel railroads into you. Before you know it, in a moment of overwhelm, your head rings — again — with the thunk! of their blade as your neck bends slightly in response to their weight.
You knew what to do! Why didn’t you do it?
As I watch my students, friends and complete strangers talk about and actually do training for sword fighting in Historical European Martial Arts (HEMA), I see some debate over the usefulness of decision trees in learning to fight.
On the face of it, there is a lot of good in a decision tree. A decision tree is a selection of options that lead to other options. It’s based on the premise that the sequence we call a fight is actually a series of logical decisions, if-then moments in which each interaction between opponents is a decision in reaction to what the other fencer is doing.
Even the masters of old often used these decisions in their teaching. “If he binds strong against your blade…” one might say, “… then do this …”
If … then …
It’s an important learning point. It’s how we fight: “If someone does this then the most efficient or effective thing to do would be this or that.”
And so we see a lot of teaching based on learning decision trees.
As anyone knows who has learnt to fight, putting the decision trees into action during an actual bout is very difficult. You get into the fight and it is hard to remember what to do; or else you remember it but it’s hard to make it happen.
In this article I deal with the first of these issues, the remembering element. The article will unpack what cognitive psychology has to say about it and explore briefly some of the motor learning theory behind what’s happening when we learn to fence. Hold on to your hats, because things are about to get academic!
Prospective Memory
When we think of memory, we usually think of recall. Recall is the process of bringing to mind something that happened in the past. It’s remembering your partner’s birthday, remembering what you were taught about warding a cut or what you were taught last week about polishing your blade. This kind of memory is known as retrospective memory.
When we enact a decision tree, we use a different kind of memory. This is known as prospective memory. Prospective memory is when we form a cue for something in the future. The idea is that this cue will trigger the memory to perform some kind of action in the future. In the literature, this is known as forming an intention (Einstein & McDaniel, 2005).
So, for example, if you need to ask your coach a question about one of the master cuts, you may set up a mental reminder to ask the question when you see them next. So far, so good.
This prospective memory is of a certain type. It’s known as event-based prospective memory because the cue that’s suppose to trigger your memory is an event. When a certain event occurs, you hope to remember to do another action.
That’s what happens with the decision tree. The if part of the if-then formulation is the intention. The then part is the thing you’re trying to remember to do.
So every time you try to actively engage a decision tree – or a branch of that tree – in a free flowing bout, you are trying to activate an event based prospective memory (EBPM).
In the words of everyone’s favourite not-left-handed swordsman, Let me explain. No, there is too much. Let me sum up.
What’s a decision tree?
A decision tree, for our purposes, is a series of options you can make in a fight. They usually centre around something like whether the opponent binds stronger or weaker than you at the sword. In that moment, you get to make a decision.
If she is strong at the sword, then you can yield your pressure, slide your blade along hers, over her tip, and slide back down along the other side, to strike them on their opening.
If she is weak at the sword, you can wind high into the Ochs and slide your blade down along their inside line, thrusting toward the chest or face.
Let’s have a look at some textual examples. Bold text is mine, to illustrate the various forms of the if-then formulation.
First, Meister Ringeck:
When you cleave-in with the wrath-hew (or otherwise high), if he displaces you with strength, then in-the-moment shove your sword’s pommel under your right arm with the left hand (Sigmund ain Ringeck, translation by Christian Trosclair, Wiktenauer).
We see here the if statement is, “When…”, which signifies the event. You are then instructed to form the mental intention (the then statement) to perform the action of thrusting your pommel under your right arm using your left.
Second, George Silver Gentleman (Silver’s Brief Instructions is replete with the if-then formulation).
If he come in upon the bastard guardant ward, bearing his hilt lower than his head, or but breast high or lower, then strike him soundly on the head which thinge you may easilye do, because his space is too wyde in due time to ward the same (Chapter 4.15, Brief Instructions upon my Paradoxes of Defense, cited in Wagner, 2003).
Notice here that the construct need not apply to when the swords are engaged. Decisions are being made at all times.
Third, Meister Joachim Meyer.
Now if he cuts from below against your left, then step out again toward his left, and cut with the long edge on top of his forte (1.31v2, The Art of Combat: A German Martial Arts Treatise of 1570, translated & edited Forgeng, 2006).
Once again, we find a device at play that relies upon this if-then formulation. It is central to the educational systems of these European martial artists.
A tree has branches
We call it a tree because it branches out. One decision leads to a response which leads to a decision, which leads to a response, and on it goes. The breadth of possibilities, in the end, can resemble a vast tree of the forest.
Of course, anyone who has tried to learn to actually use all of these possible decisions in a real time bout against a living opponent will discover that it’s not so easy to do when you’re in the middle of it.
There are several reasons for this. If you’re learning a new skill, then one of the major reasons is event based prospective memory. It’s a lot of trees and a lot of remembering we have to do and event based prospective memory (EBPM) is one of the reasons we find it so difficult to learn using them.
Working Memory and the Central Executive
Baddeley (2003) reviews the literature on the how we process initial memories. Most of us have heard of short term, medium term and long term memory. It’s usually the idea that we need to do something to get memories from short term into long term, and they go via medium term memory. Baddeley’s research has been particularly influential in understanding memory. Baddeley, however, developed and and expanded on an alternative model for memory: working memory.
Baddeley proposes (and the research seems to support) the idea of working memory as a system that temporarily stores and maintains information required for thought processes to occur. It is a limited capacity system, which means that it can only do a finite amount of work at any one time: there is an upper limit to what it can process or remember.
When it comes to EBPM and decision trees, we must use working memory to hold the information in place until such a time as we need to use it. There are four parts of working memory (known as multi-component working memory): the visuospatial sketchpad (for processing and manipulating visual representations of objects, say, for example, the shape of a sword); the phonological loop (for processing auditory signals, especially the acquisition and processing of language); the central executive (the part of the system involved in controlling attention and awareness and maintaining action when attending to something else); and the episodic buffer (a limited capacity temporary storage area, which brings together and integrates the information from the other components of the system. This could be understood as the working memory for the central executive).
Each of these components is limited in capacity. We can only hold so many visual images in our mind and manipulate them before we start forgetting the colour or shape of items or how many there are in a space. Our memory for verbal commands diminishes as we increase the number of syllables in the words remembered or when the next thing spoken (after the verbal command) is irrelevant to the task at hand (do we say more than we need to as a coach? Is what we say relevant to the goal of this training drill?).
Finally, there is the central executive. Baddeley (2003) says it has the functions of controlling attention. This amounts to focusing our attention on the relevant task, dividing our attention, switching our attention between tasks, and connecting our attention in with long term memory when necessary. Such tasks are also limited in capacity.
We know, for example, that if we have more than one task to attend to, that working memory throws some items out. Or if we have to do one task, while switching to another upon a certain cue, and then returning to the original task, performance will suffer in one way or another.
Prospective Memory and the Central Executive
You may remember, way back when you started reading this article, that I was talking about decision trees in HEMA. Full points for those who remembered! How does all of this relate to our learning a new play every week?
The decision tree is an EBPM: There is a cue (“Now if he cuts below against your left …”) and there is an intention which you must fulfil when that cue occurs (“then step out again toward his left …”). Of course, this is in the process of freeplay, so you must then return to your defensive and offensive movements, performing the other tasks required during the bout.
Every EBPM is a working memory process. You form a cue, which must then be held in your memory, while you walk out, hold your weapon, adjust your safety gear, and engage your opponent. You then, while still holding that cue in memory, engage in freeplay, wherein you don’t know what your opponent will throw at you. All the while, you have a little part of you that is trying to remember the cue (“Now if he cuts below against my left …”). So you are firstly performing a second task (freeplay) while performing the primary task of remembering the cue. Two tasks.
And this is where the central executive and EBPMs have a lot of difficulty: Any time you have to utilise an EBPM at the same time as a second task, it is highly likely that you will not remember the cue when it occurs. Further to that, the research shows that when you have to switch your attention between two different tasks (how many tasks are going in a period of freeplay?), you will probably lose the battle to remember that EBPM.
Now if your coach had given you extraneous instructions prior to going out to bout, then you have another problem: interference with the phonological loop. If you are also trying to remember the position of the opponent’s body, the length of their weapon, the placement of your body in space, then you are taxing the visuospatial sketchpad also. All of a sudden, it is no surprise that a decision tree becomes a very unreal prospect for learning to apply the teaching and learn it efficiently.
What To Do About It?
None of this is to say that decision trees are terrible or useless. Not at all. We need that knowledge. It’s not the process of a decision tree that’s the problem; it’s how we use it and what we expect from it that is.
There are a number of important things about memory that come into play here.
Central Executive Processes
The central executive seems to work off two types of memory:
- Schemas, which are patterns or habits (in our case, movements) that are triggered by contextual cues. They’re a general habit we revert to, such as when you’re driving a friend home from work but forget they’re in the car and you drive straight to your own home instead. You had formed an EBPM to take them home, but the schema for leaving work kicked in and automatically took you to where you normally go.
Or imagine you are bouting against an opponent, and you have a training goal to execute a defense against a cut. But as you absorb yourself in the moment, you find that you revert to your old habit, without even thinking — this would be a schema. There are many types of schema. - Sensory activating system (SAS), which is basically the system of attentional control I mentioned earlier. This is also a limited capacity system, and it is more a volitional, self-control system. It is the willing of something to happen. But, again, it has a limited capacity. The more self-control and deliberate attention you must use, the less you have to use.
You can imagine that you have had a big day at work, using your mind all day long, and then come to HEMA training. You’ve done the warm-ups and spent the last 45 minutes learning three new plays. Perhaps you felt a little frustrated and had to calm yourself down and then got out there to freeplay with your partner. You know what you need to do (If she hits me here, then I will do that). You’ve been practising it for some time now.
When you get out there to bout, however, it all seems to disappear. Well, you’ve had a big day. You’ve been exercising control all day long. Do you suffer from anxiety or depression and interact with people? Then you have been exercising control all day long. Do you have ongoing back pain? Then you have been exercising control all day long. Do you work in a loud, busy environment like a large shopping centre? Then you have been exercising visual and phonological control all day. This self control, this sensory activating system, is a limited capacity system. If you have been using it all day long, it makes sense, doesn’t it, that it would be a little bit worn out when it comes to playing with a sword against an opponent in real time?
Swordplay has many decisions, lots of sensory information and many things you have to keep abreast of. As a result, one of the first things the brain gets rid of is any event based prospective memory you have in store. It’s not just that learning sword fighting hard. It’s that we are already taxed, and so decision trees may not be the best thing to expect someone to use at that point in time.
Some Simple Solutions
There use to be a Kellogg’s series of commercials on television that said, “The simple things in life are often the best”. With regards to learning to use a sword, this is most definitely true.
Here are some ideas for how to use decision trees effectively in light of what is known about memory encoding, recall and attention.
1. Learn the stuff earlier in the night.
Put the hard-to-learn skills earlier in the night. Get the new stuff out while people are still fresh. Teach the new decisions trees while people are still able to concentrate and have enough attentional capacity to remember the if…then formulation better.
2. Put the stuff into practise earlier in the night.
Perhaps have two sessions for freeplay: one that focuses on putting into play the new decision tree-based skill and one for general bouting. Make a rule: you only do the new stuff in the earlier freeplay session, which should occur immediately after training the skill. This is the experimentation session. In the later freeplay session, toward the end of training, you only use things you already know — no experimenting with new material.
This earlier session cashes in on the attentional capacity: the students will still have a greater attentional capacity to learn and make wise decisions. If they are physically and cognitively fatigued, such as at the end of the night, then they will be physically unable to recall the decision tree well — it is a physiological reality. So why fight against it? Sounds foolish to me.
By training the new, harder stuff earlier, they can gain a bit more confidence, as well.
3. Distil and simplify your verbal commands
When people are training to do the new decision tree, work out the simplest way to give instructions and give them. Then, once they have their instructions, don’t — do not — say anything else. Just before they go out to experiment, repeat the instructions clearly and without apology. Coaches need to exercise discipline. Public speaking is a skill that you learn, in exactly the same way that sword fighting is a skill. Use the same discipline to speak clearly and give the right instructions in order to help your students. It’s not about you; it’s about them learning and being safe.
Perhaps you could encourage the students to verbally instruct each other. Just before engaging with their training partner, have one of them say, “When I strike from the left, remember to defend …” and so on.
In this way, we not only work with the EBPM but we also make use of some other effects (primacy and recency), which may be food for thought in another article.
4. Reduce the number of decisions
Make the learning bout about one thing only: the EBPM. We can reduce the number of decisions in a bout in many ways. Here are some, and certainly you will be able to come up with many more:
- Restrict the type of offensive strike that can be delivered. Perhaps only allow an overhead blow from the left and right and forbid all others.
- Instruct the student to only defend against the strike that warrants the new decision. This way they a still have two tasks and have to switch attention … but they only have two tasks. They don’t have to deal with how to defend against multiple types of attack, so their attention is divided a controlled number of ways, rather than an infinite number of ways.
- Only progress to a greater number of strikes as the student exhibits confidence in their ability; never before. Build their self-confidence, avoid destroying it.
- Reduce visual and auditory noise (and therefore the need to decide to not pay attention to it). Have only one pair freeply at a time when learning the new skills. Line everyone up to watch them and give positive feedback at the end. Then cycle through the pairs. Or perhaps encourage the use of ear plugs (which I encourage regardless, for health reasons). If this is impractical, then set up partitions so that they cannot see other pairs moving but have visual clarity.
- Avoid the use of masks with visual images on them. Do you know why international fencers use images of their flags on their masks? One very real reason is to create a visuo-cognitive disturbance for their opponent, which taxes their attentional capacity, delaying their ability to make decisions or remember the right thing. So when people are learning, get rid of visually novel stimuli on uniforms.
- Disallow verbal coaching from the sidelines when learning a decision tree skill.
5. Use context wisely
We recall best when the context in which we encode a memory is replicated. What this means is, if you learn a decision tree in a static context, such as lined up in two rows with the offending partners in the left row and the defending partners in the right row, then when you go to use it in the context of freeplay, of course it will be difficult to remember.
The context is completely different.
If you make the context in which a person learns the skill progressively similar to the context in which you want them to remember it, then they will be more likely to recall it when the time comes.
That’s why, when you walk into the kitchen to get a pair of scissors, but as soon as you walk in you forget why you went there, when you return to your bedroom where you first encoded the cue to get the scissors once you’re in the kitchen, you remember what it was you had to get. You have returned to the context in which you remembered the cue originally. So, with HEMA, we need to learn to safely and gradually expose people to the context in which they need to remember something. Decision trees can be helped by this.
6. Relax. Slow Down. Enjoy.
Easier said than done. But the fact is, you don’t have to learn this stuff now. You have a long time to learn it. Hopefully many decades. So enjoy it.
What we know about learning new complex skills like this is that if we are distressed, worried, anxious, and in a hurry, then we will have less attentional capacity and less ability to remember things well. There are many things to learn but trying to do it quickly won’t help you out. You’ll just end up going into a competition relying on your old schemas for movement rather than controlling what comes out in your movements.
It really is possible to look amazing when fighting. It really is possible. The problem is that people are in too much of a hurry. Be willing to take time out, sit with the movement and learn it, slowly, gradually, and very well. Learn to make the decision in a slow bout that’s safe. Then progressively increase the speed, only going as fast as you can without losing control.
Keep control, make decisions, until one day you look back at a bout and can think, “Wow! When did I learn to do that?”.
In learning to fight well with swords, there is no place for hurry. Your partner deserves better and so do you. So relax. Slow down. Be in this thing for the long haul. It really will be worth it.
Stuart McDonald is the Head Program Developer for the Glean Lachlann Estate College of Arms (GLECA) and head of research for Global Performance Testing Australia. Some areas of professional research include psychological science and performance, anxiety management and sporting syllabus development and delivery for adult audiences. He has developed the Western Martial Arts Coaching Course, which you can find out about here.
References
Baddeley, A. (2003). Working memory: Looking back ad looking forward. Nature Reviews: Neuroscience, 4, 829–839.
Einstein, G. O, & McDaniel, M. A. (2005). Prospective memory: Multiple retrieval processes. Current Directions in Psychological Science, 14(6), 286–290.
Meyer, J. (1570). The Art of Combat: A German Martial Arts Treatise of 1570 (translated by Jeffrey L. Forgeng)
Sigmund ain Ringeck (translated by Christian Trosclair), http://wiktenauer.com/wiki/Sigmund_ain_Ringeck#Long_Sword_Gloss
Wagner, P. (2003). Master of Defense: The Works of George Silver. Boulder, Co: Paladin Press.
