Exploring Differential Learning (Part 1)

Class at Bucks Historical Longsword follows a fairly regular format, we usually do about one “simple” game and one or two “complex” games (simple game being a game with fewer viable paths to success than a complex game, IE direct attack is simple, sabre march is complex), usually based on a passage from the Lew gloss that we are exploring that day. However it has become not uncommon lately to deviate from the regular structure in order to do something special for one reason or another. Sometimes we’ll do all simple drills, sometimes we’ll do an experiment of some kind, sometimes we’ll do a tournament instead of class, or various other possibilities depending on who shows up and what everyone feels like doing that day. In one class recently we did several variations of the Finnish Chicken game, and I thought it was fun and successful. 

I think this is a great game for finding the distance at which you can attack, and learning the mind game of when you should pull the trigger on your attack, when you should defend, and when you should press forward. That being said, it also dials you in on one specific distance, as do other drills like direct attack and soviet foil. This is an important distance because your direct attack to the head is a powerful weapon, but there are also benefits to changing it up, and I’ll explain why. 

The technical reason that I am going for is called a control law. Control laws are an ecological approach explanation for how we know where to move relative to an object or feature of the environment that we perceive. Imagine you are a baseball outfielder catching a fly ball, you see the ball flying through the air, and you have to run to the place where it will land in order to catch it. The information processing explanation is that the brain calculates the trajectory of the ball, and we execute a motor program to run to the place where the ball will land. In this explanation, the brain creates a mental model of the situation based on discrete cues, then we act based on that mental model. The ecological approach, using control laws, would be that we perceive the movement of the ball, and couple our own movement to it in order to arrive at the location where the ball lands at the same time as or before the ball gets there. So for IPA, we: 1) perceive a discrete cue -> 2) update our mental model, -> 3) act based on updated mental model; for EA, we 1) directly perceive specifying information (control law) from the environment -> 2) act based on specifying information.

So how does this relate to Finnish Chicken? Basically the idea would be that we are coupling our action to the relative position of our opponent, once the distance gets to a point where we think we can hit we throw our cut. The problem is that since the distance at which the fencer wants to cut is always the same or similar, they don’t actually have to use a control law here, they can come up with different ways to determine when their opponent is at the correct distance. This is called non-specifying information, since it is only applicable in this scenario. This is okay as long as you are doing only this game, but it will make it difficult to transfer the skill to unfamiliar opponents in fencing situations where the distance is highly variable. Because of this, we would prefer to solve the problem using a control law. Control laws are specifying information, which means they are generally applicable to a variety of scenarios, since they are not dependent on variables like opponent’s height, reach, sword size, etc. 

The best example I know of which illustrates this problem I heard from Sean Franklin. Imagine you are a batter trying to hit a ball, you need to know when exactly to swing the bat to hit the ball. A possible solution you can come up with is “when the ball is <x> size in your vision, swing the bat.” This will work in some cases, but it is generally not as universally applicable as establishing a control law based on the movement of the ball. It is non-specifying information, because it’s dependent on the size of the ball. To solve this, you can make the batter hit different sized balls. Then they will not be able to use their previous solution and instead be forced to couple their action to the ball’s movement using a control law. The control law is not dependent on the size of the ball, because it is directly perceived based on the change in size of the ball relative to our eyes. Therefore it is specifying information, and is more advantageous to couple our action to.

Going back to Finnish Chicken, if we want to achieve the same result, then the goal should be to vary the attack distance, and the way I tried to accomplish that was through changing targets. We started with a round of the normal version, then we did the same thing but legs only, then any target above the belt, then finally head and legs only. These differences vary the distance in different ways, some targets are physically further away so take a longer time to reach, but at the same time adding choices of targets and how to hit them makes it more difficult to parry and therefore increases the viable distance. 

Another aspect of changing targets like this is they introduce a variety of new strategies and tactics to the game, involving how to defend and how to attack the different targets. For example, with leg attacks, the rule is you are not allowed to step backwards, but that doesn’t mean you can’t pick your front leg up and move it back as long as you land it in front of where it was before so it’s technically a step forward. There were many different aspects of this, along with where and how to target and how to defend against them for people to focus on, while distance was still an aspect. I think it is a good thing that distance was not necessarily at the forefront, there was a study by Gabrielle Wulf and colleagues which showed that people learn to balance better when they are focused on a different activity while balancing than if they are focused on balancing alone. That’s not to say that they didn’t know they were learning distance, I stated the purpose of the variations as being because of distance ahead of time, but everyone had plenty of other things to focus on. 

The post-drill discussion for these variations was more lively than usual, everyone had something to share about their experience and what worked and what didn’t work. This is a good thing, and something that we usually don’t get when we do a class of playing games that everyone already knows. I think there is an intrinsic benefit to coming up with new games to give people different problems to solve, and that is a topic for another time. 

Differential learning is a topic that I have shied away from in the past, I still don’t fully understand it and I don’t want to go all in on it, but with activities like this I think I am starting to zero in on at least the idea of what it is supposed to do. My understanding of differential learning was that you need to have different conditions on each individual repetition, which we did not have here, but it is still the idea of a similar scenario with varied aspects. The idea here is to explore an idea, in this case distance, by varying what it means to have good distance. That is different from CLA, in which we are creating a situation that affords certain actions over others, with the affordances being similar to free fencing and therefore the actions being directly applicable. 

A similar idea can be applied to other games we’ve been doing lately such as Fechten im Grublein and Chess Fencing, in which there is no specific movement, action, or technique that we are trying to hone in on, but by doing them the fencers are forced to explore their solution space and come up with unique movement solutions that they might not otherwise think of. I think this may be an important aspect of the Ecological Approach beyond the scope of CLA.