It is clear that one way in which we can help facilitate the process of skill acquisition in sports is by not “throwing the learner in the deep end” right away. That is, not including all the complexities, variability and task demands they will face in competition right at the early stages of learning. This idea has been incorporated into recently developed models of periodization of skill training and has been given strong support by research showing the benefits of scaling equipment appropriately for kids, for example. In this post, I want to look at two different ways in which we can make a task easier for novice learners and how they align with prescriptive instruction vs self-organization approaches to skill acquisition.

Task Decomposition & Prescriptive Instruction

As I have discussed in detail now in my posts on using conditions vs constraints and why we add variability to practice, in the prescriptive instruction approach it is assumed that there is an ideal technique for executing a sports skill. A major consequence of this assumption is that it makes it possible to describe movement technique independently of the perceptual information. So, we can generate figures like this:

Notice how in this figure there is no reference at all to what the ball is doing (e.g., whether it is inside or outside, fastball or curveball). In the Prescriptive Instruction approach it is assumed that the majority of what it takes to be skillful involves being able to repeat the same basic pattern of movement (e.g., arms bent, elbow close to hip) regardless/despite of what is happening in the external environment. This is of course, also often reflected in the typical instructions or cues given to an athlete in this approach. 

If we accept this assumption what it also allows for is task decomposition. That is, we can break a skill down into sub-component parts and do part-training. If the movement form is going to be highly repeatable and has many aspects that occur independent of the information from the environment then we can simplify the task for a new learner by taking some of the perceptual information away. There are many examples of this in sports including hitting off a tee in baseball batting and practicing the toss in a volleyball serve without actually hitting the ball.

In both of these cases we are decomposing the skill into parts and decoupling perception from action. In the baseball example, we are treating the act of swinging as separable from perceiving the flight of the incoming pitch and in the volleyball we are treating the toss as separable from the ball strike. The idea is that once the ideal movement patterns are learned we can plug the part or component back into the whole.

Task Simplification & Self Organization

In the self-organization approach to skill acquisition it is believed that a movement cannot be separated from the information which is used to control it (aka keep em’ coupled!). So, instead of the figure above where we describe the ideal baseball swing independent of the ball trajectory we get this…

This figure shows Mike Trout swinging at two very different pitches – low and away on the left, and up and in on the right. As you can clearly see, the movement pattern is very different for these swings. In one the lead arm is straight, in the other its bent. In one the front foot is turned in, in the other it’s turned out. In the one the back elbow is out away from the body, in the other it is pulled into the hip. So, in other words, there is no one ideal swing like illustrated in the figure of Jose Altuve above. Instead, a good hitter will need a bunch of different movement solutions that vary with the task constraints (e.g. the pitch trajectory). In the self-organization approach this is called movement degeneracy – achieving the same goal (hitting the ball) with different patterns of coordination. Why are these swings so different? Because the perceptual information about ball trajectory that the hitter is using to regulate the swing is very different. Thus, in the ecological approach, it is argued that we can only understand a movement by also considering the information. For this reason task decomposition (in particular, decoupling perception and action) is strongly discouraged. Movements and coupled to information and information is coupled to movement. If you decompose a skill into part training the performer will not develop a way to use the information to come up with a movement solution and/or they won’t educate their attention to the specifying information. Movement solutions only make sense when considered in the context of informational constraints so they should be trained that way.

Evidence in support of this view can be seen in a study of volleyball training by Davids et al (1999). Shown in the figure below are ball trajectories for whole, coupled training in which the server had to toss and strike the ball (a) and uncoupled, part training when they only had to perform the toss like in the video above (b):

As you can see, the ball trajectories are very different in the two cases. The uncoupled tosses were higher, had greater acceleration and were more variable. There are many other similar published findings including research examining hitting off a ball projection machine vs a bowler in cricket and research looking at saving a penalty kick in soccer in uncoupled and coupled conditions. Although, there are also counter-examples that can be seen (for example, research showing effective transfer of training for anticipation training using temporal occlusion) it has been argued that even greater training benefits and expertise differences can be seen if these types of training don’t use task decomposition. For example, instead of having a baseball batter make a passive verbal response to a pitch recognition video, this activity represents a step away from task decomposition:

So, if we don’t want to decompose a task what is the alternative way to make it easier for a novice? In the Self-Organization approach the preferred method is task simplification. In task simplification, the whole movement is always completed and the movement and information always coupled. To make the movement easier for a new learner we scale down the skill while keeping its basic structure. For example, we could reduce the ball speed, the distance and/or number of different pitch locations in batting; use a lighter, softer ball and hit to a lower net in volleyball, or reduce the number players and/or increase the size of the goal in team sports like soccer. The two keys here are that: (i) we want to allow the performer to develop the same type of information-movement coupling they will use for the full skill and (ii) we want to make sure that we still have some degree of variability in practice conditions so that the performer has to learn to solve movement problems (plural!) and to prevent them from developing a solution based on non-specifying information that may not transfer to the fully, scaled up skill. A good example of this can be seen in the aforementioned work on scaling equipment for kids (which is a form of task simplification). Even though the kids are using smaller rackets and lower compression tennis balls, perception and action are obviously still coupled and there are still variations in the time to contact, bounce height and direction. We will be seeing a lot more examples of this approach in the upcoming Practice Activity Analysis I will be doing on the Perception & Action Journal Club: