Verbal Instruction in Sports Coaching

A few days ago I posted a Twitter thread about the role of verbal instruction in sports coaching. It's a thing that comes up a lot as a key point of contention between ecological and non-ecological types, so I wanted to think it through. This post collects what I said in the thread, and adds a few things that have occurred to me since. 

Coaches want to be able to give their athletes instructions. Usually, this is about technique; ‘place your feet here’, ‘angle your club like this’, etc. This fits with the idea of coaching as imparting knowledge. Ecological coaching approaches tend to veer away from verbal instruction like this, and focuses on creating constrained environments players find their own way through. This becomes a key point of contention. From the traditional point of view, it makes no sense to not verbalise instructions. 

So what’s the ecological motivation for avoiding this?

First, verbal instructions about technique come from coach intuitions and experience about how best to execute a movement. There isn’t really any overarching evidence base that constrains their choices. Experience is great, but using it as your only source of ideas is a problem, because our intuitions about where behaviours come from and how they work are generally pretty bad. This is why we have to do science on this stuff! 

A specific example comes from the literature on learning & transfer of learning. This is particularly relevant because coaching is about exactly these things; promoting skill acquisition during training in a way that shows up on game day. 

People have studied learning and transfer for 100+ years. It swings in and out of fashion, as the field tries to figure out when learning would transfer to a different context, fails to do so, loses interest, then tries again 30 years later using slightly different words but not fundamentally changing the approach. 

Learning transfers when two tasks have meaningful overlap. From Thorndike all the way through to analogy researchers in the 1980s, everyone has tried to map that overlap based on common sense task analyses. For example, surely learning to balance on a slack line and balance on a beam both require ‘balance’; or surely learning to program should train logical thinking in the wider world. 

Turns out, no. Like, not at all. Something is wrong.

What’s wrong is the task decomposition. The actual mechanisms of behaviour are not particularly accessible to common sense, and so common sense is a very limited basis for identifying important parts. This implies that verbal coaching instruction based on common sense task analyses won’t help the athlete. Worse, it will actively get in the way! 

We need a more evidence based way to decompose tasks. Science is what we do when we need to develop knowledge beyond the scope of common sense. So what does the science say?

First, the science says there is no such thing as ‘the correct technique’. Because of motor abundance, we have many ways to achieve any action. Some are better than others, some are just wrong, but even the right basic solution has huge flexibility available. 

Second, there are no behaviours, only behaviours-in-contexts. Remember ‘balancing’? Turns out there is really ‘balancing-on-slack lines’ and ‘balancing-on-beams’ and they have so little in common training on one doesn’t help on the other (Serrien et al, 2017)

Third, if context is part of the makeup of a behaviour, then we need to think about perception and how it connects us to those contexts, and we need to include those contexts in our instruction. Note that coach intuition rarely engages with this fact, because it isn't intuitive! It's so unintuitive that notions like the extended mind and perception-action loops that include the environment are still considered radical notions. 

So what’s a task, if not what we think it is?

This is where ecological approaches begin to show promise. Our notion of task is not rooted in intuition but in an analysis of the ecological scaled physics of the context. What is there? What does it afford? What information does it create? All real things, not intuitive guesses

We call this analysis task dynamics  and it works gangbusters, at least for tasks we can handle in the lab such as coordination (e.g. Leach et al, 2020) or throwing (reviewed in Wilson et al, 2018). Task dynamical analyses are bloody hard, and get harder fast as the complexity of the context increases. But even work on simple lab tasks teach us lessons that coaches can start to apply. Ecological dynamics and the constraints-led approach are these applications. 

These analyses reveal that tasks don’t split into parts that fit intuitions. They also reveal the crucial importance of information in supporting learning & transfer (e.g. Snapp-Childs et al, 2015). Couple this to motor abundance and you stop talking about finding the correct way to move and start talking about  increasingly skilled perception-action engagement with contexts.

Can verbal instruction still be a tool? Perhaps, if treated as a way to implement constraints designed to guide attention to real aspects of actual tasks. But right now, that’s not what non-ecological coaches are doing. What they need is our task analysis to guide their use of language.

Lots of science still to do, as always (some already being done by Keith Davids). But that’s the issue as I see it right now; this is why ecological people think verbal instruction is problematic and overrated, and why we shift our attention to developing constrained but representative training contexts.