In a discussion about my article, Tom Tabaczynski raised an interesting question: Do we think about what we are doing when we drive a car? Do we consciously think about grammatical rules when we speak? Why, then, should we assume that dancing is based on conscious decisions made according to rules previously learned in lessons?

At first, this is a good objection. An experienced driver normally no longer thinks about which foot has to move where when starting the car. When speaking, we likewise do not construct our sentences by first retrieving grammatical rules consciously. And an experienced dancer should not spend the entire dance thinking: now I shift my weight, then I rotate my pelvis by a certain angle, and after that I take the next step.

Complex skills that have been learned well are performed to a considerable extent automatically. That is an important part of learning.

The argument becomes problematic, however, when this correct observation is used to derive a second claim: because a skill is later performed largely without conscious control, it must therefore also have been learned only in a non-conscious way. This is precisely the step Tabaczynski takes when he writes that dancing on a crowded floor is “the only way to learn,” because this cannot be learned consciously in a lesson.

Two different questions are being merged here: How is a developed skill performed? And how is that skill acquired?

The fact that I no longer think about operating the clutch while driving does not mean that I was never consciously taught how a clutch works. The fact that I do not think about every grammatical rule while writing does not mean that explanations, corrections, or conscious engagement with language could not have contributed to my language development. And the fact that a dancer eventually stops thinking about the organization of every movement does not prove that instructions, tasks, or teaching could not have contributed to its development.

Automatization can be an outcome of learning. It is not a description of the only possible way to learn.

Motor-learning research, in particular, does not support a simple opposition between “consciously learned” and “unconsciously learned.” Motor learning can involve both implicit and explicit processes. Information that was initially acquired consciously can become automatized with increasing experience. At the same time, there are good reasons to be critical of excessive conscious control of a movement that has already become automatic. Someone who suddenly tries to consciously control every component of a fluent movement may indeed make that movement worse.

But that still does not mean that instruction or teaching is inherently harmful. It means, rather, that teaching should not be confused with permanent conscious control of movement.

A teacher can, for example, define a task without prescribing its solution. The teacher can direct attention towards a particular effect, restrict the available space, create different situations, or alter the conditions under which learners have to find solutions themselves. In contemporary approaches to motor learning, precisely this deliberate design of the learning environment is regarded as an important part of coaching and teaching.

This brings us to the crowded milonga. I find it entirely plausible that a densely populated dance floor challenges and develops certain abilities. A dancer has to perceive available space, regulate speed, interrupt movements when necessary, and continually adapt to the movements of other couples. A crowded floor can therefore provide an extremely rich learning environment.

But “this context is very well suited to learning” does not logically imply “this context is the only way to learn.”

At least some of the relevant conditions can also be created deliberately in a teaching situation. Several couples can dance in a restricted space. Unexpected stops or changes of direction can be introduced. The available space can be altered. Tasks can be given in which no particular sequence of steps is prescribed. The learners still have to generate their own solutions. The only difference is that the conditions have been deliberately designed.

That is still teaching.

There is another interesting aspect of Tabaczynski’s own learning history. He himself describes having trained in Contact Improvisation while he was learning tango and says that this experience helped him move beyond the movement patterns he had acquired in tango classes. He may therefore already have possessed abilities that an ordinary tango beginner would first have to develop: perceiving movement through contact, responding continuously to a partner, maintaining a flowing movement process, and reorganizing one’s own movement in response to a changing shared situation.

What eventually appears self-evident and “natural” to an experienced Contact dancer may therefore be anything but self-evident to a beginner. A skill can become so deeply automatized that we even forget that at some point it had to be learned.

None of this makes the idea of implicit learning wrong. On the contrary, it is extremely relevant to tango. It speaks against teaching that overloads dancers with rules and attempts to make every movement consciously controllable. It supports exploration, variation, perception, rich context, and tasks whose solutions are not predetermined.

What it does not establish, however, is the claim that teaching cannot fundamentally contribute to the acquisition of such abilities.

Perhaps the decisive distinction is therefore not between teaching and non-teaching, but between different forms of teaching. Teaching can prescribe movements and demand their reproduction. But it can also create conditions in which learners discover their own solutions.

A task can be prescribed without prescribing its solution.