Nehaveigur

No Limit in Sight: Trainable skills beyond the distribution

We can train our brains to perform feats that seem supernatural. There are people who have learned to sense their surroundings using echolocation like bats. Others can memorize 10,000 random numbers. The feats professional athletes fall in the same category, even though they’re extremes on a continuum rather than one of a kind.

Those abilities get mistaken as a freak talent, acquired by a mutation like the superpowers of in X-Men. In reality, this is rarely the case, and genetics only plays a small role. Mathematician David Bessis makes this point in his book, Mathematica, but it equally applies to domains outside of maths.

If our brains are so plastic, and it’s possible to train those abilities, why can so few of us be bothered to do so? There are several explanations that aren’t mutually exclusive. The simplest one is that braining our brains is hard work. In many cases, it may be too hard for anyone but the most driven. The other explanation is the perception that there aren’t that many skills that can be acquired through training, and a mismatch between training methods and skills.

It may be helpful to systematically explore all the supernatural-seeming skills that humans have acquired, and then to map out the skills that could be acquired in principle but which haven’t been attempted yet. This is an idea I outlined here.

The other obstacle to the intentional acquisition of skills is that there is a lot of confusion about how to do this. Neuro-linguistic Programming (NLP) is an example: It’s a plausible-seeming technique to acquire skills, but it doesn’t work as promised. At the core of NLP, according to illusionist Derren Brown in his book Tricks of the Mind, involves understanding the mental patterns of those whose skills we wish to acquire. Practitioners believe that this is a shortcut to skill acquisition. It sounds plausible, but doesn’t work well.

The types of training that actually work seem to involve repetition and some element of immediate feedback. Not every skill is acquirable this way, and for many there may be a way to acquire them this way, but it hasn’t been devised yet. AI may make it possible to devise new ways of training skills that have previously not be amenable to repetition-and-feedback training.