Opinion – Suzana Herculano-Houzel: Muscles change, actions remain

With loaded arms and hands, pointing at an object of interest to get someone’s attention is foolhardy. But there’s no problem: in the absence of hands, without thinking twice we extend an elbow, a foot, or even our chin in the desired direction to show “there, oh”.

The abstract action is the same, designated with the same word –to point–, but the objective action is completely different, depending on the part of the body used, involving different muscles, limbs of different dimensions and weights. For a robot whose actions are coded one by one, the equivalence of movements is not trivial. Why, then, does it seem so simple to the brain?

​The advent of magnetic resonance imaging, which makes it possible to detect places in the brain whose activity changes as actions are planned and executed, showed years ago that there is a hierarchy of structures in the cortex, from those that implement movements punctually (where different neurons directly control different muscles, making them contract) to those that represent abstract actions, such as “reaching”, “taking”, “taking to the mouth”.

The key to the equivalence between “reaching with the hand” and “reaching with the foot” appears to lie in an intermediate region, whose neurons represent for the first time combinations of actions and their sensory consequences. Activating these neurons causes not a simple contraction of one muscle or another, but combinations that generate targeted movements – but always from the same part of the body, such as the right hand or left foot.

Abstract equivalence, which makes “pointing” a verb that, for the brain, dispenses with an object, appears later, in the frontal cortex, whose neurons represent in their activity patterns what “point with the hand” and “point with the foot” have in common: the part about extending a part of the body, point. The details on which part to use are up to those other regions who take care of the details, hey.

A transatlantic study involving teams in the US, Belgium and Italy recently showed that this scheme of transformation between specific and abstract actions also works in people born without arms or hands, and who have learned to use their feet for what others would call manipulating. the world. In the dysplastic brain, all actions at a distance are carried out by neurons that control legs and feet – but “point” or “grab” are orders represented by the same frontal regions as those born with hands.

It is unlikely that equivalence is already written in the genes; like so many other things in the brain, abstractions are probably built up bit by bit, according to the objective use of the parts. It’s an arrangement worthy of the attention of artificial intelligence people who expect everything to be born ready. Even robots need a chance to learn…