Motor behavior as collective intelligence: Physics as autonomy of the parts

Motor behavior is an example of collective intelligence, where a system of parts interact to produce group-level solutions to group-level problems. The members of a collective intelligence are autonomous: they solve their own problems in their own problem spaces, and produce collective solutions as a result of how the autonomous members are connected via a cognitive glue. So the members of the system that produces motor behavior must be autonomous as well.

It’s relatively easy to understand how the members of a collective intelligence can be autonomous when the model examples are organisms and the economy. The members of these collectives are cells and people, respectively. Cells and people obviously have some amount of autonomy, or the ability to make decisions to pursue their own goals instead of just waiting inertly to be told what to do by something else. But the members of the collective that produces motor behavior are things like muscles, joints, body segments, and bones. Yes, these things are made of cells, but the cellular composition is unnecessary: human-like motor behavior can be produced with nonorganic materials.

So inorganic matter has to be able to behave autonomously. But all that it does is physics—so what’s the catch?

A possible solution is that physics is the autonomy of the parts. For the purposes of collective intelligence, it doesn’t matter whether the members of the collective make choices (at least not choices as typically understood). Instead, what really matters is whether signals shared throughout the collective function to warp the energy landscape of the parts so that when they do what we perceive as physics, the result is a coordinated system that produces group-level outcomes.

The simplest way of understanding what it means for inorganic, low-agency parts to be autonomous is that you can’t give them commands or instructions. If an apple falls from a tree, and you tell it to stop, it will keep falling anyway. To get the apple to do what you want, you have to interact with it in such a way that when the apple does whatever it’s going to do in the situation, the result is an outcome you want. For example, if you want the apple to fly into your hand and rest there, you can put your hand under the falling apple and grasp it. The apple is just doing its thing, but its thing is now your thing as well.

The tricking to getting a system of low-agency parts to produce collective intelligence is to connect them in such a way that when they interact with each other and the environment, they all push and pull on each other, and otherwise warp each other’s tendencies, so as to produce instances of the desired behavior, such as motor behavior. The spring-like, pendulum-like nature of the legs, for example, naturally produce swinging motions in many circumstances that drag each other forward as energy is stored and released. No commands have to be given; the system reliably produces the walking behavior just by how all the autonomous parts interact with each other and the environment.