What can collective intelligence theory do for the science of motor behavior?

In “Nature of Motor Control: Perspectives and Issues”, Turvey and Fonseca outline different approaches to motor behavior. I think there are a lot of important connections between motor behavior and collective intelligence. But beyond simple comparisons, collective intelligence theory can resolve some of the tricky conceptual issues in understanding motor behavior.

Turvey and Fonseca outline four perspectives on how motor behavior works. The first is neural control: the brain commands the body to do a particular thing, and the body does it. The second is a puppeteer controlling a puppet on strings: the puppeteer is controlling the puppet, but this takes work because the puppet has physics that have to be accounted for; you can’t simply command outcomes. The third perspective adds in the environment as a factor. The fourth perspective eliminates any concept of an executive: motor control does not reside anywhere in particular. There is no puppeteer, however disguised or distributed.

This last perspective is very conducive to collective intelligence. Instead of thinking of intelligence as being situated somewhere in particular, we think of intelligence as emerging from the coordinated interactions of many parts. “Intelligence” is a label humans put on the results; it is a decision about how to categorize observations. Intelligence isn’t localized to anywhere in particular for the same reason that emotion isn’t localized to anywhere in particular.

In motor behavior, this last perspective has been associated with dynamic systems theory and related thermodynamic ideas. There are two challenges with these approaches to understanding motor behavior. First, they attribute the patterns created by the motor systems to the properties of the underlying mathematical dynamics. In other words, robust and reliable patterns emerge because “that’s how dynamic systems are”, which isn’t satisfying. Why are dynamic systems this way? We don’t know. Relatedly, a second problem is how the dynamic plan is actually constructed and maintained.

The planning-execution conceptual divide is difficult to maintain in the perspective of Figures 1c and 1d. The singular dynamical, self-organizing language required to capture the time-evolution of neural, body, and environmental states incorporates preparing and doing. A central and challenging issue for the theory implied by Figure 3 is how to craft the dynamics of planning so as to express its continuous development and seamless transformation into the dynamics of execution.

I.e., you somehow need a system where the planning of the motor behavior flows into the execution of the motor behavior without clear divide or distinction, and which somehow continuously accounts for changes everywhere within and without the body, to consistently achieve patterns that are written nowhere. How can this happen?

Collective intelligence is the answer. The key is that collective intelligence exploits the knowledge and planning abilities of the members of the collective. Planning and execution aren’t distinct because the plans are the predictions produced by the coordinating model constructed by the members of the collective. Because each member of the collective is pursuing its own self-interest, it necessarily brings to bear its own personal knowledge and competencies to the task. The result is a system that continuously updates the plan to maintain a pattern based on lots of decentralized information because there are many members of the collective each continuously updating its own plan to maintain a pattern based on the information they personally have access to.

Therefore, I think that collective intelligence theory can usefully be thought of as an extension of dynamic systems theory, expanding on concepts about agential problem-solving parts coordinating in an economic fashion to explain the observations made by dynamic systems theory in a more satisfactory manner.