Mechanisms disappear when you look closely

My introduction to the economics of biology involved getting my butt kicked by abscisic acid. I was trying to learn about how trees enter and exit dormancy in anticipation of winter and spring, respectively. Science is one of those things that looks very solid from far away and is messy and riddled with holes up close, so at the time, knowing nothing about biology beyond “mitochondria is the powerhouse of the cell”, I figured there had to be a really good, clean explanation of exactly how trees enter and exit dormancy. It’s not exactly an exotic phenomenon, after all.

Things looked promising at first. If you google “what causes dormancy”, you get a nice, clear explanation: It’s abscisic acid. The production of abscisic acid leads to dormancy via clear mechanistic channels.

But when you look closer, the story becomes a lot more complicated. The “mechanisms” by which abscisic acid leads to dormancy seem to involve a tremendous amount of anticipatory regulating activity from many other genes, molecules, pathways, etc. When you look at dormancy from a distance, there’s a clear mechanistic cause of dormancy: the production of abscisic acid. When you look up close, you have a much more confusing, complex, and frankly organic situation.

Consider an analogous problem in economics. Workers usually do what their bosses tell them to. Imagine an alien studying humanity asks the question, “What is the mechanism that causes workers to obey their bosses?” They might be surprised to discover that there is no molecule that bosses release that forces workers to act out their wills. As we know, there is no mechanism for getting workers to do what their bosses tell them to. Instead, there is an incentive structure, a shaping of the energy landscape that workers navigate: do what I say and get money, or don’t do what I say and don’t get money.

What is the mechanism that causes people to pursue money? Again, a naive alien scientist might imagine that there’s something about the wavelength of green light that acts as a mechanism to cause people to chase after money, but nothing could be further from the truth. People want money because of how they expect to be able to use it to fulfill their plans—and to create bigger, better plans.

Talk of mechanisms treats causality as if it proceeds from one object to another. Press a button here, and something happens over there, so the button is the mechanism that caused the thing that happened. But economics know this isn’t how causality works—causality is inherently reciprocal, an idea that developmental psychologists independently generalized into multicausality. No part of a system can push on another part of the system without that latter part being there and able to be pushed on. This means that attributing causality to the pusher and not the pushee is an economic decision rather than a physical observation in the naive sense. (Observations being decisions themselves in real life.)

It’s like a light switch. A light switch is the mechanism that turns on a light in the sense that, given the rest of the system that produces the light, the switch is a part you can easily interact with to produce the outcome you want. But if you take out other parts of the system, the light switch that was previously “mechanistically” producing the light on/light off outcomes now seems to do nothing at all.

If economists studied a firm as if it was made of mechanisms, they would try to explain how a CEO’s directives lead to company-wide outcomes by carefully examining every single step in the process: this person talks to this person who sends this piece of paper to a third person—paper hypothesized to mechanistically induce particular behavior by some yet-unknown property of cellulose—which then leads to more, and more, and more, until they had enough details to fill a hundred books, and if asked, “But how did the CEO make that happen?” they still wouldn’t know.

It’s fine to ask for a mechanistic explanation of something, but what’s being asked for is a practical actionable perspective on a situation. Mechanistic explanations dissolve when you look closely at them, leaving behind a multicausal, polycomputable, relationally real system made of economic decisionmaking.