Consciousness May Be the Override Switch on a Crude Emotional Calculator

Books about consciousness spend a great deal of time on what it is and almost none on what it is for. When the second question is raised, the usual answer is that it is for nothing: a side effect of a complicated brain, the hum of the machinery. I do not find an immediate flaw in that theory. But I want to propose a more direct one, and see how far it goes.

The Emotional Calculator

Start with how decisions are actually made, most of the time.

We decide by feeling. Harari puts it well: emotions are a kind of measurement instrument. Every consideration for and against an action is fed in, each carrying a weight set by evolution and experience, and out comes a verdict, fight or flee, approach or avoid, trust or suspect. It arrives as a feeling, and the feeling is the answer.

This is a remarkable piece of engineering and it works well for the situations it was built in. But notice its properties. It is fast. It is roughly linear: more of this input pushes the verdict this way, more of that pushes it the other. And it is fixed by a process that knew nothing about the world you now live in.

The Spear

Here is the difficulty that I think consciousness was built to solve.

Suppose your ancestor meets a rival and the calculator runs: his size, his stance, your condition, the terrain. Out comes a verdict. Fine. Now suppose your ancestor has just invented the spear.

How much should the spear tilt the calculation toward fight? A lot? A little? Does it depend on whether the other man has one too, and on how good he is with it, and on whether the ground allows a throw? Evolution has no idea. The calculator was tuned over millions of years to a world without spears, and the spear changed the payoffs overnight. The feeling that arrives will be the pre-spear feeling, and it will be wrong.

The calculator cannot update itself fast enough to keep up with a creature that invents things. Something else is needed: a mechanism that understands that its own verdicts are being produced by an emotional algorithm, that knows the algorithm has not accounted for the spear, and that can override the verdict.

That is my candidate for what consciousness is for. Not to feel, but to sit above the feeling and, when the feeling is out of date, to overrule it. To resist the genes, so to speak, when the genes are wrong about the present.

Why the Override Is Partial

If that were the whole story, we would expect consciousness to have complete control, and it plainly does not. We know our feelings are wrong and we act on them anyway; that is the ordinary human condition. Why would evolution build an override and then not let it override?

Two reasons, I think, and both are good ones from evolution's point of view.

First, speed. Some situations require reaction before deliberation, and it is not obvious in advance which ones. A system that always waited for the override to weigh in would be eaten by the things the calculator was built to flee from. So the override has to be slow and the calculator has to be able to act without it, and the line between them cannot be clean.

Second, and more interesting: a creature with full rational control of itself has flown the perch. It might decide not to reproduce. It might decide not to play tit-for-tat, not to flee the lion, not to value what evolution needs it to value. From the genes' point of view, a fully rational organism is a catastrophe, because rationality serves the organism's goals and the genes' goals are not the same. So evolution built an override and then kept it on a leash, and the tension between the two is the tension everyone feels between what they know and what they want.

What Follows

If this is right, three things follow, and I find all of them uncomfortable.

The first is that logical thought is difficult for a reason. It is not the native mode of the system; it is the exception, the override, and overrides are costly by design. The experimental record on human rationality reads exactly as you would expect if reasoning were a late, effortful patch on an emotional engine.

The second is that our decisions have become badly detached from reality, and this is not a moral failing. The modern world is the spear multiplied by a million. Almost nothing in it resembles the environment the calculator was tuned for, so the feelings it produces are wrong almost everywhere, and the override cannot keep up.

The third is prescriptive, and I state it knowing it sounds cold. If consciousness is the override, then the right use of it is to override. To treat the feeling as a reading from an instrument known to be miscalibrated, and to correct for it deliberately. Not to suppress emotion, which is impossible and would remove the only fast system we have, but to stop mistaking it for a conclusion.

What I Am Not Claiming

I have argued elsewhere that pain is what an override needs in order to be safe, and that consciousness is probably not all-or-nothing. This post is narrower: it is a proposal about function, and it is an idea rather than a result. I have not seen it argued in this form and would like to hear it attacked.