Free will, or agency, is an extremely confusing, self-contradictory, seemingly incoherent, concept — which is why in the past I simply denied that it existed at all or that it was even a meaningful idea. That was a cop-out, though. It very obviously is a meaningful idea, and even a philosophically necessary one. Everyone acts as if they had free will — which implies that, on an intuitive level, we all understand what that means — which, in turn, implies that there is something to understand.
Granted, no one actually does understand free will in an explicit way, but that’s not such an unusual situation. No one fully and explicitly understands all the syntactic rules of his own native language, either, but the fact that we effortlessly follow those rules every day is proof that they exist and are, in principle, intelligible.
So I’ve decided to tackle the problem of free will instead of evading it. The first step is to admit the truth — there is such a thing as free will, and we don’t have a clue what it is — and to resist the temptation to deny either part of that statement, either to pretend free will is nonsense or to pretend that any of the lame “explanations” of it is satisfying.
As a name for the problem to be solved, I currently find the term agency to be more helpful than free will. This is partly due to my Mormon upbringing, and partly to the more recent influences of Goethe — whose declaration “In the beginning was the Act” seems deeper and deeper the more I think about it — and of Gurdjieff.
Fundamental to Gurdjieff’s philosophy is the distinction between things that people do and things that just happen. Happening is mechanical; doing — that is, action — is conscious. When a cloud drops rain or an arrow flies through the air, neither the cloud nor the arrow is doing anything. Raining happens; flying happens. The role of the cloud or the arrow is entirely passive; each simply receives pre-existing causal influences and passes them on in a mechanical, predetermined manner. In the strictest sense, neither the cloud nor the arrow can even be said to exist as an individual thing; each is simply part of that amorphous “it” to which we ascribe mere happenings (as when we say “it is raining”). Gurdjieff believed that most of what most people “do” most of the time is also mere happening. We go through our lives on autopilot, mechanically responding to stimuli and running the preprogrammed subroutines of habit. Most people most of the time have no real consciousness and no real will; they do not even exist as individuals. Rather, “it wants or it does not want,” and things happen accordingly.
To a pure determinist, that’s the whole story. But Gurdjieff also believed that true will — doing — agency — was possible, as it indeed it must be. A universe of pure happening — in which everything is caused, everything is predetermined, but nothing is caused or determined by anything — is unintelligible. At the very least, there must have been an agent to set the ball rolling — the First Mover of Aristotle. But there is no reason to assume only one agent, and common sense and our immediate subjective experience tell us that humans, too, are or can be agents. Gurdjieff may have been right that most of the time “it happens” — but sometimes we act.
One key feature of agency is decision. When an agent acts, he decides something that was undecided before. Prior to his action, two or more different courses of events were possible, and his action consists in the realization of one of them. Predicting which of them would be realized, using only information about the world prior to the act, would have been impossible even in principle. (An action which can be predicted from past information is not a true action because it does not decide anything; everything has already been decided, and all that remains is for it to happen.)
The other key feature of agency is the agent. Things are not just decided; they are decided by somebody. And this is where the difficulty — the seeming antinomy which once led me to reject the whole idea of agency — comes in. The decision must “come from” an agent who already existed prior to the decision, and it’s hard to know what “come from” can mean here if it does not refer to a relationship of cause-and-effect. But if a decision is determined by some previous state of affairs, then it is not really a decision. Everything was already decided before the so-called decision was made.
This, then, is the fundamental antinomy. The idea of an agent doing something seems to entail cause-and-effect, meaning that the state of affairs at any given time can be derived from the state of affairs at a previous time. However, the idea of decision seems to entail the opposite: unpredictability-in-principle — the idea that a thing becomes determined at a particular point in time, and that prior to that point it was not determined. Yet agency necessarily includes both of these seemingly incompatible elements. Agency requires that decisions can be made by agents — which means that one and the same event is both caused an uncaused, derived from a past state of affairs and yet not predictable from that past state of affairs. Unless both of these contradictories can be incorporated, we don’t have agency; we just have things happening — happening either mechanically (if no decisions are made) or randomly (if decisions are made, but not by anyone), but either way, still just happening — not being done.
It appears that what we need is a non-deterministic form of causality — some way to say that P caused Q, but that P could just as easily have caused not-Q instead. This is complete nonsense logically — it means that “P, therefore Q” and “P, therefore not-Q” are equally correct. Is there any way to make such an idea intelligible without destroying the meaning of “therefore”?