Weakness of will

Why do people knowingly do the very thing they have already judged to be worse for them?

You decided last night, calmly and with all the facts, that you would not have the second drink. Tonight you have it. Nothing was learned in between; no new argument arrived; you would still, if asked in the abstract, give the same answer you gave last night. And yet the hand moves.

Notice this is a different puzzle from asking whether your choices are yours at all. Grant that the choice was fully yours — nobody pushed you, you were not compelled. The oddity is internal: how can a person act, deliberately and knowingly, against a judgement they still hold?

Socrates' answer in Plato's Protagoras is to deny the phenomenon. Nobody, he argues, goes willingly toward what they believe is worse. What actually happens is a distortion of size: as with objects, the nearer pleasure looks bigger than the distant one, so the person is not defying their own judgement but miscalculating. What we need, he concludes, is an art of measurement — a technique for weighing what is near against what is far. Weakness of will is ignorance under another name.

That has a satisfying tidiness, and it fails a simple test. Give the person the measurement — the numbers, the outcome, their own agreement out loud that the second drink is worse. Then watch. If akrasia were ignorance, information would be its cure, and it very often is not. Aristotle takes the phenomenon seriously in Nicomachean Ethics VII and locates the failure more precisely: the akratic possesses the knowledge but is not using it, applying the general principle to the particular in front of him in the way a drunk recites a poem he does not understand.

So the knowledge is there but inert. What makes it inert? Here the modern account earns its place. Suppose the value you place on a reward shrinks with its distance from you — uncontroversial so far. Everything turns on the shape of the shrinkage. If it shrinks by a constant proportion per unit of delay, then the ordering of two options is preserved no matter where you stand: whichever you prefer at a distance, you prefer up close. If instead it shrinks steeply at first and then flattens — roughly hyperbolically, as George Ainslie argued — then the ordering is not preserved. A small reward that is nearly here can overtake a large reward that is further off, purely because it approached.

Sit with what that means, because it is the crux. Nothing has to change in your beliefs for your preference to reverse. You need learn nothing, forget nothing, deceive yourself about nothing. Only the distance changed. Socrates was half right — there is a distortion of magnitude, exactly as he said — and half wrong, because it is not the kind of error information corrects.

It also explains the shape of the actual experience: the calm decision at a distance, the reversal at close range, the retrospective agreement with your earlier self once the small reward has receded again. That is not the pattern of ignorance. It is the pattern of a crossing.

Now the strongest objection, and it is aimed at the discounting story rather than at Socrates. If your preference genuinely reverses in the moment, then in the moment you judge the drink to be better — and someone who does what they judge best is not acting against their judgement at all. On that reading, hyperbolic discounting does not explain akrasia. It abolishes it, and hands the puzzle back to Socrates in modern dress.

That objection has force, and there are two serious replies. Donald Davidson, in "How Is Weakness of the Will Possible?" (late 1960s, date recalled), argued that the akratic really does act against an all-things-considered judgement, because the judgement that weighs all the reasons and the unconditional judgement that issues in action are distinct, and nothing guarantees the second follows the first. What the akratic violates, on his account, is not a belief but a principle of continence — the requirement to let your all-out judgement track your all-things-considered one.

Richard Holton makes a different move. He separates two things usually run together: acting against your best judgement, and abandoning a resolution too readily (Willing, Wanting, Waiting, 2009, recalled). Most everyday failures, he suggests, are the second. You do not revise your view of the second drink; you fail to hold a prior resolution against pressure, and holding it is a distinct capacity — effortful, exercised rather than deduced. That makes willpower an ability rather than a belief-state, which is why practice and rules help where argument does not.

Two honesty notes. First, the popular corollary that willpower is a depletable resource rests largely on the ego-depletion literature, which has replicated poorly at scale; treat the effortfulness as observed and the fuel-tank picture as unsettled. Second, the functional form of discounting is itself contested — hyperbolic and quasi-hyperbolic models compete, and the fit varies by domain — so the reversal is well evidenced while its exact curve is not.

Whether clear-eyed akrasia, in Davidson's strict sense, ever actually occurs is genuinely open. What turns on it is practical: if most failures are reversals and resolution-collapses rather than defiance of a live judgement, then the useful intervention is not better arguments in the moment but precommitment and bright-line rules — arrangements that remove the choice from the point where the curves cross.

• Whether the same crossing explains procrastination, or whether avoidance of an aversive task works differently from pursuit of a nearer reward.
• How a resolution is actually maintained, given that reconsidering it is sometimes exactly the rational thing to do.

Every term the collection defines is gathered in the glossary.