[Excerpt] Commitment Problems

Excerpt From: Frank (1988). Passions within Reason.
See Also: Breakdown of Will, Intro to Prisoner’s Dilemma 
Content Summary: 1600 words, 8 min read.

The Kidnapper’s Dilemma

Thomas Schelling described a kidnapper who suddenly gets cold feet. He wants to set his victim free, but is afraid he will go to the police. In return for his freedom, the victim gladly promises not to do so. The problem, however, is that both realize it will no longer be in the victim’s interest to keep this promise once he is free. And so the kidnapper reluctantly concludes that he must kill him.

Schelling suggests the following way out of the dilemma: “If the victim has committed an act whose disclosure could lead to blackmail, he may confess it; if not, he might commit one in the presence of his captor, to create a bond that will ensure his silence.” (Perhaps the victim could allow the kidnapper to photograph him in the process of some unspeakably degrading act.) The blackmailable act serves here as a commitment device, something that provides the victim with an incentive to keep his promise. Keeping it will still be unpleasant for him once he is freed; but clearly less so than not being able to make a credible promise in the first place.

Commitment Problems

In everyday economic and social interaction, we repeatedly encounter commitment problems like the one confronting Schelling’s kidnapper and victim. Being known to experience certain emotions enables us to make commitments that would otherwise not be credible. The clear irony here is that this ability, which springs from a failure to pursue self-interest, confers genuine advantages. 

Granted, following through on these commitments will always involve avoidable losses not cheating when there is a chance to, retaliating at great cost even after the damage is done, and so on. The problem, however, is that being unable to make credible commitments will often be even more costly. Confronted with the commitment problem, an opportunistic person fares poorly.

Everyday Examples

Deterrence. Jones has a $200 leather briefcase that Smith cabinets. If Smith steals it, Jones must decide whether to press charges. If he does, he will have to go to court. He will get his briefcase back and Smith will spend 60 days in jail, But the day in court will cost Jones $300 in lost earnings. Since this is more than the briefcase is worth, it is clearly not in his material interest to press charges. (To eliminate an obvious complication, suppose Jones is about to move to a distant city, so deterrence is not a relevant consideration). Thus, if Smith knows that Jones is a rational, self interested person, he is free to steal the briefcase with impunity. Jones may threaten to press charges, but his threat will be empty. 

Now suppose that Jones is not a pure rationalist; that if Smith steals the briefcase, Jones will become enraged, and think nothing of a day’s lost earnings, or even a week, in order to see justice done. If Smith knows that Jones will be driven by emotion, not reason, he will let the briefcase be. People expect us to behave rationally in response to theft of property, we will sell them need to behave irrationally in practice, because it is not in their interest to steal it. And predisposed to respond irrationally serves much better here than being guided only by material self interest.

Cheating. Two persons, Smith and Jones, can engage in a potentially profitable venture, say, a restaurant. Their potential for gain arises from the natural advantages inherent in the division and specialization of labor. Smith is a talented cook, but is shy and an incompetent manager. Jones, by contrast, cannot boil an egg, but is charming and has shrewd business judgment. Together, they have the necessary skills to launch a successful venture. Working alone, however, their potential is much more limited. Their problem is this: Each will have opportunities to cheat without possibility of detection. Jones can skim from the cash drawer without Smith’s knowledge. Smith, for his part, can take kickbacks from food suppliers. If only one of them cheats, he does very well. The non-cheater does poorly, but isn’t sure why. His low return is not à reliable sign of having been cheated, since there are many benign explanations why a business might do poorly. If the victim also cheats, he, too, can escape detection, and will do better than by not cheating; but still not nearly so well as if both had been honest. Once the venture is under way, self-interest unambiguously dictates cheating. Yet if both could make a binding commitment not to cheat, they would profit by doing so. 

Bargaining. In this example, Smith and Jones again face the opportunity of a profitable joint venture. There is some task that they alone can do, which will net them $1000 total. Jones has no pressing need for extra money, but Smith has important bills to pay. It is a fundamental principle of bargaining theory that the party who needs the transaction least is in the strongest position. The difference in their circumstances thus gives Jones the advantage. Needing the gain less, he can threaten, credibly, to walk away from the transaction unless he gets the lion’s share of the take, say $800. Rather than see the transaction fall through, it will then be in Smith’s interest to capitulate. Smith could have protected his position, however, had he been able to make a binding commitment not to accept less than, say, half of the earnings. One possible way of accomplishing this would be to sign a contract that requires him to contribute $500 to the Republican party in the event he accepts less than $500 from his joint venture with Jones. (Smith is a lifelong Democrat and finds the prospect of such a gift distasteful.) With this contract in place, it would no longer be in his interest to give in to Jones’s threat. (If Smith accepted $200, for example, he would have to make the $500 contribution, which would leave him $300 worse off than if he hadn’t done the job with Jones at all.) Jones’s threat is suddenly stripped of all its power.

Marriage. As a final example of the commitment problem, consider the difficulty confronting a couple who want to marry and raise a family. Each considers the other a suitable mate. But marriage requires substantial investment, which each person fears could be undercut if the other were to leave for an even more attractive opportunity in the future. Without reasonable assurance that this will not happen, neither is willing to make the investments required to make the most of their marriage. They could solve their problem if they could write a detailed marriage contract that would levy substantial penalties on whichever of them attempted to leave. They are, after all, willing to forego potentially attractive opportunities in the future in order to make it in their interests to invest in the present effort of raising a family. It would serve their purposes to take steps now that would alter the incentives they face in the future.

Life, it seems, is rife with commitment problems. 

Emotions as Commitment Devices

My claim is that specific emotions act as commitment devices that help resolve these dilemmas. 

  • Retaliation and Deterrence. Consider a person who threatens to retaliate against anyone who harms him. For his threat to deter, others must believe he will carry it out. But if others know that the costs of retaliation are prohibitive, they will realize the threat is empty. Unless, of course, they believe they are dealing with someone who simply likes to retaliate. Such a person may strike back even when it is not in his material interests to do so. But if he is known in advance to have that preference, he is not likely to be tested by aggression in the first place. 
  • Proportionality vs Bargaining. Similarly, a person who is known to “dislike” an unfair bargain can credibly threaten to walk away from one, even when it is in her narrow interest to accept it. By virtue of being known to have this preference she becomes a more effective negotiator.
  • Guilt vs Cheating. Consider, too, the person who “feels bad” when he cheats. These feelings can accomplish for him what a rational assessment of self-interest cannot–namely, they can cause him to behave honestly even when he knows he could get away with cheating. And if others realize he feels this way, they will seek him as a partner in ventures that require trust.
  • Love and Marriage. It is no surprise that the marriage problem is better solved by moral sentiments than by awkward formal contracts. The best insurance against a change in future material incentives is a strong bond of love. If ten years from now one partner falls victim to a lasting illness, the other’s material incentives will be to find a new partner. But a deep attachment will render this change in incentives irrelevant. (Indeed, research has shown the “active ingredient” of romantic and non-romantic attachment is derogation of alternatives). 

Signaling Your Commitment

Emotions qua commitment devices would not have evolved unless a person can reliably signal possessing them. Consider:

One fall day, almost twenty years ago, black activist Ron Dellums was speaking at a large rally. But at least one young man was not moved by Dellums’s speech. He sat still as a stone on the steps of Sproul Plaza, lost to some drug, his face and eyes empty of expression. Presently a large Irish setter appeared, sniffing his way through the crowd. He moved directly to the young man sitting on the steps and circled him once. He paused, lifted his leg, and, with no apparent malice, soaked the young man’s back. He then set off again into the crowd. The boy barely stirred.

Now, the Irish setter is not a particularly intelligent breed. Yet this one had no difficulty locating the one person in that crowd who would not retaliate for being sprayed. Facial expressions and other aspects of demeanor apparently provide clues to behavior that even dogs can interpret. And although none of us had ever witnessed such a scene before, no one was really surprised when the boy did nothing. Before anything even happened, it was somehow obvious that he was just going to go right on sitting there.

Without doubt, however, the boy’s behavior was unusual. Most of us would have responded angrily, some even violently. Yet we already know that no real advantage inherent in this “normal” response. After all, once the boy’s shirt was soaked, it was already too late to undo the damage. And since he was unlikely ever to encounter that particular dog again, there was little point in trying to teach the dog a lesson. On the contrary, any attempt to do so would have courted the risk of being bitten.

Our young man’s problem was not that he failed to respond angrily, but that he failed to communicate to the dog that he was so predisposed. The vacant expression on his face was somehow all the dog needed to know he was a safe target. Merely by wearing “normal” expressions, the rest of us were spared.

There are numerous behavioral clues to people’s feelings. Posture, the rate of respiration, the pitch and timbre of the voice, perspiration, facial muscle tone and expression, and movement of the eyes, are among the signals we can read. We quickly surmise, for example, that someone with clenched jaws and a purple face is enraged, even when we do not know what, exactly, may have triggered his anger. And we apparently know, even if we cannot articulate, how a forced smile differs from one that is heartfelt. At least partly on the basis of such clues, we form judgments about the emotional makeup of the people with whom we deal. Some people we sense we can trust, but of others we remain forever wary.


  1. A commitment problem is one where maximizing immediate material welfare is maladaptive in the long run.
  2. A commitment device hijacks one’s own reward mechanism towards irrational behavior (insensitive to one’s material welfare).
  3. Many emotions (including guilt, rage, love) and intuitions (including fairness) may be best seen as commitment devices.
  4. A commitment device must be advertised to work effectively: the organism must emit (hard to fake!) signals of their presence.

Until next time.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s