Let’s begin by trying to make sense of the very idea of self-control and the circumstances in which it is and isn’t rational.75 First we must set aside pure selfishness—doing something that helps oneself but hurts others—and focus on self-indulgence—doing something that helps oneself in the short term but hurts oneself in the long term. Examples abound. Food today, fat tomorrow. Nicotine today, cancer tomorrow. Dance today, pay the piper tomorrow. Sex today, pregnancy, disease, or jealousy tomorrow. Lash out today, live with the damage tomorrow.
There is nothing inherently irrational about preferring pleasure now to pleasure later. After all, the You on Tuesday is no less worthy of a chocolate bar than the You on Wednesday. On the contrary, the You on Tuesday is more worthy. If the chocolate bar is big enough, it might tide you over, so eating it on Tuesday means that neither You is hungry, whereas saving it for Wednesday consigns you to hunger on Tuesday. Also, if you abstain from chocolate on Tuesday, you might die before you wake, in which case neither the Tuesday You nor the Wednesday You gets to enjoy it. Finally, if you put the chocolate away, it might spoil or be stolen, again depriving both Yous of the pleasure.
All things being equal, it pays to enjoy things now. That is why, when we lend out money, we insist on interest. A dollar tomorrow really is worth less than a dollar today (even if we assume there is no inflation), and interest is the price we put on the difference. Interest is charged at a fixed rate per unit of time, which means that it compounds, or increases exponentially. That compensates you exactly for the decreasing value of the money coming back to you as time elapses, because the decrease in value is also exponential. Why exponential? With every passing day, there is a fixed chance you will die, or that the borrower will abscond or go bankrupt and you’ll never see the money again. As the probability that this will not have happened dwindles day by day, the compensation you demand multiplies accordingly. Going back to pleasure, a rational agent, when deciding between indulging today and indulging tomorrow, should indulge tomorrow only if the pleasure would be exponentially greater. In other words, a rational agent ought to discount the future and enjoy some pleasure today at the expense of less pleasure tomorrow. It makes no sense to scrimp all your life so that you can have one hell of a ninetieth birthday bash.
Self-indulgence becomes irrational only when we discount the future too steeply—when we devalue our future selves way below what they should be worth given the chance that those selves will still be around to enjoy what we’ve saved for them. There is an optimum rate of discounting the future—mathematically, an optimum interest rate—which depends on how long you expect to live, how likely you will get back what you saved, how long you can stretch out the value of a resource, and how much you would enjoy it at different points in your life (for example, when you’re vigorous or frail). “Eat, drink, and be merry, for tomorrow we die” is a completely rational allocation if we are sure we are going to die tomorrow. What is not rational is to eat and drink as if there’s no tomorrow when there really is a tomorrow. To be overly self-indulgent, to lack self-control, is to devalue our future selves too much, or equivalently, to demand too high an interest rate before we deprive our current selves for the benefit of our future selves. No plausible interest rate would make the pleasure in smoking for a twenty-year-old self outweigh the pain of cancer for her fifty-year-old self.
Much of what looks like a lack of self-control in the modern world may consist of using a discounting rate that was wired into our nervous systems in the iffy world of our pre-state ancestors, when people died much younger and had no institutions that could parlay savings now into returns years later.76 Economists have noted that when people are left to their own devices, they save far too little for their retirement, as if they expect to die in a few years.77 That is the basis for the “libertarian paternalism” of Richard Thaler, Cass Sunstein, and other behavioral economists, in which the government would, with people’s consent, tilt the playing field between their current and future selves.78 One example is setting an optimal retirement savings plan as the default, which employees would have to opt out of, rather than as a selection they would have to opt into. Another is to shift the burden of sales taxes onto the least healthy foods.
But weakness of the will is not just a matter of discounting the future too steeply. If we simply devalued our future selves too much, we might make bad choices, but the choices would not change as time passed and the alternatives drew near. If the inner voice shouting “dessert sooner” outvoted the one whispering “fat later,” it would do so whether the dessert was available for consumption in five minutes or in five hours. In reality the preference flips with imminence in time, a phenomenon called myopic discounting.79 When we fill in the room service card at night and hang it on the hotel doorknob for the following morning’s breakfast, we are apt to tick off the fruit plate with nonfat yogurt. If instead we make our choices at the buffet table, we might go for the bacon and croissants. Many experiments on many species have shown that when two rewards are far away, organisms will sensibly pick a large reward that comes later over a small reward that comes sooner. If, for example, you had a choice between ten dollars in a week and eleven dollars in a week and a day, you’d pick the second. But when the nearer of the two rewards is imminent, self-control fails, the preference flips, and we go for smaller-sooner over larger-later: ten dollars today over eleven dollars tomorrow. Unlike merely discounting the future, which makes sense if the discount rate is properly set, myopic discounting, with its reversal of preferences, is not in any obvious way rational. Yet all organisms are myopic.
Mathematically minded economists and psychologists explain the myopic preference reversal by saying that organisms engage in hyperbolic discounting rather than the more rational exponential discounting.80 When we depreciate our future selves, instead of repeatedly multiplying the subjective value of a reward by a constant fraction with every unit of time we have to wait for it (rendering it half as valuable, then a quarter, then an eighth, then a sixteenth, and so on), we multiply the original subjective value by a smaller and smaller fraction (which renders it half as valuable, then a third, then a quarter, then a fifth, and so on). This insight can also be expressed in a more intuitive, qualitative way. A hyperbola is a mathematical curve with a bit of an elbow, where a steep slope looks like it has been welded onto a shallow one (unlike an exponential curve, which is a smoother ski jump). That jibes with a psychological theory that myopic discounting arises from a handoff between two systems inside the skull, one for rewards that are imminent, another for rewards that are far in the future or entirely hypothetical.81 As Thomas Schelling put it, “People behave sometimes as if they had two selves, one who wants clean lungs and long life and another who adores tobacco, or one who wants a lean body and another who wants dessert, or one who yearns to improve himself by reading Adam Smith on self-command . . . and another who would rather watch an old movie on television.” 82 Freud’s theory of the id and the ego, and the older idea that our lapses are the handiwork of inner demons (“The devil made me do it!”) are other expressions of the intuition that self-control is a battle of homunculi in the head. The psychologist Walter Mischel, who conducted classic studies of myopic discounting in children (the kids are given the agonizing choice between one marshmallow now and two marshmallows in fifteen minutes), proposed, with the psychologist Janet Metcalfe, that the desire for instant gratification comes from a “hot system” in the brain, whereas the patience to wait comes from a “cool system.” 83
In previous sections we have caught glimpses of what the hot and cool systems might be: the limbic system (whose major parts are exposed in figure 8–2) and the frontal lobes (seen in figure 8–3). The limbic system includes the Rage, Fear, and Dominance circuits that run from the midbrain through the hypothalamus to the amygdala, together with the dopamine-driven Seeking circuit that runs from the midbrain through the hypothalamus to the striatum. Both have two-way connections to the orbital cortex and other parts of t
he frontal lobes, which, as we saw, can modulate the activity of these emotional circuits, and which can come between them and the control of behavior. Can we explain self-control as a tug of war between the limbic system and the frontal lobes?
In 2004 the economists David Laibson and George Loewenstein teamed up with the psychologist Samuel McClure and the neuroimager Jonathan Cohen to see if the paradox of myopic discounting could be explained as a give-andtake between two brain systems: as they put it, a limbic grasshopper and a frontal lobe ant.84 Participants lay in a scanner and chose between a small reward, such as five dollars, which would be available in the nearish future, and a larger reward, such as forty dollars, which would only be available several weeks later. The question was, did the brain treat the choice differently depending on whether it was “$5 now versus $40 in two weeks” or “$5 in two weeks versus $40 in six weeks”? The answer was that it did. Choices that dangled the possibility of immediate gratification in front of a participant lit up the striatum and the medial orbital cortex. All the choices lit up the dorsolateral prefrontal cortex, the part of the frontal lobes involved in cooler, more cognitive calculations. Even better, the neuroimagers could literally read the participants’ minds. When their lateral prefrontal cortex was more active than their limbic regions, the participants held off for the larger-later reward; when the limbic regions were as active or more active, they succumbed to the smaller-sooner one.
As the front-heavy brain in figure 8–3 reveals, the frontal lobes are massive structures with many parts, and they carry out self-control of several kinds.85 The rearmost margin, which abuts the parietal lobe, is called the motor strip, and it controls the muscles. Just in front of it are premotor areas that organize motor commands into more complex programs; these are the regions in which mirror neurons were first discovered. The portion in front of them is called the prefrontal cortex, and it includes the dorsolateral and orbital/ventromedial regions we have already encountered many times, together with the frontal pole at the tip of each hemisphere. The frontal pole is sometimes called “the frontal lobe of the frontal lobe,” and together with the dorsolateral prefrontal cortex, it is active when people choose a large late reward over a smaller imminent one.86
Traditional neurologists (the doctors who treat patients with brain damage rather than sliding undergraduates into scanners) were not surprised by the discovery that it is the frontal lobes that are most involved in self-control. Many unfortunate patients end up in their clinics because they discounted the future too steeply and drove without a seat belt or bicycled without a helmet. For the small immediate reward of getting on the road a second sooner or feeling the wind in their hair, they gave up a large later reward of walking away from an accident with their frontal lobes intact. It’s a bad bargain. Patients with frontal lobe damage are said to be stimulus-driven. Put a comb in front of them, and they will immediately pick it up and comb their hair. Put food in front of them, and they will put it in their mouths. Let them go into the shower, and they won’t come out until they are called. Intact frontal lobes are necessary to liberate behavior from stimulus control—to bring people’s actions into the service of their goals and plans.
In a collision with a hard surface, the frontal lobe bangs against the front of the skull and is damaged indiscriminately. Phineas Gage’s freak accident, which sent a spike cleanly up through his orbital and ventromedial cortex and largely spared the lateral and frontmost parts, tells us that different parts of the frontal lobes implement different kinds of self-control. Gage, recall, was said to have lost the equilibrium “between his intellectual faculties and animal propensities.” Neuroscientists today agree that the orbital cortex is a major interface between emotion and behavior. Patients with orbital damage, recall, are impulsive, irresponsible, distractible, socially inappropriate, and sometimes violent. The neuroscientist Antonio Damasio attributes this syndrome to the patients’ insensitivity to emotional signals. He has shown that when they gamble on cards with different odds of gaining and losing money, they don’t show the cold sweat that normal people experience when they bet on a card with ruinous odds.87 This emotionally driven self-control—what we might call apprehension—is evolutionarily ancient, as shown by the well-developed orbital cortex in mammals such as rats (see figure 8–1).
But there are also cooler, more rule-driven forms of self-control, and they are implemented in the outer and frontmost parts of the frontal lobe, which are among the parts of the brain that expanded the most in the course of human evolution.88 We have already seen that the dorsolateral cortex is engaged in rational calculation of costs and benefits, as in the choice between two delayed rewards, or in the choice between diverting a runaway trolley onto a side track with a single worker or letting it proceed to the main track with five.89 The frontal pole sits even higher in the chain of command, and neuroscientists credit it with our suppleness in negotiating the competing demands of life.90 It is engaged when we multitask, when we explore a new problem, when we recover from an interruption, and when we switch between daydreaming and focusing on the world around us. It is what allows us to branch to a mental subroutine and then pop back up to the main thing we were trying to accomplish, as when we interrupt ourselves while cooking to run to the store to get a missing ingredient, and then resume the recipe when we return. The neuroscientist Etienne Koechlin summarizes the functioning of the frontal lobe in the following way. The rearmost portions respond to the stimulus; the lateral frontal cortex responds to the context; and the frontal pole responds to the episode. Concretely, when the phone rings and we pick it up, we are responding to the stimulus. When we are at a friend’s house and let it ring, we are responding to the context. And when the friend hops into the shower and asks us to pick up the phone if it rings, we are responding to the episode.
Impulsive violence could result from malfunctions in any of these levels of self-control. Take the violent punishment of children. Modern Western parents who have internalized norms against violence might have an automatic, almost visceral aversion to the thought of spanking their children, presumably enforced by the orbital cortex. Parents in earlier times and other subcultures (such as mothers who say, “Wait till your father gets home!”) might modulate the spanking depending on how serious the infraction is, whether they are at home or in a public space, and, if they are home, whether there are guests in the house. But if they are weak in self-control, or are inflamed by what they see as egregious naughtiness, they might lose their tempers, which means that the Rage circuit shakes free of frontal lobe control, and they thrash the child in a way they might regret later.
Adrian Raine, who previously showed that psychopaths and impulsive murderers have a small or unresponsive orbital cortex, recently carried out a neuroimaging experiment that supports the idea that violence arises from an imbalance between impulses from the limbic system and self-control from the frontal lobes.91 He scanned a sample of wife-batterers as they tried to ignore the meanings of printed words for negative emotions such as anger, hate, terror, and fear and just name the color in which they were printed (a test of attention called the Stroop task). The batterers were slowed down in naming the colors, presumably because their background anger made them hypersensitive to the negative emotions the words spelled out. And compared to the brains of normal people, who can examine the print without getting distracted by the words’ meanings, the batterers’ limbic structures were more active (including the insula and striatum), while their dorsolateral frontal cortex was less active. We may surmise that in the brains of impulsive assailants, aggressive impulses from the limbic system are stronger, and the self-control exerted by the frontal lobes is weaker.
Most people, of course, are not so lacking in self-control that they ever lash out in violence. But among the nonviolent majority some people have more self-control than others. Aside from intelligence, no other trait augurs as well for a healthy and successful life.92 Walter Mischel began his studies of delay of gratification (in which he gave
children the choice between one marshmallow now and two marshmallows later) in the late 1960s, and he followed the children as they grew up.93 When they were tested a decade later, the ones who had shown greater willpower in the marshmallow test had now turned into adolescents who were better adjusted, attained higher SAT scores, and stayed in school longer. When they were tested one and two decades after that, the patient children had grown into adults who were less likely to use cocaine, had higher self-esteem, had better relationships, were better at handling stress, had fewer symptoms of borderline personality disorder, obtained higher degrees, and earned more money.
Other studies with large samples of adolescents and adults have documented similar payoffs. Adults can wait indefinitely for two marshmallows, but as we have seen, they can be given equivalent choices such as “Would you rather have five dollars now or forty dollars in two weeks?” Studies by Laibson, Christopher Chabris, Kris Kirby, Angela Duckworth, Martin Seligman, and others have found that people who opt for the later and larger sums get higher grades, weigh less, smoke less, exercise more, and are more likely to pay off their credit card balance every month.94
Baumeister and his collaborators measured self-control in a different way.95 They asked university students to divulge their own powers of self-control by rating sentences such as these:I am good at resisting temptation.