GEORGE AINSLIE

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION ? (Accepted September 3, 1999)

INTRODUCTION

This paper examines whether a relatively new line of scientific inquiry may clarify some age-old puzzles about addiction. Although the most productive research on addictions in recent years has been the study of their brain mechanisms, I won’t say much about it, because it doesn’t do much to change our concepts of these puzzles. It only lets us see in detail the reward process that used to happen inside a sealed box. Rather I’ll talk about a phenomenon which has less obvious implications, but which might wholly recast the way we conceive addiction, the will, and indeed the self. I mean an obscure but remarkably robust finding of quantitative behaviorism called Herrnstein’s matching law. Few nonspecialists have heard of it, and few who have heard of it have considered its ramifications. Several of these seem inevitable, although I have to admit up front that deduction runs quite a ways ahead of experimental confirmation. The issue of addiction has always been a quandary for the law. The nature of the injury to others is cloudy, and any affront to morality itself is usually defined in ad hoc fashion, making it difficult to find a malum per se: Pot is moral in India but not in the U.S.; alcohol is moral in moderate doses, except for Muslims and Baptists, but immoral in high doses; tobacco started out immoral, became moral to the point of normality, and is now on its way to becoming immoral again. ?

Presented at the National Humanities Center Workshop on Addiction and the Law September 25–27, 1998. I thank John Monterosso for his comments. The U.S. Government’s right to retain a non-exclusive, royalty free licence in and to any copyright is acknowledged. Law and Philosophy 19: 77–115, 2000. © 2000 Kluwer Academic Publishers. Printed in the Netherlands.

78

GEORGE AINSLIE

The basic incentive for proscribing addictive behaviors seems to be that addicts are extremely provocative to the people around them. It’s not just that addicts steal and sometimes get violent. Most addictive substances cost little to make; most of their cost, which leads to stealing, is a consequence of our attempts to eradicate them. Furthermore, most addictive substances don’t cause violent behavior. Alcohol, the legal one, is the agent that causes the most by its own properties. Sometimes the excitement of speed or cocaine leads to violence; but again, the greatest cause of violence is the illegality that makes them the province of organized crime. Likewise, our irritation with addicts probably does not much depend on the idea that they’re a burden on public resources. Supporting a non-criminal addict would cost no more than welfare generally. Addicts do have a higher rate of medical illness than agematched peers. But some of that, including all of HIV, is a byproduct of illegality; and illnesses that cause addicts to die young actually reduce the burden on social security. Rather, addicts’ provocativeness is empathic. Addicts are selfabsorbed, a trait that’s mildly humorous in small doses, as when you see someone staring intently into a mirror. In larger doses selfabsorption seems to be disturbing, even frightening, the sensation that attaches stigma to schizophrenia and led to the persecution of masturbation in previous centuries. People take it as a major affront in its own right. At the same time, addicts’ experiences are a temptation to other people, at least early in their careers. Many people are drawn to experiment with drugs. Addiction seems like a bodysnatching force that threatens to lure our children into the ranks of the self-absorbed. Later in their careers, addicts’ social regression and self-neglect are painful to watch, an outcome that seems to have limited Swiss and other European experiments with legalization. Again, it’s our empathy that cries out for relief. The trouble is that empathic injuries have a dubious status in law. We punish haphazardly and without any consistent philosophical basis. It’s inconsistent to punish a person who kills dogs for sport, while regarding butchers as upstanding citizens, but we do this in deference to prevailing empathic habits. Likewise, the law punishes

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

79

some addictions and not others, not on the basis of their harm but according to accidents of social history. Punishment deters some people from using addictive substances, and this has been enough to sustain the strategy. Even campaigns that became distasteful to the public reduced the amount of addiction; Prohibition was a success in terms of numbers.1 However, punishment doesn’t deter thoroughly. Also, punishment makes those who are not deterred more wretched than they were before. Thus criminalization raises the stakes, but may not improve the average welfare of the population at risk. The comparative failure of criminalization casts doubt on addicts’ rationality. Criminals have often been called compulsive, but addicts seem even more so. The robustness of addiction has brought forth another response to our empathic discomfort, which is to see addiction as a disease like pneumonia, in which the victim is passive and needs treatment. However, the victims of other diseases want treatment. Addicts often refuse treatment, or, even more perplexingly, accept it and then work to defeat it. So the law is caught between conflicting interpretations of addiction as voluntary vs. involuntary. Voluntary injuries, even to such a vague thing as our fellow-feeling, can be sanctionable; diseases shouldn’t be. At a level even more basic than the law, addicts’ families experience the same ambivalence between blame and sympathy. Punishment and blame imply someone who is rational at least to the point of calculating the outcomes of her choices. Treatment and sympathy are appropriate for people who can’t do this. Thus the fundamental question the law asks about addiction is whether – and when, and to what extent – its victims’ choices are voluntary. Part and parcel of this question is how to understand addicts’ own ambivalence: ingesting their substance while saying they don’t want to. They’re clearly seeking the increasingly wellexplained rewards of ingesting, but why do they sometimes buy ways to make themselves stop at the same time? Especially, is there any way to shore up the side of their ambivalence that supports sobriety?

1

Fisher and Brougham, 1930; Merz, 1930/1968.

80

GEORGE AINSLIE

Addiction has many other specific properties that are not explained by conventional theories of either utility or disease. To name four: − Recovering addicts are often extremely legalistic. AA members are notorious for seeing the world in black and white terms; many chapters forbid their members all psychiatric drugs, for instance, on the theory that they may be a slippery slope back to drinking. People who see themselves as food addicts sometimes adopt the highly legalistic cognitive style of anorexia nervosa. These traits don’t represent a simple return to rationality. − Willpower often backfires when tried on addictions, leading to greater failure and demoralization. A major movement in treating addictions aims to reduce the rigidity of addicts’ wills. This is the thrust of the “relapse prevention” that was developed at the University of Washington twenty years ago.2 Similarly, it’s the gist of AA’s tenet, “one day at a time.” − Addicts often fail to notice basic facts about their addictions, such as how much they ingest or that ingestion under some circumstances counts as part of their addiction. In the extreme, they may develop whole dissociated personalities like Jekyll and Hyde. − Even addicts who have been sober for years often say they miss their addictions. Indeed, society itself has mixed feelings about addiction, and hence about coercing addicts. Intoxication is often shown positively in fiction, and people express strong sentiments against variants of Prohibition. These puzzles suggest the complexity of the addiction problem, which no widely accepted concept has given us the tools to explore. However, basic research in motivation during the last forty years has produced findings that will let us return to all of them with at least a starting hypothesis about why they might occur.

2

Marlatt and Gordon, 1980.

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

81

ADDICTION ENTAILS TEMPORARY PREFERENCE

Let’s examine first the possibility that addicts are simply following the best available strategy to maximize utility as they have experienced it. Utilitarian writers in fields from economics to abnormal psychology depict people as simply trying to maximize their satisfactions. To these writers, that’s what reason’s role is. It’s just straight thinking; the enemy isn’t passion per se, or the misuse of logic, but the simple miscalculation of your main chance. A theory of “rational” addiction adopts the discipline that your options have to compete for your favor on the basis of some elementary motivational quality – call it reward – a common dimension along which all options that can be substituted for each other have to compete. Reward is supposed to operate on your choices the way that natural selection operates on the animals in a species: It keeps the successful ones and drops the others. Utility theory is widely accepted in the behavioral sciences, but it has always had trouble saying why people regularly defeat their own best interests, often knowingly and in anticipation of regret. Why do people often fail to choose their best deal among familiar alternatives, even though rats in a maze succeed? The substance addictions may be the most striking example of this puzzle. Observers often blame the biological properties of the substance, even though experienced addicts knowingly re-addict themselves after they’re sober. Addictions that don’t depend on substances, like gambling and credit card abuse, have the same characteristics as drug addictions when carried to the same extent: the promise of a rush of feeling that pre-occupies the addict, the relentless narrowing of alternative opportunities through alienation of family, friends and employers, and even the adaptation of brain chemicals involved in pleasure, so that deprivation of the activity leads to symptoms of withdrawal – nausea, sweating, and other cholinergic side effects.3 These non-substance addictions form a conceptual link to a large class of ordinary “bad habits,” habits that people say they want to be rid of even while indulging in them: episodic rage, chronic procrastination, a bent for destructive relationships – all the patterns that the classical Greeks would have 3

Wray and Dickerson, 1981.

82

GEORGE AINSLIE

called akratic, incontinent.4 Unlike substance abuse, these behaviors can’t be blamed on the distortion of “natural” motivation by a molecule. Ironically, the straightforward simplicity of utility theory keeps it from providing an explanation for irrational behavior. If choice is just a matter of estimating maximal reward, the role of motivation in bad choices will be trivial, since any failure of maximization can come only from an error in the estimating process. Utility theorists have been at great pains to find a way around this conclusion. Is there a commonsense explanation for addiction? A common solution is to suggest that addiction occurs only in two special cases – when you don’t know its consequences, or when you don’t care about them. In the first case, you’re seen as committing yourself to the poorer but more conspicuous alternative before you know its cost – the hangover or damage to relationships is supposed to be a surprise. Certainly this could be a factor in how people first get addicted to something. Despite the fact that most smokers, for instance, know the dangers of smoking before they start, it may be that they can’t picture how strong their craving will become once they’re addicted. It’s also believable that alcoholics and drug addicts can’t imagine how their other options will narrow after they’ve indulged in their habit for a while. Thus there probably is a “primrose path” that leads many people to addiction.5 However, the popular impression that addictions trap people by the threat of withdrawal, so that once you start you can’t stop, has proven not to be true. In reality addicts stop many times, even withdrawing deliberately to cheapen their habits. Furthermore, the primrose path theory doesn’t explain why sober addicts have such a tendency to be re-addicted, or why active addicts try to stop by committing themselves in advance – for instance by taking Antabuse, a drug that makes alcohol sickening. Once an addict has become familiar with her options, conventional utility theory requires that any craving strong enough to lead to a relapse has to be strong enough to command consistent preference for the addiction. People 4

Mele, 1987. Smokers know dangers: Kessler, 1995; primrose path theory: Herrnstein and Prelec, 1992b. 5

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

83

may be deceived into taking up an addictive habit, but not into restarting a familiar addiction once they’ve escaped. For similar reasons we can rule out the theory that addicts just don’t care much about the future – that their addiction comes, in economic terms, from a steep discount curve for delayed events. Their discount curves may well be steeper than usual, but the “rational addict” – as pictured, for instance, by economist Gary Becker,6 wouldn’t ever try to kick her habit, and certainly wouldn’t try to restrict her own future range of choice by taking Antabuse. The rational addict who thinks that the high is worth the consequences when the opportunity is at hand should think so at a distance as well. Far from committing herself, she should always keep her options open in case new information makes another choice seem better. Straightforward value estimators have no reason to bind themselves. If your struggle with temptation can be bypassed simply by insight into your mental bookkeeping methods, volition as a distinct process becomes superfluous. Your calculation of your main chance flows smoothly into action, and your need for a will disappears. Facing this somewhat counterintuitive conclusion, utility theorists have looked outside of the evaluation process for a way to explain why self-control seems to take effort. In effect, they looked for disease processes to explain how people could develop preferences that were intrinsically temporary. Two areas of research have sometimes looked promising: Does classical conditioning cause the urges of addiction? Perhaps some factor causes unpredictable changes in utility.7 When a stimulus picked at random is regularly followed by the emotionally meaningful events that can trigger reflexes – food that produces salivation, pain that produces a racing heart, etc. – the random stimulus starts to produce behavior very similar to the reflex, without the experimenter rewarding or punishing this behavior. This is “classical” conditioning, the kind Pavlov discovered. At first glance this model promises a way out of the knowingself-harm enigma. If appetites can be conditioned, the argument goes, then maybe conditioned stimuli can impose motives on a per6 7

E.g. Becker et al., 1992. This possibility is proposed in Loewenstein, 1996 and 1999.

84

GEORGE AINSLIE

son that she otherwise doesn’t want, and those conditioned motives can overwhelm her normal, “rational” ones. A person might drink because the sight of a bottle elicits conditioned craving great enough to overcome her other motives. There’s an intuitive appeal to this scenario. Many people have had the experience in a restaurant of resolving not to order dessert, only to have their resolve broken by the appearance of the dessert cart. Couldn’t it be the conditioned appetite aroused by the sight and smell of the desserts that makes the difference? Certainly the appearance of desserts can induce cravings, as can the sight of a needle if you’re an IV drug addict.8 However, there’s nothing about being conditioned that should make a motive different from other motives. Once we’ve become familiar with dessert carts there’s no reason our motives shouldn’t develop in the usual way: The delicacies on the cart should give rise to appetite in proportion as (1) they can be eaten soon and (2) we expect to actually eat them. Indeed, it may be that all appetites become conditioned.9 Even if something were to prevent our anticipating this appetite, there’s nothing in conventional utility theory that would make us regret its occurrence; it represents just a sudden source of income, as it were, which, if it could have been anticipated at a distance, would have led to consistent preference in proportion to the intensity of the rewarding experience it made possible. Under conventional utility theory, it’s the resolution not to respond to new information about rewardingness that doesn’t make sense. My point is that although the experiences that get called conditioned clearly happen, they don’t explain temporary preferences. Waiters usually don’t ask if we want to see the dessert cart – they just bring it – because they know that we know what effect it will have. We know in advance that we’ll be lured by the desserts, try to forestall this by not asking for the cart, and are foiled when the cart comes without our asking. This is an example of temporary preference, all right, and the mere fact of the cart as well as its close8

O’Brien et al., 1986. Elsewhere I make extensive arguments (1992, 1999, and in press) that the reinforcement for conditioned appetites and other responses is no different from the reinforcement for goal-directed activities – i.e. utility or reward – but this isn’t important for the present discussion. 9

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

85

ness seems to play a role in producing appetite; but we still have to explain why we change our valuation of the appetite –> consumption sequence as it gets closer. In any case, people with major weight problems don’t mostly overeat because they keep getting surprised by appetizing stimuli. Does addiction come from brain chemistry? Many scientists hope that our expanding knowledge of the reward process itself will provide an answer. Since the 1950s neurophysiologists have been using brain electrodes to find how and where reward happens in the brain. They have found that most or all recreational substances, from alcohol and marijuana to cocaine and heroin, exert their rewarding effect by stimulating dopamine release in one small part of the midbrain, the nucleus accumbens, which is the same site where normal rewards like food and sex occur. Different strains of animals have widely different susceptibilities to drugs, and these can be modified by stimulating, ablating, or chemically treating neurons in the nucleus accumbens. These susceptibilities also change over time as an addicted animal consumes more of the drug, and the changes correlate with alterations in cell chemistry. Human addictions also have genetic predispositions. It seems likely that some addicts have been at least partly disposed to their habits by their neurological endowments.10 But these findings still don’t explain temporary preference.

THE CAUSE OF TEMPORARY PREFERENCES

These findings suggest that addiction may well be a disease, in the sense that the addict may have learned to generate cravings in surroundings associated with her substance, and/or may have been born with a brain that gets unusually great reward from the substance or little reward from alternative activities; she may even have deadened her brain to other kinds of reward through the addiction itself. However, it must be a disease of motivation: Pneumonia doesn’t come from or generate a desire for congestion in the lungs, but addictions 10

Animal brain mechanisms: Gardner, 1997; Hereditary basis of human addiction: Parsian and Cloninger, 1995; Nestler, 1992.

86

GEORGE AINSLIE

come from strong motives and their indulgence strengthens these motives in turn. The crucial point is that strong motives in themselves shouldn’t lead to temporary preferences, at least according to conventional utility theory. They don’t explain why an addict should fail while trying to stop, or should want to commit herself to stop in the future. They don’t say why an alcoholic should buy Antabuse. There is an innate tendency to form temporary preferences. Another solution to the self-harm puzzle has always been logically possible, but it has been seriously considered only recently: People may indeed maximize their prospective reward, but discount their prospects in a curve shaped differently from the function that bankers use. Few utility theorists question the assumption that people discount future utility the way banks do: by subtracting a constant proportion of the utility there would be at any given delay for every additional unit of delay. If a new car delivered today would be worth $10,000 to me and my discount “rate” is 10% a year, then the prospect of guaranteed delivery today of the same car would have been worth $9,000 to me a year ago, $8,100 two years ago, and so on (disregarding inflation, which merely subtracts another fixed percentage per unit of time). Utility theory operates the same way for internal reward, although it has to use a fanciful unit of measure like the “utile.” If drinking a bottle of whisky was worth 100 utiles to me right now and my discount rate for drinking was 20% per day, the prospect of today’s drinking would have been worth 80 utiles to me yesterday, 64 utiles the day before, and so on. Furthermore, if the drinking has a cost of 120 utiles that has to be paid the day after in the form of a hangover, reproaches from my family, etc., the net utility of drinking today will be 100 – (120 × 80%), or 4 utiles. So I should decide to drink. If I foresaw this episode from a day away, the net value would have been (100 × 80%) – (120 × 80% × 80%), or 80 – 76.2, or 3.2 utiles. At that point I would still have decided to drink. This arithmetic seems to describe the way many people behave toward large sums of money – bankers, at least, and everyone who is regarded as deciding “rationally”; but it misses the mark for drinkers, or least for people whose drinking is serious enough to

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

87

Figure 1. Exponential discount curves from two rewards due at different times. The taller, later line might represent the value of a pair of water skis in a future spring, and the shorter, earlier line the value of a sum of money at a point in the preceding fall. If the money at that point is worth less than the skis are at that point, it will always be worth less, even many years in advance.

involve hangovers and reproaches from their families. People who are strongly drawn to drinking – or taking drugs, or gambling, or kleptomania, or any other thrills of the kind that people regret – typically experience swings of preference between indulging their habit and giving it up. And the swings are often influenced by how close an opportunity for indulgence is. People trying to control a bad habit tend to keep a distance between themselves and opportunity – avoid the streets with their favorite bars on them and strategies like that.11 The car-buying and drinking illustrations assumed that the internal market in choices discounted future goods according to the same formula that financial markets do, at a constant rate per unit of delay. Pictured over time, this discounting is described by an exponential curve (Figure 1).12 If the internal market of choice didn’t do this, theorists have argued, it would impair the organism’s 11

For most people, the availability of a good excuse is even more important; but we’ll have to establish why anybody would need an excuse to herself before we can talk about that. 12 It is called exponential because it calculates value by an exponential, or power, function of the discount rate: Value = “Objective” value X(1 − Discount rate)Delay An outcome that loses 10% of its value for every unit of time it is delayed is worth (1.00–0.10),1 or 0.90 of its value, at one unit of delay (1.00–0.10),2 or 0.81 of its value, at two units of delay, 0.39 at ten units, 0.00003 at 100 units, and so on.

88

GEORGE AINSLIE

Figure 2. Hyperbolic discount curves from the same rewards as in Figure 1. At a distance they are valued in proportion to their heights, as with exponential discounting. However, at some point (in the fall, for instance) these curves may make the objectively smaller reward (the money) temporarily worth more than the objectively greater one (the skis). A person with these hyperbolic curves would sell the skis for that much money at that point, and a person with exponential curves would buy them. In the spring, however, both would value the skis at the same, undiscounted rate; the hyperbolic discounter would have to pay at least that to buy them back.

adaptiveness. If someone devalued future goods proportionately to their delay, for instance, their discount curve would be hyperbolic – roughly similar to the banker’s curve, but more deeply bowed, so that goods at both very short and very long delays would be valued more than they would with exponential curves, and goods in between valued less (Figure 2).13 In buying and selling between someone who discounted exponentially and someone who discounted hyperbolically, the latter would be at a disadvantage: Mr. Exponential could buy Mr. Hyperbolic’s water skis cheaply every fall, for instance, because the distance to the next summer would depress Mr. H’s valuation of it more than Mr. E’s. Mr. E could then sell the skis back to Mr. H 13

The simplest hyperbola is: Value = Objective value/Delay

However, this formula could probably never describe a natural process, since it would make value infinite at zero delay. A hyperbolic formula that makes “objective” value equal to discounted value at zero delay is: Value = Objective value/(1 + Delay)

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

89

every spring when the approach of summer sent Mr. H’s valuation of it into a high spike. Because of this mathematical pattern, only an exponential discount curve will protect a person against exploitation by somebody else who uses an exponential curve. Thus exponential curves seem not only rational, in the sense that they are consistent, but also adaptive. At first glance, it looks as if natural selection should have weeded out any organism that didn’t discount the future exponentially. Nevertheless, there’s a great deal of evidence that people’s natural discount curve is nonexponential. The experiment to test whether a subject’s discount curves cross is simple: You offer subjects a small reward at delay D versus a larger reward of the same kind that will be available at that delay plus a constant lag, L. A subject gets the small reward at delay D from the moment she chooses, or the larger reward at delay D + L. If she chooses the larger reward when D is long, but switches to the smaller reward as D gets shorter, she is showing the temporary preference effect that implies a discount curve more bowed than an exponential one. When experimenters have used rewards that the subject experiences (“consumes”) on delivery, people have shown a persistent tendency to reverse their preferences as D changes, evidence that their basic discount curves cross: Subjects exposed to noxious noise and given a choice between shorter, earlier periods of relief and longer, more delayed periods choose the shorter periods when D is small and the longer periods when D is long. College students show the same pattern when choosing between periods of access to video games. Retarded adolescents show it in choosing between amounts of food.14 Certainly at the gut level people’s discount curves cross. More remarkably, whatever cognitive methods people may have learned in order to compensate for hyperbolic discounting don’t spoil temporary preference experiments. Even money turns out to sometimes be chosen differently depending on D. Various groups of subjects have shown the change of preference over a range of D’s. Some excellent recent work by Leonard Green and his co-workers, and by Kris Kirby and the late Richard Herrnstein, have made it possible to describe the exact shape of subjects’ discount curve 14

Noise: Solnick et al., 1980; Navarick, 1982; Video games: Millar and Navarick, 1984; Food: Ragotzy, Blakely and Poling, 1988.

90

GEORGE AINSLIE

in similar amount-versus-delay experiments. It’s clearly hyperbolic for all age groups, although older subjects discount the future less steeply than younger ones.15 Hyperbolic discounting is even more evident in lower animals, which shows that it isn’t some quirk of human culture. In scores of experiments they’ve always chosen rewards in inverse proportion to their delays – and punishments in direct proportion. Animals also do what crossing discount curves predict: In amount-versus-delay experiments they choose the smaller, earlier reward when D is short, and the larger, later reward when D is long. It was the consistency of animal findings that led psychologist Richard Herrnstein forty years ago to propose a universal law of choice, which he called the matching law: that the proportion of rewards chosen matches their relative immediacies (as well as their amounts and frequencies of occurrence). tend to be chosen in direct proportion to their size and frequency of occurrence, and inverse proportion to their delay.16 Many researchers have since offered variations to fine-tune the matching law to describe individual differences in impatience, but the best seems to be one of the simplest:17 Value = Amount/(Constant1 + (Constant2 × Delay)) In practice the constants seem to stay close to 1.0, which simplifies the equation still further. When I discuss the likely consequences of hyperbolic discounting, I’ll be using this formula. Compensating for hyperbolic curves isn’t effortless. Most people would probably argue that since only exponential curves produce consistent preferences, they’re the ones that are objectively true, and that people should learn to correct their spontaneous valuations to fit them. After all, the intensity of many other subjective experiences 15 Ainslie and Haendel, 1983; Green, Fry and Myerson, 1994; Kirby and Herrnstein, 1995. 16 Herrnstein, 1961, 1997. The word “matching” comes from his original experimental design, in which pigeons pecked to get food on two independent keys that paid off at different rates. He found that relative rates of pecking matched the amounts, frequencies, and immediacies of reward. 17 Generalized from Mazur, 1987. I compare the possible formulas in Ainslie, 1992, pp. 63–76.

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

91

has long been known to follow hyperbolic curves,18 and people can learn to correct those impressions. It soon becomes second nature to a child that the telephone pole down the street is as tall as the one nearby, even though it forms a smaller image on her retina. Even where spontaneous impressions are misleading, people learn to trust instruments for measuring objective size – warmth by a thermometer, distance to travel by an odometer or map, etc. – without feeling that they’re wrestling with some inner resistance. They develop “object constancy.” Can’t people learn to value reward in proportion to its objective amount, just as we learn to gauge objective temperature and distance? Especially, can’t addicts learn not to be misled by the lure of an imminently available high? That’s what conventional utility theory calls for; but despite the availablity of objective ways of measuring time and value, such adjustment seems to occur irregularly, sometimes not at all. It usually takes some kind of effort (sometimes called “willpower”) to evaluate a lesser present good as less desirable than a greater one in the future. This is where the analogy of delay to other sensory impressions like length breaks down: A person may move through time toward a goal just as she moves through space toward a telephone pole. The matching law formula describing her spontaneous valuation of a goal is close to the formula for the retinal height of the pole.19 But the pole doesn’t seem to get larger as it gets closer, whereas the goal often seems to get more valuable. Insofar as the person fails to make the correction in value that corresponds to her correction of retinal height, poorer goals that are close can loom larger than better, distant goals. She fails to develop a faculty for “utility constancy.” Hyperbolic curves make it easy to explain addiction. It emerges as the state of nature or original sin, everyone’s innate challenge. Heredity or other factors may make the challenge bigger for some than for others, but people tend to recognize it as qualitatively the same as they themselves face, and resent granting addicts exemption 18

Psychologist J. Gibbon has pointed out that the matching law seems to be only one example of the principle by which many different physical qualities are sensed, known since the nineteenth century as the Weber/Fechner law (1977). 19 Y = l/X, where Y is the magnitude in question and X is the distance to the pole, or goal.

92

GEORGE AINSLIE

from the common responsibility. Before discussing this issue, we need to look at what lets people sometimes not give in to their short range preferences. The interests that rewards shape bargain among themselves to create a “self.” Is it possible for a marketplace of hyperbolically discounted choices ever to look like a single individual? The problem is that a mind which discounts reward hyperbolically can’t be the straightforward value estimator that an exponential discounter is supposed to be. Rather it will be a succession of estimators whose conclusions differ; as time elapses these estimators shift their relationship with one another between cooperation on a common goal and competition for mutually exclusive goals. Ulysses planning for the Sirens must treat Ulysses hearing them as a separate person, to be influenced if possible and forestalled if not. Consider one of the most widespread examples: A person may hate to go to bed at a prudent hour, even though she hates even worse getting up in the morning without enough sleep. Her mind this morning curses her mind of last night and tries to forestall her expected mind of tonight, but runs up against the effect of hyperbolic discount curves: Her mind holds a population of reward-seeking processes that have grown to survive in contradiction to each other, and that endure despite each other. She keeps on staying up late when the chance is at hand and the morning is far away – unless she can do someting to bring the incentives to get sleep, which are larger in the aggregate, to bear on the moment she chooses whether to go to bed. Any explanation of her options has to account for our observations not only of unity but of varying degrees of disunity, ranging from preference reversals in “normal” people to selves that actually look like Jekyll and Hyde. The prime suspect is the deeply concave shape of the discount curves in Figure 2, which limits what the market of choice can do to unify a person’s purposes over time. Ulysses’ wish to sail home and his wish to hear the Sirens will be integrated only for individual moments; this integration will make different options dominant when it occurs at different times, producing a regular conflict between the mental operations that win out when the lure of the Sirens’ song is dominant and those that win out when the prospect of finishing the journey is dominant.

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

93

You could call the mental operations selected for by a particular kind of reward the person’s interest in that reward – Interests within the person are very like interests within a social group, those factions that are rewarded by (“have an interest in”) the goal that names them (e.g. “the petroleum interest,” “the arts interest”). Since a person’s purposes can be coherent except where conflicting rewards dominate at successive times, it makes sense to name distinct interests only in that case. An alcoholic wouldn’t be said to have separate bourbon and scotch interests, even though they’re often alternatives, because at the time when she prefers bourbon she doesn’t increase her prospective reward by forestalling a possible switch to scotch. But she may have a drinking interest and a sobriety interest, such that each increases prospective reward in its own time zone by reducing the likelihood of the other’s dominance. Put another way, she doesn’t increase her prospective reward in either the long or short range by defending her choice of bourbon against the possibility that she may change to scotch; but she increases her prospective long range reward by defending her sobriety against drinking, and she increases her prospective short range reward by finding evasions of her resolve for the sake of drinking. Whichever faction promises the greatest discounted reward at a given moment gets to decide her move at that moment; the sequence of moves over time determines which faction ultimately gets its way. Where the alternative rewards are available at different times, each will build its own interest, and one interest will be able to forestall the other only if it can leave some enduring commitment that will prevent the other reward from intruding: If the person’s sobriety interest can arrange for her not to get too close to the opportunity to drink, the discounted prospect of drinking may never rise above the discounted prospect of the rewards for sobriety, and the sobriety will effectively have won. However, whenever the value of drinking spikes above that of sobriety, the drinking interest may undo the effect of many weeks of sobriety. The ultimate determinant of a person’s choice is not her simple preference, any more than the determinant of a legislature’s action is simple voting strength; in both processes, strategy is all. This process – power bargaining constrained by limited access to your means of expression – may be all that unifies a per-

94

GEORGE AINSLIE

son. Philosophers and psychologists are used to speaking about an organ of unification called the “self” that can variously “be” confident, divided, conscious, fragile, well-bounded, etc.; but this organ doesn’t have to exist as such. The only factor needed to impose unity on the various behavioral tendencies that grow from a person’s rewards may be the fact that they are, in effect, locked up in a room together. Four kinds of device can commit future choices. This brings us to the importance of this model of the self as a marketplace: If a person is a population of this kind of roommates, each clamoring to control the use of the room, how does the matter get decided? One interest can’t eliminate a competitor simply by providing more reward than the other does, either at one time or on the average, since the competitor might undo that interest’s choice when it became dominant at a later time. On the other hand, to continue to exist each interest has to be the highest bidder at some time; to achieve this each may have to constrain others, and not be too constrained by them. Just because an interest is dominant at one moment doesn’t mean it will get its intended reward; while an interest is dominant it has to forestall conflicting interests long enough to realize the reward on which it’s based. For long range interests this usually means committing the person not to give in to short range interests that might become dominant in the future. Long range interests don’t usually conflict with each other, except in the trivial sense of being close choices, because the effect of distant rewards tends to be proportional to their “objective” size; the less well-rewarded of two equally long range interests tends not to survive, but there is no time that this interest includes an incentive to resist this fate, that is, no time when such resistance would increase the person’s prospective discounted reward. For short range interests, survival usually means evading these precommitments. However, short range interests are also served by committing the person not to act in other, incompatible short range interests; and sometimes they can even commit the person to disobey long range interests. While on an eating binge a person avoids information about calories that might remind her of a diet, for instance, and is incidentally forestalled from having a sexual adventure.

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

95

There seem to be four kinds of tactics you can employ to precommit future choice:20 1. You can make it physically impossible to choose a future alternative, or arrange for additional outside incentives that will influence a future self. 2. You can try to avoid information that would change your mind. If you already know that a seductive reward is available, you can try to avoid thinking about it: “If you speak of the Devil, he’ll appear.” This is the advice that was most respected in our culture before Freud pointed out the bad side-effects of repression. 3. You can cultivate or inhibit the motivational processes that have intrinsic momentum – for instance, nip an emotion in the bud or deliberately provoke a contrary one. These processes can change how the expectation of reward influences your choice, at least in the near future. Once your appetite for a particular satisfaction is aroused it has a committing effect that lasts for a while. 4. You can make a personal rule to stick to your current plan. This is by far the most powerful and adaptable weapon against addictions, but also the most mysterious. What is there about “making a rule” that adds anything to your power to resist changing motivation?

A SPECIFIC MECHANISM FOR WILL

Conventional utility theory doesn’t suggest any role for personal rules – but of course, this theory doesn’t recognize a temporary preference problem to begin with. People commonly speak of a kind of self-control that doesn’t involve any of the first three devices, but their reports aren’t revealing. When they give up smoking or climb out of debt they mostly say they “just did it.” Words like will, character, intention, and resolve are often applied, but don’t suggest how they help somebody resist overvaluing an imminent reward. However, several authors dating back to classical Greece have proposed that the active ingredient in will is the property of uniting actions under a common rule. Aristotle spoke of universal 20

Described more fully in Ainslie, 1992, pp. 130–155, and in press.

96

GEORGE AINSLIE

“opinions” as leading to continence more than particular ones, for instance. The Victorian psychologist Sully said that the function of the will is to “unite . . . particular actions . . . under a common rule,” so that “they are viewed as members of a class of actions subserving one comprehensive end.”21 This has recently been proposed anew by two research psychologists. Gene Heyman has reported that pigeons facing one of the usual schedules that elicit hyperbolic valuations will move in the exponential direction if given a special signal that predicts additional reward every time they do so. As Heyman interprets it, they have come to make their choices in an “overall” context instead of a “local” one, although the additional reward blurs the implications of this finding. Similarly, Howard Rachlin has said that self-control comes from choosing “patterns” of behavior over time rather than individual “acts.” He also designed a pigeon experiment, in which subjects made an impulsive choice significantly less often when thirty previous nonimpulsive choices were required than when it stood by itself.22 The effect of choosing in categories doesn’t seem to depend on the intricacies of human cognition. It does require that the impact of repeated rewards accumulate somehow so that the rewarded behavior can be weighed against alternatives. This accumulation hasn’t been studied much, but some experiments suggest that the discounted effects of each reward in a series simply add, at least in pigeons.23 If this is at least roughly correct, then we have more reason to believe that the shape of the basic discount curve is hyperbolic, or at least deeply bowed. Bundling choices together wouldn’t affect the direction of preference if discount curves were exponential. Consider a series of larger, later rewards and their smaller, earlier alternatives, as in Figures 3 and 4. Exponential discount curves stay proportional, even when a whole series is added together. But with hyperbolic discounting, the reward for choosing among categories of reward will be the expected value of series of larger rewards like those depicted in Figure 4; the reward for a choice that seems unique will be just the curves from the single pair (as in Figure 2). Thus if discounting is hyperbolic, choosing behaviors in whole cate21 22 23

Aristotle, Nichomachean Ethics, 1147a24–28; Sully, 1884, p. 631. Heyman, 1996; Rachlin, 1995. Mazur (1986) confirmed an earlier report of McDiarmid and Rilling (1965).

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

97

Figure 3. The effect of summing the exponential discount curves from six pairs of rewards like the pair in Figure 1. The summed curves are higher than single curves, but they stay proportional to each other, just like the curves from a single pair or rewards.

Figure 4. The effect of summing the hyperbolic discount curves from six pairs of rewards like the pair in Figure 2. The summed curves are not only higher than the single curves, they no longer cross in the period before the first small-early reward is due.

gories will lead to less impulsiveness. This explains how people with fundamentally hyperbolic discount curves can sometimes learn to choose as if their curves were exponential; summing hyperbolic curves bends their shape in the direction of exponential curves (Figure 5).

98

GEORGE AINSLIE

(A)

(B) Figure 5. Comparison of (A) exponential and hyperbolic curves from a single reward, vs. (B) summed hyperbolic curves from series of rewards and an exponential curve from the earliest of these rewards. The summed hyperbolic curve moves significantly closer to an exponential shape than did the hyperbolic curve from the single reward.

The will is vulnerable to special cases. The trouble with will is that the insight about choosing according to principle doesn’t eliminate the attraction of small, immediate rewards; it offers only a discipline that a long range interest would benefit from at the expense of short range interests, if only the person were consistently motivated to follow it. People who have learned a “higher” or “richer” principle of choice aren’t thereby freed from temptation. We aren’t very old before we discover a perverse truth: If behaving according to categorical principles promises more discounted, expected reward than making isolated choices does, then making an isolated choice now and expecting to act by rule in the future promises still more. The tough question is not how molar bookkeeping recruits motivational support for long range interests, but how this process defends

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

99

itself from short range interests, sometimes unsuccessfully. Acting in my long range interest, how do I keep a short range interest from repeatedly proposing an exception to my rule, “just this once?”: by realizing that exceptions set precedents, and if I go too far I’ll lose my expectation of getting anything like the series of larger, later rewards. In this way hyperbolic discount curves make self-control a matter of self-prediction. This effect will be especially noticeable where self-control is tenuous. The hyperbolic discounter can’t simply estimate whether she is better off dieting or eating spontaneously, and then follow the best course. Even if she figures dieting is better from a perspective of distance, she won’t know whether or not she’ll regularly prefer to eat ad lib when she’s hungry. If she expects to eat ad lib, her long range perspective will be useless to her unless she can use one of the first three kinds of precommitments I described above – not a rich selection – or make a personal rule. Her expectation that her rule will succeed may be enough to motivate sticking to it, but only insofar as she thinks it will be enough. If she then violates the diet and loses faith in it, this condition will magically stop being enough. Personal rules are a recursive mechanism; they continually take their own pulse, and if they feel it falter, that very fact will cause further faltering. Even pigeons can learn to peck a key early in a trial to prevent a temptation from appearing subsequently. Soon after I first reported this finding, the constitutional scholar Laurence Tribe wrote that it was the simplest example of constitution-making, the way a whole nation binds itself not to be impulsive.24 Certainly hyperbolic discount curves should create such a motive in species from the lowest to the highest. But a constitution depends either partly or wholly (as with Britain’s) on honoring precedent. Precedent was irrelevant to my pigeons, as it is to the alcoholic relying on Antabuse. It is the very core of the will, which could be looked upon as an individual’s version of an unwritten constitution. Will is cooperation in an intertemporal prisoner’s dilemma. The will to stick to a diet has the same properties as the “will” of the nations in World War II not to use poison gas, or of those since not to use 24

Ainslie, 1974; Tribe (second edition), 1988, pp. 11–12.

100

GEORGE AINSLIE

nuclear weapons. This will is a bargaining situation, not an organ. In fact it can be well described in terms of bargaining theory: The relationship of bargaining agents who have some incompatible goals but also some goals in common is sometimes called limited warfare. Countries want to win trade advantages from each other while avoiding a trade war; a person today wants herself to stay sober tomorrow night, and tomorrow night will want herself to get drunk, but from neither standpoint wants to be an alcoholic. Whether the parties are countries or individuals or interests within an individual, limited warfare describes the relationship of diversely motivated agents who are somehow in the same boat. Among agents engaged in limited warfare with each other, there’s a practical mechanism for peace: Their mixture of conflicting and shared motives creates the incentive structure of a well-studied bargaining game, “the prisoner’s dilemma”: Two or more parties do better if they all cooperate than if they all don’t; but if only one defects against the others, she will do better still. Countries at war will do better by all not using poison gas than by all using it; but if one country surprises the others, it will do better still. Why did no country ever use gas throughout World War II? How, in fact, does this situation offer a road to peace? The key fact is that most parties don’t meet to fight only once, and the combatants were conscious of future battles. Since following suit is both the most obvious strategy and the most successful one in repeated prisoner’s dilemmas, each side could reasonably expect the other to do so, knowing only that the payoffs are in a prisoner’s dilemma pattern. As long as using gas would set a precedent that predicts your disposition in future battles, and your opponent hasn’t shown how she’s disposed to choose, you have an incentive to avoid using gas.25 But how does this model apply to intertemporal bargaining, where the limited warfare is among successively dominant interests? Their situation seems different. Successive motivational states are transient, so that by the time the person has entered a contrary frame of mind, she can’t retaliate against the earlier self that betrayed her. In one sense, a person can’t meaningfully be said to bargain with herself at a later (or earlier) time. You have no way to reward or 25

Schelling (1960, pp. 53–80) gives a clear description of limited warfare.

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

101

punish a bygone “past self” or threaten a future one who will exist when your present self is bygone in turn. However, successively dominant interests do have stakes in each others’ behavior that are very close to the ones in a literal prisoner’s dilemma: The threat that weighs on your current self’s choice in a repeated prisoner’s dilemma is not literally retroactive retaliation by a future self, but the risk of losing your own current stake in the outcomes that future selves obtain. Any means of grouping choices in a prisoner’s dilemma pattern has the potential to succeed as principle of self-control. Any set of parties to a limited war – countries, individuals, or successive motivational states within an individual – will seize upon such a grouping if it has this property: that all can see it as dividing areas of mutually advantageous cooperation from areas where the hope of cooperation is unrealistic. The resulting tacit agreements about areas of cooperation constitute rules – in the intertemporal case, personal rules. In effect this committing tactic creates a side bet: The players stake their whole expectation of getting the benefits of cooperation on each choice where cooperation is required; this expectation is the kitty of the side bet, and it may be much larger than what was originally at stake in a given choice. As a solution to temporary preferences, intertemporal bargaining gets both its strength and its weakness from the openness to interpretation of what constitutes cooperation: what choices set precedents for what others, and what choices are exceptions. The person facing a choice about sticking to a diet, or getting drunk, or indulging her temper, or procrastinating, is always tempted to say, “This choice is unique; it can stand on its own. I don’t have to worry about how I’ll look back on it when I face this other kind of choice later on; it’s just not part of what I bet myself I’d do.” This opportunity lets people have some flexibility in their commitments – If we encounter a kind of choice where the rules we set up seem to make us worse off, we can redraw those rules on the spot so they don’t apply to this kind of choice. But in the middle of a choice between a small, early and larger, later reward, the urge to see your way clear to take the early one is great, which leads people to gamble on claiming exceptions to their personal rules on

102

GEORGE AINSLIE

shaky grounds. The same problem exists in interpersonal bargaining: Cooperation between the countries in the choice about poison gas is threatened by the possibility of choices that are marginal in their aptness to be seen as precedents, such as the use of gas against a third nation or the use of an explosive that happens to give off toxic fumes. A country might hope to engage in such marginal behaviors without being seen as having betrayed its tacit agreement to cooperate. Because of this hope, there is substantially more risk that the country will engage in them and find it has hoped falsely than that it will commit an unambiguous betrayal. Recursive bargaining creates the experience of will. We like to think of ourselves as having an organ like the ego, which resembles a sailor, navigating by balancing stronger forces than her own: If I’m tacking against the wind, the force of the wind is much stronger than my arms. I can choose only how to set my sails against it. But in this analogy the strength I steer with is still my own, and unrelated to the force of the wind. To model strict utility theory a sailor would have to also steer by the force of the wind, which would mean that the rationale of the steering mechanism would somehow have to come from the properties of the wind, not the extraneous wishes of a person blown by it. The ultimate dilemma of the utilitarian is how to describe such a steering mechanism without making it seem improbable that the person whose will it represented would feel human – that she would experience authentic doubt, self-esteem, and other subtle feelings, particularly the one necessary for legal responsibility: freedom of will. Such a description has, in effect, to evoke the feeling of being a sailor from the logic of the wind. Going on conventional assumptions, any maximizer seems like a mere calculating machine, a throughput, as philosopher Martin Hollis once characterized it.26 However, in an intertemporal bargaining model, will is a recursive process. The person herself can’t be sure of what she’ll do in the future, and makes her current choice based on her best prediction. But this choice also affects her prediction, so that before she has acted on her choice she may predict

26

Hollis, 1983, p. 250.

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

103

again, and may then change her previous prediction and thus her choice. Although clearly pulled by identifiable motives, a person’s choice in a recursive process can’t be predicted with certainty. Nevertheless, choice is as strictly determined as the weather. This model predicts that a person’s choice will feel free to her, not in the sense of being random but as something she participates in but can neither predict with certainty nor explain from only her incentives themselves. Even in a familiar choice where she “knows” what she’ll do, her experience has also taught her that a small change – even a taunt from a friend about her predictability – can reverse the balance. The inscrutable human spirit that transcends mere obedience to our ostensible motives – what keeps the person from being Hollis’ “throughput” – comes from the partial responsiveness of the intertemporal crowd that she senses around her present self, which is made up of her own choices but of which she forms only a small part at any given moment.

THE NEGATIVE SIDE-EFFECTS OF WILL

All self-control devices can impair your reward-getting effectiveness: If you have yourself tied to a mast you can’t row; if you block attention or memory you may miss vital information; and if you nip emotion in the bud you’ll become emotionally cold. Unfortunately, the most powerful and flexible strategy against the effects of hyperbolic discounting, personal rules, also has the greatest potential for doing harm. A person’s perception of the prisoner’s dilemma relationship – and the valued willpower that results from this perception – doesn’t simply cure the problem of temporary preference. Intertemporal bargaining doesn’t make your spontaneous preferences consistent over time. Rather it formalizes internal conflict, making some selfcontrol problems better, but some worse. Cobbled together from properties of hyperbolic discounting that apparently didn’t much affect evolution before humans appeared, it remains something of a stopgap. Willpower may be the best way we know to stabilize choice, but it turns out to have serious side-effects.

104

GEORGE AINSLIE

We don’t usually recognize these side-effects as part of using willpower. Will seems to be a source of pure strength, with no relationship to such abnormal symptoms as loss of emotional immediacy, abandonment of control in particular areas of behavior, blindness toward one’s own motives, or decreased responsiveness to subtle rewards. But I argue that these four phenomena are part and parcel of a reliance on personal rules. They may go so far as to erase the net attractiveness of willpower, even for a person’s longest range interest. 1. Willpower overshadows spontaneous experience. The perception of a choice as a precedent often makes it much more important for its effect on future expectations than for the rewards that literally depend on it. When this is true, your choices will become detached from your immediate options and take on an aloof, legalistic quality. It’s often hard to guess whether you’ll look back on a current choice and see it as a lapse. Did eating that sandwich violate my diet or not? Under the influence of an imminent reward you may claim an exception to a rule, but later think you fooled yourself, that is, see yourself as having had a lapse. Conversely, you may be cautious beyond what your long-range interest requires, for fear that you’ll later see your choice as a lapse. This rationale will make you compulsive. Every lapse reduces your ability to follow a personal rule, and every observance reduces your ability not to. Errors in either direction impose costs that would never result from conventional exponential curves, since those curves wouldn’t lead to recursive decision-making in the first place. For a person resisting the lure of addiction, these costs take the form of excessive legalism, one of the traits of recovering addicts that we set out to explain. 2. Willpower is disproportionately vulnerable to lapses. As the Victorian psychologist Sully observed, “every gain on the wrong side undoes the effect of many conquests on the right.”27 This is probably true because impulses need only a brief moment of dominance to capture behavior, while the will needs consistent control. The motivation needed for a sooner reward to spike above the summed curves of later rewards will be greatest for the first time this happens; each 27

1884, p. 669.

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

105

time it happens your expectation of getting the whole set of delayed rewards becomes less, and thus requires less alternative motivation to pierce it the next time. Awareness of this snowball effect in turn will make a lapse seem like more of a disaster, which will make the loss of prospective larger-later reward even greater. The obvious way to repair a lapse is to abandon the particular circumstance in which it occurred. This means attributing the lapse to a particular aspect of your present situation, even though it will make self-control much more difficult in future cases where that aspect is present. For instance, you may decide that you can’t resist the urge to panic when speaking in public, or to lose your temper at incompetent clerks, or to smoke after meals. If you no longer hope that your rule will hold in these cases, these urges will seem to command obedience automatically, without an intervening moment of choice. Your discrimination of this special kind of area has a perverse effect, since within it you see only failure predicting further failure. I’ve called this area, where a person doesn’t dare attempt efforts of will, a lapse district, by analogy to the vice districts in which Victorian cities tolerated the vice they couldn’t suppress. Where the encapsulated impulses are clinically significant, a lapse district gets called a symptom. Thus the perception of repeated prisoner’s dilemmas stabilizes not only long range plans but lapses as well.28 Here we have an explanation for how willpower can backfire, another hitherto puzzling aspect of addiction. 3. Willpower motivates misperception. Personal rules depend heavily on perception and recall – noticing and remembering your choices, the circumstances in which you made them, and their similarity to the circumstances of other choices. And since personal rules organize great amounts of motivation, they naturally can create a great incentive to suborn the perception and recall processes. When a lapse is occurring or has occurred, it will often be in both your long and short range interests not to recognize that fact: Your short range interest is to keep the lapse from being detected so as to let it continue. Your long range interest is also at least partially to keep 28

Discussed further in Ainslie, 1992, pp. 193–197.

106

GEORGE AINSLIE

the lapse from being detected, because acknowledging that a lapse has occurred would lower the expectation of self-control that you need to stake against future impulses. After a lapse, the long range interest is in the awkward position of a country which has threatened to go to war in a particular circumstance that has then occurred. The country wants to avoid war without destroying the credibility of its threat, and may therefore look for ways to be seen as not having detected the circumstance. Your long range interest will suffer if you catch yourself ignoring a lapse, but perhaps not if you can arrange to ignore it without catching yourself. This arrangement, too, must go undetected, which means that a successful process of ignoring must be among the many mental expedients that arise by trial and error – the ones you keep simply because they make you feel better without your realizing why.29 As a result, money disappears despite a strict budget, and people who “eat like a bird” mysteriously gain weight. Clouding of consciousness in the face of temptation has been familiar to observers from Aristotle to the present day.30 Here’s a motivational pattern that could easily create a black market, indeed an underworld, of those interests that can control attention so as to block your notice and recollection. In addicts this distortion of perception ranges from their notorious denial to the dissociation of whole personalities, a finding that is rare in the general population.31 4. Willpower can make you unresponsive to subtle rewards. You’d think that judging choices in whole categories rather than by themselves would have to improve your overall rate of reward. Whenever it didn’t, you ought to be able to call off the side bet that enforced the rule. However, a number of modern writers have warned that your sense of will may decrease rather than increase if you bind yourself too extensively to rules. 29 As Erdelyi has pointed out (1990), the unconscious but goal-directed effort to forget which the psychoanalysts call repression does not differ in nature from the conscious kind (suppression). I would suggest that its unconsciousness is shaped by the incentive to avoid losing the stakes of personal rules. 30 Aristotle: Bogen and Moravcsik, 1982; for a modern example, see Sjoberg and Johnson, 1978. 31 Dunn et al., 1993; Ross et al., 1992.

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

107

Theologians have long known of the dangers of ‘scrupulosity,’ the attempt to govern yourself minutely by rules. The philosopher of religion Paul Ricoeur has pointed out that freedom of will is encroached upon not only by sin but by moral law, through the ‘juridization of action’ by which “a scrupulous person encloses himself in an inextricable labyrinth of commandments.” Psychotherapists have embraced this insight. Most of the schools of therapy that developed in Freud’s wake made overgrown personal rules their chief target. Martin Hollis’ recent criticism of ‘economic man’ raised the same concern: A Rational Economic Man . . . must be able to reflect on whether the upshot of his calculations is truly the rational course of action. This is to raise a query about the base of the calculation . . . A person may find himself locked into his preferences against his real interests.

These writers’ message is that personal rules can become prisons.32 How can this happen? Why should anyone who wasn’t in the thrall of an immediate temptation ever conclude that she was trapped by her rules, and even hire a psychotherapist to free her? The supposed imprisoning process is made out of reward – It’s the stake of an intertemporal prisoner’s dilemma. If a person regrets the existence of such a stake even from the perspective of distance, but remains influenced by it, we’d have to conclude that her personal rule doesn’t serve her longest range interest. It might seem from the logic of summing discount curves (Figure 4) that cooperation in a repetitive prisoner’s dilemma would have to serve the players’ long range interests, or else they’d never adopt this strategy. But in everyday life a person can discern many possible prisoner’s dilemmas in a given situation; and the way of grouping choices that finally inspires her cooperation need not be the most productive, largely because of the selective effect of countability: Personal rules operate most effectively on countable goals. Thus the ease of comparing all financial transactions lets cash prices fluctuate much less over time than, say, the value of an angry outburst, or of a night’s sleep. The motivational impact of a series of moods must be less than that of an equally long series of cash purchases. 32

Ricoeur, 1971; many therapists and others reviewed in Ainslie, 1992 and in press; Hollis, 1983, p. 260.

108

GEORGE AINSLIE

The impact of having rewards marked by discrete stimuli can be seen in experiments where the cost of bad choices is lower payoffs vs. longer delays to payoffs.33 Amounts are eminently countable; delays are a matter of intuition unless someone specifically measures them. Accordingly, subjects achieve rational behavior toward amounts much more readily than to delays. By the same logic, when mid-range interests are based on well-marked rewards, and their richer, longer range alternatives are harder to specify, the more concrete options may win out by using personal rules. Rules that will prove too concrete from a long-run perspective may still be attractive to someone trying hard to avoid her impulses, especially someone who has had a conspicuous addiction – for instance, the person who diets to the point of anorexia nervosa to end a history of overeating. It’s easier to enforce specific rules about diet than more subtle rules like “eat what you need” or “eat what you’ll be glad of in retrospect,” though if the latter were adequate rules they’d permit the most reward in the long run. When you seek the comparative safety of having the most clearcut criteria for your personal rules, you may be forestalling not only short range impulses but also your chances for the richest reward in the long run. So the most workable rules for intertemporal cooperation don’t necessarily bring the most reward in the long run. The mechanics of policing this cooperation may produce the intrapsychic equivalent of regimentation, which will increase your efficiency at rewardgetting in the categories you’ve defined, but reduce your sensitivity to less regular kinds of reward. Both hyperbolic discounting and the personal rules that compensate for it have distorting effects. Personal rules confront us with the paradox of definition: that to define a concept is to alter it, in this case toward something more mechanical. A person who concludes that she should maximize money becomes a miser; one who rules that she should minimize her openness to emotional influence develops the numbing insensitivity that clinicians have named alexithymia; if she concludes she should minimize risk, she becomes obsessively careful; and so forth. The logic of rules may come to so overshadow your responsiveness to experience that your behavior becomes formal and inefficient. A 33

For instance “melioration” experiments (Herrnstein and Prelec, 1992b).

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

109

miser is too rigid to optimize her chances in a competitive market, and even a daring financier undermines the productiveness of her capital if she rules that she must maximize each year’s profit. Similarly, strict autonomy means shielding yourself against exploitation from others’ ability to invoke your passions; but alexithymics can’t use the richest strategy available for maximizing emotional reward, the cultivation of human relationships. And avoidance of danger at any cost is poor risk management.34 Modern culture has been slow to recognize the dilemma of selfcontrol: that we’re endangered by our willpower as well as by our impulses. For instance, writers now wring their hands both about the average citizen’s rising body mass index and about the prevalence of dieting in the young, without noting that this combination means that the enemy is now approaching from two opposing directions. This and the previous three side effects may sometimes keep the exercise of willpower from seeming better than an addiction, even from the perspective of distance, which gives the truest reading of “objective” value. Personal rules may sometimes not be worth the trouble. In these four ways personal rules sharpen the basic conflict of successive motivational states and raise its stakes. Rules we create in our long range interest may or may not wind up advancing this interest against shorter range ones. After the need for clarity has taken its toll on subtlety, and overcaution has reduced flexibility, and undercaution with its consequent lapses has eroded resolve and corrupted selfobservation – in short after the makeshift nature of our attempt at deciding according to principle has caught up with us – the conflict between global and local approaches to choice-making is no longer simply one of long vs. short range interests. Sometimes a particular global approach is less productive, even in the long run, than local, myopic alternatives. In the interpersonal realm the dangers of rules are much better known. The English long ago established courts of equity to correct distortions that arose from laws, and that great social rule-maker, Jeremy Bentham, cautioned that rules shouldn’t be fully binding. 34

Alexithymia: Nemiah, 1977; the cost of rules for maximizing annual income: Malekzadeh and Nahavandi, 1987.

110

GEORGE AINSLIE

A recent review by legal scholar Cass Sunstein makes it clear that social control by rules creates side-effects analogous to my problems 1, 3, and 4: The need for preserving precedents makes them too rigid, this rigidity “drives discretion underground” into transactions that aren’t a matter of record, and the need to use available bright lines between what is and isn’t permissible both forbids innocuous activities and licenses cleverly defined harmful ones.35 Problem 2 is also evident: Some potential drug addicts may be protected by legal deterrence, but many who are not deterred become identified criminals who are worse off than they would be if drugs weren’t illegal. The robustness of suboptimal rules may sometimes make addictions more attractive. Better to be fat, someone might think, than anorectic. Your will may become so confining that a pattern of regular lapses actually makes you better off in the long run. The lore of addictionology often attributes bingeing to a patient’s inhibitedness in the rest of her life; her general overcontrol is said to set up periodic episodes of breaking loose. The model of intertemporal bargaining predicted by hyperbolic discount curves provides a specific rationale for this pattern. Alcoholics are sometimes described who become nicer, or more genuinely creative, or more fully human when drunk. Furthermore, some addicts plan binges in advance. Such people may believe that their binges are undesirable – indeed, “rationality” will almost certainly dictate such a belief – but the therapists they hire find them mysteriously unresponsive to treatment. The patient who arranges for drinking several days in advance – goes off her Antabuse, for instance, or brings bottles to her rehabilitation program for later use – can’t simply be yielding to a short range impulse. This is behavioral evidence that she experiences her rational plans as overly narrow rules which, even at a distance, appear to need hedging. This phenomenon suggests why a simplistic policy of “the more willpower, the better” contradicts the experience of many addicts. They’re able to listen to reason only when reason, represented by personal rules, stops starving their own long range interests.

35

1995, pp. 991–996; he discusses Bentham on pp. 1006–1007.

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

111

IMPLICATIONS FOR LEGAL INTERVENTION

If the self is a population of interests conducting limited warfare with each other to a greater or lesser extent, and the will is a bargaining stance in a somewhat volatile intertemporal prisoner’s dilemma, how should the law intervene? First there is the question of whether sobriety is always a person’s better option. There may be some emotionally inept people who face a choice between being a drunkard, say, and adopting what would be the equivalent of the compulsive overcontrol in anorexia nervosa. Obviously a third option, emotional growth, is better than either, as it is in the conflict between anorexia and obesity; but it may not seem possible, and in any case represents a separate question. It could be argued that the United States as a whole made a similar choice when it rejected Prohibition, and for reasons that look like the negative side-effects of will writ large: the legalism in what had been an area of spontaneous choice, the transformation of the undeterred into criminals, the creation of an underworld from the resulting secrecy, and simply missing the freedom to drink in moderation without conflict. This is not to argue that most people, or any people, are better off addicted, but to point out that the question is not always crazy. Next, given that it’s in a person’s longest range interest to be sober, how can legal policy support this interest against her short range interests in addiction? It’s no surprise that the intertemporal bargaining approach finds both conventional interpretations wanting. To regard the addict as helpless is to abandon the greatest tool of her long range interest, her will.36 To regard her as rational is to deny the disportionate power even ordinary rewards have if immediate, much less those created by a predisposed brain. The best conclusion is that addiction is just an outsized case of a vulnerability that everyone has, and that it may have become outsized either from genetic endowment or a history of bad choices, or both. The problem for legal policy-making is that there seems to be no natural 36

AA’s seeming abandonment of the will may actually be a way of tacitly appealing to it, much as the Calvinists’ doctrine of predestination is said to have increased the effective willpower of its adherents – by discouraging the hedging that usually erodes the will’s power. I have developed this argument elsewhere (1992 and in press).

112

GEORGE AINSLIE

line between addiction and ordinary weakness of the flesh. It may be useful to recognize a special status for intractable addictions, but the threshold for such a status has to be arbitrary, much like that for bankruptcy. The seeming intractability of addiction comes from the person’s recursive reaction to failures of her will, creating a lapse district for the addiction. Even so, preferences in someone’s short range interest are temporary by definition, so that the addict is not a wholehearted opponent, but a conflicted one. How do we shore up one faction in someone else’s ambivalence? Policeman and therapist alike become like envoys to a sovereign country. Addiction may have torn the country with dissension, but it is still ready to unite – indeed, probably eager to unite – against outside coercion. Coercion undermines a person’s will to do what we demand. It replaces the incentive for the person to maintain her credibility to herself with external incentives, thus reducing the motivational basis of her will and at the same time offering a game much like cops-androbbers: If someone else has taken responsibility for watching her behavior, she may be comfortable trying to get away with as much as she can. That is, if an external agent is acting for the “cop” half of her ambivalence, she can act wholeheartedly for the “robber” half. As in the international situation, the efficient solution is diplomacy, which involves winning the trust by both sides that adequate rewards in both short and long ranges are possible. This is psychotherapy. Diplomacy failing, war may be necessary, but it should be decided upon with the knowledge that it may fatally harm a given addict’s long range interest. The consideration that limits our exercise of that option is the same one that started us intervening in addictions to begin with, that is, empathy – the way that any addict can hold our own feelings hostage. At the practical level, these are not new insights. The theoretical value of deducing the implications of hyperbolic discounting for deliberate self-defeating behaviors is to make them not paradoxical, not an exception to general laws of motivation. This in turn may offer ways out of the tired stalemate between disease and utility models of addiction.

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

113

REFERENCES Ainslie, G., ‘Impulse Control in Pigeons’, Journal of the Experimental Analysis of Behavior 21 (1974): 485–489. Ainslie, G., Picoeconomics: The Strategic Interaction of Successive Motivational States Within the Person (Cambridge: Cambridge University Press, 1992). Ainslie, G., Breakdown of Will (New York: Cambridge University Press, in press). Ainslie, G. and Haendel, V., ‘The Motives of the Will’, in E. Gottheil, K. Druley, T. Skodola and H. Waxman (eds.), Etiology Aspects of Alcohol and Drug Abuse (Springfield, IL: Charles C. Thomas, 1983), pp. 119–140. Bain, A., The Emotions and the Will (New York: Appleton, 1859/1886). Becker, G. S., Grossman, M. and Murphy, K. M., ‘Rational Addiction and the Effect of Price on Consumption’, in G. Loewenstein and J. Elster (eds.), Choice Over Time (New York: Russell Sage, 1992). Bogen, J. and Moravcsik, J., ‘Aristotle’s Forbidden Sweets’, History and Philosophy 20 (1982): 111–127. Dunn, G. E., Paolo, Anthony M., Ryan, J. J. and VanFleet, J., ‘Dissociative Symptoms in a Substance Abuse Population’, American Journal of Psychiatry 150 (1993): 1043–1047. Erdelyi, M. H., ‘Repression, Reconstruction and Defense: History and Integration of the Psychoanalytic and Experimental Frameworks’, in J. L. Singer (ed.), Repression and Dissociation: Implications for Personality Theory, Psychopathology, and Health (Chicago: University of Chicago, 1990). Fisher, I. and Brougham, H. B., The ‘Noble Experiment’ (New York: Alcohol Information Committee, 1930). Green, L., Fry, A. and Myerson, J., ‘Discounting of Delayed Rewards: A Lifespan Comparison’, Psychological Science 5 (1994): 33–36. Gardner, E. L., ‘Brain Reward Mechanisms’, in J. H. Lowinson, P. Ruiz, R. B. Millman and J. G. Langrod (eds.), Substance Abuse: A Comprehensive Textbook (Baltimore: Williams & Wilkins, 3d ed. 1997), pp. 51–85. Gibbon, J., ‘Scalar Expectancy Theory and Weber’s Law in Animal Timing’, Psychological Review 84 (1977): 279–325. Herrnstein, R., ‘Relative and Absolute Strengths of Response as a Function of Frequency of Reinforcement’, Journal of the Experimental Analysis of Behavior 4 (1961): 267–272. Herrnstein, R. J., The Matching Law: Papers in Psychology and Economics, in H. Rachlin and D. I. Laibson (eds.) (New York: Sage, 1997). Herrnstein, R. J. and Prelec, D., ‘Melioration’, in G. Loewenstein and J. Elster (eds.), Choice Over Time (New York: Sage, 1992a), pp. 235–264. Herrnstein, R. J. and Prelec, D., ‘A Theory of Addiction’, in G. Loewenstein and J. Elster (eds.), Choice Over Time (New York: Sage, 1992b), pp. 331–360. Heyman, G. M., ‘Resolving the Contradictions of Addiction’, Behavioral and Brain Sciences 19 (1996), pp. 561–610. Hollis, M., ‘Rational Preferences’, The Philosophical Forum 14 (l983): 246–262. James, W., Principles of Psychology (New York: Holt, 1890).

114

GEORGE AINSLIE

Kessler, D. A., ‘Nicotine Addiction in Young People’, New England Journal of Medicine 333 (1995): 186–189. Kirby, K. N. and Herrnstein, R. J., ‘Preference Reversals Due to Myopic Discounting of Delayed Reward’, Psychological Science 6 (1995): 83–89. Malekzadeh, A. R. and Nahavandi, A., ‘Merger Mania: Who wins? Who loses?’ Journal of Business Strategy 8 (1987): 76–79. Marlatt, G. A. and Gordon, J. R., ‘Determinants of Relapse: Implications for the Maintenance of Behavior Change’, in P. O. Davidson and S. M. Davidson (eds.), Behavioral Medicine: Changing Health Lifestyles (Elmsford, N.Y.: Pergamon, 1980), pp. 410–452. Mazur, J. E., ‘Choice Between Single and Multiple Delayed Reinforcers’, Journal of the Experimental Analysis of Behavior 46 (1986): 67–77. Mazur, J. E., ‘An Adjusting Procedure for Studying Delayed Reinforcement’, in M. L. Commons, J. E. Mazur, J. A. Nevin and H. Rachlin (eds.), Quantitative Analyses of Behavior V: The Effect of Delay and of Intervening Events on Reinforcement Value (Hillsdale, NJ: Erlbaum, 1987). McDiarmid, C. G. and Rilling, M. E., ‘Reinforcement Delay and Reinforcement Rate as Determinants of Schedule Preference’, Psychonomic Science 2 (1965): 195–196. Merz, C., The Dry Decade (Seattle: University of Washington, 1930/1968). Millar, A. and Navarick, D. J., ‘Self-control and Choice in Humans: Effects of Video Game Playing as a Positive Reinforcer’, Learning and Motivation 15 (1984): 203–218. Navarick, D. J., ‘Negative Reinforcement and Choice in Humans’, Learning and Motivation 13 (1982): 361–377. Nemiah, J. C., ‘Alexithymia: Theoretical Considerations’, Psychotherapy and Psychosomatics 28 (1977): 199–206. Nestler, E. J., ‘Molecular mMchanisms of Drug Addiction’, Journal of Neuroscience 12 (1992): 2439–2450. Parsian, A. and Cloninger, C. R., ‘Emerging Markers of Predisposition to Alcoholism’, in B. Tabakoff and P. L. Hoffman (eds.), Biological Aspects of Alcoholism (Seattle: Hogrefe and Huber, 1995). Rachlin, H., ‘Self-control: Beyond Commitment’, Behavioral and Brain Sciences 18 (1995): 109–159. Ragotzy, S. P., Blakely, E. and Poling, A., ‘Self-control in Mentally Retarded Adolescents: Choice as a Function of Amount and Delay of Reinforcement’, Journal of the Experimental Analysis of Behavior 49 (1988): 191–199. Ricoeur, P., ‘Guilt, Ethics, and Religion’, in J. Meta (ed.), Moral Evil Under Challenge (New York: Herder and Herder, 1971). Ross, C. A., Kronson, J., Koensgen, S., Barkman, K. et. al., ‘Dissociative Comorbidity in 100 chemically dependent patients’, Hospital and Community Psychiatry 43 (1992): 840–842. Schelling, T. C., The Strategy of Conflict (Cambridge, Mass: Harvard University Press, 1960).

A RESEARCH-BASED THEORY OF ADDICTIVE MOTIVATION

115

Sjoberg, L. and Johnson, T., ‘Trying to Give up Smoking: A Study of Volitional Breakdowns’, Addictive Behaviors 3 (1978): 149–167. Solnick, J., Kannenberg, C., Eckerman, D. and Waller, M., ‘An Experimental Analysis of Impulsivity and Impulse Control in Humans’, Learning and Motivation II, 61–77. Review (1980): 217–225. Sully, J., Outlines of Psychology (New York: Appleton, 1884). Sunstein, C. R., ‘Problems with Rules’, California Law Review 83 (1995): 953– 1030. Tribe, L. H., American Constitutional Law (Mineola, NY: The Foundation Press, 1988), pp. 11–12. VA Medical Center Coatesville, Pennsylvania 19320 U.S.A. (E-mail: [email protected])

A Research-based Theory of Addictive Motivation

This paper examines whether a relatively new line of scientific inquiry may clarify some age-old puzzles about addiction. Although the most productive research on addictions in recent years has been the study of their brain mechanisms, I won't say much about it, because it doesn't do much to change our concepts of these ...

146KB Sizes 0 Downloads 251 Views

Recommend Documents

A Research-based Theory of Addictive Motivation
explained rewards of ingesting, but why do they sometimes buy ways to make ..... nature to a child that the telephone pole down the street is as tall as the one ...

Theory of Human Motivation
in motivation theory as the more basic, unconscious goals. 5. Any motivated ..... under a system which has at least a skeletal outline of rigidity, In which there is a.

A Theory of Human Motivation - Abraham H Maslow - Psychological ...
A Theory of Human Motivation - Abraham H Maslow - Psychological Review Vol 50 No 4 July 1943.pdf. A Theory of Human Motivation - Abraham H Maslow ...

The social motivation theory of autism
an incentive motivation promoting approach seeking and consumption of the reward) [10]. .... confirmed that lack of social support constitutes a major health risk,.

Motivation in Education: Theory, Research, and ...
1 of 1 people found the following review helpful. Little Real-World. ApplicationsBy smtchiroI purchased this book because it was a requirement for my graduate ...