Smoking, Addiction And Decision-making

J Richard Eiser University of Exeter. UK

Smoking, addiction and decisionmaking

There can be few, if any, forms of behavior both as widespread and as damaging to health as is cigarette smoking, yet behavioral medicine research has so far produced only very partial improvements in our understanding of such behavior or of techniques for its modification. When one considers the contributions which psychologists have made, one is by and large looking at specific applications of specific approaches, and the goal of an integration of behavioral and biomedical science knowledge (cf. Gentry, 1981) still seems a long way off. In short, the field still lacks a coherent conceptual framework, and as a result there is a multiplicity of views on which research questions should be given the highest priority. This multiplicity is not in itself a matter for regret, but it makes it difficult for any single paper to represent fairly all current areas of research activity. This paper claims no such representativeness. I shall not consider directly the potential applications of techniques of behavior modification in smokers’ clinics and similar person-to-person encounters between therapists and individual smokers (cf. Raw, 1978). Nor shall I deal here with the extremely important area of prevention of smoking among adolescents (cf. Evans, 1981). The immediate effects of nicotine or deprivation on attention, mood and task performance likewise will not be described specifically. Instead, I shall concentrate on what seems to me to be one of the most critical conceptual contrasts in current research. This contrast is that which arises from two divergent approaches. The first approach, with a strong reliance on physiological evidence concerning smokers’ attempts to regulate their intake of nicotine, emphasizes International Review of Applied Psychology (SAGE, London, Beverly Hills and New Delhi), Vol. 32 (1983). 11-28

12

J Richard Eiser

the addictive nature of cigarette smoking. The second approach, relying more on measures of attitudes and behavioral intentions, emphasizes smokers’ beliefs and decisions as predictors of their behavior. In the context of a broader consideration of the relevance of cognitive social psychology to this field, I shall argue that there is no necessary contradiction between these approaches. To understand how physical dependence affects smokers’ behavior, one must also consider how it affects smokers’ decisions, and if one can understand the decision processes of dependent smokers, this must help our understanding of the concept of addiction itself. Why do people smoke?

To ask why people smoke cigarettes is not to ask a single question. One could concentrate on a variety of more specific questions, all worthy of study in their own right - why do people take up smoking, why do they find it difficult to give up, why do they often relapse after periods of abstinence, why do they smoke the particular brands they do, why do they inhale (or not), why do they keep to a fairly regular level of consumption, why do particular situations make them want to smoke more than others, what are the factors which contribute most to the effectiveness of any cessation treatment? The list could go on and on. The point is simply that smoking is not a single behavior or a single behavioral problem and that approaches which may provide insights in the context of one type of question may be much less relevant to another. To take just one example, the addictive nature of nicotine hardly seems the most obvious starting-point for an explanation of teenagers or younger children accepting their first cigarette. Contrasting sharply with the specificity required by any behavioral analysis is the generality of much of the information communicated to the public in the form of government health warnings, anti-smoking advertisements and the like. Here ‘smoking’ is typically treated as an undifferentiated whole. The degree of effectiveness of most such attempts at dissuasion can be most charitably described as undetectable by the crude evaluation techniques typically employed, and in the light of such disappointments, the question, ‘Why do people smoke?’ often takes on a more plaintive form ‘Why do people carry on smoking when they’ve been told they shouldn’t?’

Smoking, addiction and decision-making

13

The answer I shall suggest to this form of the question is a simple one: ‘Because they decide to’. Clearly, such an answer requires further explanation of what kinds of decisions may be involved. Yet it is no mere truism. The admissibility of any concept of decisionmaking in this context implies that one is dealing with behavior at least in some sense under volitional control. However, to suggest that smoking is in some sense voluntary runs directly counter to what might seem to be implied by describing smoking as an addiction. Before looking at how the notion of decision-making has been treated in cognitive social psychology, and how this might apply to smoking, it is therefore important to look briefly at the argument that smoking is addictive, and what this means.

The concept of addiction The labeling of a behavior as an addiction is not simply a matter of drawing conclusions from pharmacological evidence. It is to take a position which can have a number of repercussions - political, legal, commercial, ethical - which extend well beyond any purely scientific debate. It also places the behavior, in this case cigarette smoking, within a particular medical category - that of dependence disorders (Russell, 1971) - which is itself a subspecialty of psychiatry. Thus Jaffe (1977) has felt entitled to classify cigarette smoking as a ‘mental disorder’. From such a definition, it is potentially a short step to saying that smokers smoke because they are mentally ill, and not responsible for their behavior. Cruder definitions are not uncommon in the context of addiction, and philosophers are fond of quoting the old idea that opium sends one to sleep because of its virtus dormitiva. It is therefore vital to treat the behavioral and pharmacological criteria of addiction separately from questions of harmful consequeqces, if any, and more tendentious notions of diminished choice, responsibility and rationality. On behavioral and pharmacological grounds, there now seems every reason to view cigarette smoking as at least as addictive as the use of many other licit and illicit substances, such as alcohol and opiates. Some of the strongest evidence in support of this position is that which shows that the smoking patterns of established smokers function in such a way as to regulate the level of nicotine in their bloodstream, for example in response to changes in the nicotine

14

J Richard Eiser

yield of cigarettes (Russell, 1977) or to changes in their urinary (and salivary) pH which affects absorption and excretion of nicotine (Schachter et al., 1977). Russell (1977) also points out the extreme efficiency of the manufactured cigarette as a device for selfadministering nicotine: the nicotine from an inhaled puff hits the brain in about seven seconds - faster than the heroin from an intravenous injection. Nicotine itself has been identified as having the properties of a central nervous system stimulant, but, depending on dose and other factors, can also act as a relaxant. There is evidence of the development of tolerance to its effects, as well as of withdrawal symptoms when it is made unavailable. The success rates of smokers who come to clinics for help in giving up are unimpressive (Hunt and Matarazzo, 1973; Raw, 1978) as with other drugs. Cigarette smoking, then, clearly is an addiction. Of course, not all smokers will be addicted to the same extent, and some perhaps not at all, but the same might be said of any drug. For most adult smokers, smoking is not a ‘take it or leave it’ activity. Compared with how most people use alcohol, for example, smokers will feel deprived if they cannot smoke regularly, every day, and throughout the day. To call smoking ar, addiction might seem to imply that smoking is a behavior quite beyond volitional control and hence one to which an analysis of decision processes is irrelevant. This implication is based on an image of the addict as someone completely helpless in the face of his craving and in constant terror of withdrawal. Is this image correct? Even in the archetypical case of heroin, there is reason to suppose it is not. Robins, Davis and Goodwin (1974) conducted an important study of opiate use among US servicemen returning from Vietnam. Included in their sample were 495 whose urines were positive for opiates when they left Vietnam. When interviewed 8-12 months later, only 7 per cent of this sample still showed signs of opiate dependence. Other data from a sample chosen to be representative of the general army population showed that almost half had tried heroin or opium in Vietnam, but less than 1 per cent were addicted to opiates by the time of the follow-up. The myth of life-long addiction as an inevitable consequence of opiate use (at least in the absence of medical intervention) is simply unsupported by such evidence, which suggests rather that one is dealing with a learned response to a more or less specific situation. One factor that might distinguish the street addict from

Smoking, addiction and decision-making

1:

the addicted serviceman, however, may be the lack of any prospect of ever leaving the situation in which drug use has been learned and may at first have been felt to be adaptive. If addictive behaviors may be less than totally involuntary, is there also a sense in which they may be less than totally irrational? Like many other terms that permeate discussions in this area, ‘irrational’ carries clear value connotations. There is a temptation to think that anyone who engages in behavior which has foreseeably potentially terrible consequences must be acting irrationally. As Aronson (1972, p. 9) has put it, however, ‘People who do crazy things are not necessarily crazy’. Rather than dismiss some 40 per cent of the adult population as insane or imbecilic, we should ask ourselves whether the relative unmodifiability of smoking behavior may not be a more predictable outcome of the ways in which human beings make decisions. Social cognition and the question of rationality

Of the many streams of research in cognitive social psychology in recent years two deserve special mention for this discussion. The first is concerned particularly with individuals’ subjective explanations for events and has been organiLed around the various formulations of attribution theory (e.g. Kelley and Michela, 1980). The second is concerned with questions of how individuals make predictions and inferences under conditions of uncertainty, and places particular emphasis on notions such as ‘cognitive heuristics’ (Tversky and Kahneman, 1974; Nisbett and Ross, 1980). Both streams of research share a common interest in the fundamental questions of how individuals simplify complex information about social events to the point where these events become interpretable. In both areas, however, researchers have tended to be rather coy about predicting the behavioral consequences of particular cognitive processes. Kelley and Michela distinguish ‘attribution theories’, which are concerned with how people interpret the causes of events, from ‘attributional theories’ which deal with the consequences of such causal attributions. In a somewhat more apologetic tone, Nisbett and Ross (1980, p. 11) write: ‘we share our field’s inability to bridge the gap between cognition and behavior, a gap that in our opinion is the most serious failing of modern cognitive

16

J Richard Eiser

psychology’. Both of these areas may be distinguished from another tradition, derived from normative models of economic decision-making, which does attempt to relate cognitions and values to behavior, but arguably fails to give an accurate representation of the processes involved at an individual level. It is this tradition which has most influenced psychological definitions of rationality. At its simplest, a decision may be defined as ‘rational’ if it is seen, on balance, as more likely to have good consequences than bad, in comparison to other available options. In other words, the maximization of expected profit is the measure of rationality. Adopting this approach, one can calculate the subjective expected utility (SEU) of any option by taking each identifiable possible consequence of the option, obtaining an evaluation of its desirability to the individual (utility), multiplying this evaluation by its perceived probability, and summing the products over the set of identified consequences. The greater the SEU of a given option, the more likely it should be to be chosen, and in many situations researchers have found support for this prediction. Is cigarette smoking ‘rational’ according to any such definition? Bearing in mind that the SEU approach defines rationality in terms of subjective probabilities and evaluations of outcomes, the fact that some people smoke and others do not might simply be attributable to smokers assigning lower importance and/or probabilities to the risks involved. Mausner and Platt (1971) calculated smokers’ SEUs for stopping smoking and continuing and indeed found some evidence that smokers who had higher SEUs for stopping as compared with continuing were more likely to try to stop. More troublesome for the SEU approach, however, was the finding that smokers’ SEUs for stopping were reliably greater than their SEUs for continuing. In other words, these smokers were prepared to acknowledge that smoking was bad for them, on balance (given a particular way of calculating this ‘balance’), but just seeing it as ‘bad on balance’ was not enough to make them decide to stop. This is consistent with other evidence (e.g. Eiser, Sutton and Wober, 1979) which suggests that, even though smokers may be more sceptical of the health hazards of smoking than non-smokers, they still d o not deny them entirely. Attempts to make sense of such findings whilst preserving the main assumptions of the SEU approach have taken the line that one must be far more specific about the particular kind of decision

Smoking, addiction and decision-making

17

one is considering. In one application of this approach, Eiser and Sutton (1977) looked at variables influencing whether adult smokers would accept or decline the offer of free treatment at a smokers’ clinic, and Sutton (1979) has argued more fully that the choice confronting smokers in such situations is not only whether to continue or to stop smoking, but whether to continue or to try to stop. In other words, the expected benefits of actually stopping have an indirect influence on such decisions; what matters is the expected benefit or cost of trying to stop. Someone who ‘knows for sure’ that he is bound to fail in any attempt at giving up, and also expects any such attempt to involve a great deal of discomfort, would be acting quite ‘rationally’, in terms of SEU theory, if he made no attempt to stop, even though he might acknowledge that the benefits of successfully stopping would be tremendous. When one tries to take account of smokers’ own expectations of success at giving up, the concept of addiction reasserts itself, but under a new guise. Those who feel they are addicted will have a low expectation of success at giving up, and, according to the above line of reasoning, would be less likely to try. What matters here, though, is not addiction as an identifiable physiological state, but addiction as an attribution which people make for their own behavior. It is here that the popular stereotype of the addict, as someone incapable of desisting from drug-taking without the help of medical intervention, can profoundly influence behavior even if it is profoundly inaccurate. If believed, it can be self-fulfilling. Robinson (1972) has discussed some of the paradoxical implications of this concept of the addict or ‘sick role’ (Parsons, 1951) for the treatment of alcoholism. He points to the tendency of practitioners treating alcoholics to demand ‘motivation’ from their clients, in spite of the view that people defined as ‘sick’ are not usually held personally responsible if they fail to make themselves better. In a series of studies, we have investigated some of the factors associated with whether smokers do or do not see themselves as addicted to cigarettes (Eiser, Sutton and Wober, 1977; 1978). Most importantly, those who see themselves as addicted are more likely to think that they would fail if they tried to give up, and are thus less likely to try. They are also more likely to fall into the category which McKennell and Thomas (1967) call ‘dissonant smokers’ specifically those who say that they would like to give up smoking ‘if they could do so easily’. McKennell and Thomas claimed that such individuals would be in a state of ‘cognitive dissonance’ with

18

J Richard Eiser

respect to their smoking, but I have argued elsewhere (Eiser, 1978, 1981), that such a self-attribution of addiction may function as a way of reducing dissonance, in that dissonance, according to theory, should not occur in the absence of perceived freedom of choice. Besides, the image of this large proportion of smokers all desperate to give up but unable to do so is not altogether in tune with their endorsement of the statement ‘I don’t think I’m really prepared to give up smoking if it proves too difficult or distressing’ (Eiser, 1981). Related to the SEU approach is the Fishbein and Ajzen (1975) model of attitudes and behavioral intentions. A central part of this model is the notion that attitudes toward an act are predictable from the salient ’evaluative beliefs’ about that act. As in the SEU model this involves a quasi-economic calculation of expected benefit based on a summation of products of evaluations (utilities) and beliefs (subjective likelihood ratings). A person’s intention to perform the act is then predicted from the person’s attitude toward the act and also from what is termed the ‘subjective norm’ - the person’s impressions of how performance of the act would be evaluated by other people, weighted by how much the person is motivated to comply with such evaluations (again the score is calculated as a summation of products). The actual performance of a behavior - provided the behavior is defined at the appropriate level of specificity - then is typically quite predictable from the measure of intention. Fishbein (1981) describes a number of applications of this model to the issue of smoking behavior. An understanding of a smoker’s subjective norms can contribute considerably to making sense of behavior which might seem ‘irrational’ in terms of expected utilities alone. Fishbein also makes clear that he views his approach not merely as a predictive model, but more generally as a ‘theory of reasoned action’. Applied to smoking, Fishbein argues that smoking can be seen as a behavior under volitional control in that smoking intentions can be quite good predictors of smoking behavior. Decision-making and cognitive simplification It seems, then, that various theoretical approaches - SEU theory, attribution theory, and Fishbein’s theory - can all be applied reasonably successfully to aspects of smoking behavior. Even if one

Srnokinn, addiction and decision-rnakir7.g

19

grants that there are rarely hard-and-fast criteria for interpreting one’s subjects’ responses as evidence of rationality rather than rationalization, at least one is not forced to abandon entirely one’s search for an adequate decision-making analysis. There is a difference, though, between showing, on the one hand, that smokers’ decisions are predictable from their evaluations and expectations of certain perceived consequences and, on the other hand, being able to say why some perceived consequences appear more desirable and/or important than others, and why they are seen as more or less probable. This is not simply an empirical matter of determining the values in a specific instance of weights and variables in a formal model it is a question of the adequacy of the formal model itself as a description of the decision process rather than as a prediction of the outcome of that process. Viewed as descriptions of process, both SEU theory and Fishbein’s model may be seen to make rather stringent assumptions about people’s abilities to combine probabilistic and evaluative judgments in ways compatible with normative principles. Do individuals really sum the products of expectancies and utilities in their heads to come up with an overall preference for each behavioral option before deciding on a course of action? Do they combine their judgments of different consequences together as SEU theory and Fishbein’s model suggest? In fact, in laboratory tests involving choices between hypothetical gambles and similar kinds of stimuli, the basic axioms of SEU/expected value theory have been continually violated (Coombs and Huang, 1976; Kahneman and Tversky, 1979). The focus of this research has been on decision-making involving risk - that is, uncertain outcomes involving possible loss or reduction of profit - and this focus on uncertainty is clearly relevant to decisions involving dangers to health on the one side, and imperfectly effective strategems for changing behavior on the other. In spite of the success of the SEU/expected value approach at making global predictions in many instances, therefore, such predictions may turn out right for the wrong reasons. If we want to describe how smokers arrive at their decisions, a different kind of model may be required. One of the key themes in attempts at finding an alternative model of decision-making is the idea that individuals may rely on simplificatory cognitive strategies - what Tversky and Kahneman (1 974) call ‘heuristics’ - so as to make decisions, as it were, by rule

20

J Richard Eiser

of thumb rather than by working through some kind of internal representation of an economic balance sheet or a statistical equation. Such heuristics can relate to the handling of probabilities and the impact of particular kinds of information on decisions. One line of research relates to people’s tendency to neglect information about base-rate probabilities, when assigning an object to a particular category or assessing the likelihood of an event. In other words, people’s subjective inferences do not typically accord with normative principles such as Bayes’ theorem (Nisbett and Ross, 1980). This has important practical implications when one is presenting probabilistic information about health risks, as, for instance, in the form of mortality statistics, where ordinary people may have only the haziest idea of how many people annually die from any given cause, or from all causes combined (Harding, Eiser and Kristiansen, 1981). Contrasting with this tendency to neglect statistical information is a tendency to rely more heavily on concrete instances, or individual cases (Nisbett and Ross, 1980). Thus the smoker who queries the smoking-cancer link on the grounds that he personally knows one (or more) heavy smokers who have survived to a ripe old age is exhibiting a familiar cognitive strategy used both by smokers and non-smokers across many situations. To label such a cognitive process as irrational or maladaptive in general on the grounds that it can lead to errors in specific instances is to beg many important questions about how far abstract statistical information is to be trusted over more concrete, immediate and (for all one knows) replicable, personal experiences. The question of what kinds of information have most effect on judgment may be narrowed down to the question of what kinds of expected consequences have most effect on decision-making. If individuals are selective in terms of their attention to probabilities, they surely are likely to be selective in terms of their attention to different kinds of outcomes. This might appear to be purely an empirical matter of determining for each decision after it has made the relative importance of different possible criteria. Indeed, procedures have been developed, based on multiple regression, for calculating the relative weights of different attributes of alternative options as predictors of preference on an ad hoc basis (Hammond, Stewart, Brehmer, and Steinmann, 1975). But are we completely in a theoretical vacuum here, without any guidelines from either research or intuition regarding the kinds of consequences which in general will be more influential?

Smoking, addiction and decision-making

21

Just as concrete information may have more influence on decisions than abstract information, so might consequences which are immediate be more influential than consequences which are more remote. In crude terms, this is a reappearance in cognitive terms of the idea that reinforcements are more effective when they are contiguous with the response. Smoking and its perceived consequences

Let us return now to our main question ‘Why do people carry on smoking when they’ve been told they shouldn’t?’ Before writing off such behavior as an example of perversity, stupidity or mental illness, we should ask what impact the familiar health warnings might have at the level of individual decision-making. Let us make a further assumption - that often, if not always, when people are given general information about the effects of particular behavior, they are likely to check out this information for plausibility in the light of other knowledge and personal experience. What form might this checking-out take? The first step might be for smokers to ask themselves the question ‘What is the observable relationship between any stopping or continuing smoking and the occurrence of dire consequences to my health?’ The answer they might give themselves might run like this: ‘Well, honestly, I can’t see that it has very much effect. Maybe my cough is worse when I’ve been smoking heavily, and I get a bit short of breath; but I can’t ever really say “this cigarette’s the one that’s going to kill me - if I smoke this cigarette, it’s going to make me less healthy than if I don’t”. Anyway, everyone has to die of something and you can get killed crossing the road’. Then they might ask themselves ‘What would I expect to happen if I tried to stop?’ The answer might then be: ‘Well, I’ve tried before, remember? It was pretty rotten, wasn’t it? I felt irritable and couldn’t work properly, and I kept nibbling things and putting on weight (which couldn’t have been too good for my health). Sure, if I actually managed to give up for good, that would be great, but that’s a bit irrelevant really as I know I’d start again. Perhaps I’ll feel differently in a few years’ time. Right now, 1 feel I’d be punishing myself if I tried to give up’. I am not attempting to give the above caricature any quasiuniversal status, but hopefully it should serve to illustrate the

22

J Richard Eiser

distinction between the immediate contingencies of the act of smoking any single cigarette and the long-term consequences of continuing or giving up smoking as a habit. The immediate consequences are certain and familiar. The long-term benefits of stopping are based on hearsay, and may be both unreliable and unattainable. This distinction also brings out the question of the difference in urgency between the kind of decisions involved in saying ‘I want a cigarette’ and ‘I want to stop smoking’. This difference is such that the two statements can be made together without any real inconsistency. What is more, the problem is not one of merely delaying gratification - of doing without a small reward now for a large reward later. One does not knowforsure that there will be a large reward in the long term. Essentially, the smoker who gives up is buying an ‘insurance’ against the risks of continued smoking, at the ‘price’ of certain discomfort and expenditure of effort in the short and medium term. The question of how high a ‘price’ a smoker will or should be prepared to pay for such ‘insurance’ is not at all obvious, and depends very much on how relative risks are perceived - although most lung cancer victims are smokers, most smokers do not get lung cancer. The point then is that the immediate and even medium-term reinforcement contingencies all point to the smoker smoking another cigarette (and another, and another.. .). To decide to give up smoking is to decide to act in spite of such contingencies, and this, although not impossible, is very difficult indeed to do. This problem of being trapped by immediate contingencies is an extremely general one. It has parallels in such unlikely areas as interpersonal conflict, and experimental games such as the Prisoner’s Dilemma (Eiser, 1980), where, because of problems of interpersonal trust, individuals have to run risks of short-term loss - and indeed reject a strategy of minimizing loss and maximizing profit on any given trial - in order to achieve the goal of mutual cooperation which is more profitable in the long term. Pruitt and Kimmel(l977) have argued that the achievement of mutual cooperation requires a cognitive change from short to long-range thinking. Broader implications have been discussed by Platt (1973) in his paper on ‘social traps’. Both in applied and laboratory contexts, this literature shows that this longer term goal is very difficult to achieve, even when its ultimate desirability is acknowledged. Considering decisions in this way, one may begin to gain some better idea of the sense in which an addictive behavior may still be

Smoking, addiction and decision-making

23

under volitional control. A smoker may want to give up cigarettes, but the fact that he wants to and has not done so, would seem to imply that it is not the only thing he wants. He also wants a cigarette, and at frequent and regular intervals this want is likely to be the more pressing. What then of the concept of addiction? Too often the concept has been wheeled in to account for nothing more substantial than verbal expressions of regret. To understand addiction, however, it is more important to understand why a smoker says ‘I want a cigarette’ than to try and unravel the epistemological tangles in statements such as ‘I wish I didn’t want cigarettes’. If it were not for the urgent short-term reinforcement contingencies, there would be no need for expressions of regret. Clearly we do not want to get trapped into saying that a heavy smoker is never addicted until he wishes he wasn’t. Regret, or ‘dissonance’ in the McKennell and Thomas (1967) sense, therefore, cannot be a criterion of addiction, even though it may be an effect. Addiction and reinforcement

The concept of addiction here proposed is very simple - in learning theory terms, an addictive behavior is one which is associated with strong short-term positive reinforcement contingencies. In other words, it is something that, at the time a person very much wants to do, and/or very much does not want to do without, whatever the longer-term consequences may be. To say a person needs a great deal of determination and/or help to overcome an addiction is to say no more and no less than that it is extremely difficult to act deliberately, or be made to act, in a way that runs counter to overlearned reinforcement contingencies. Does this mean that any reinforcing activity can be addictive, whether or not it is drug-related? In principle, there is no reason why not. Over-eating in the case of obese people may be one example. Compulsive gamblers may talk about their gambling much like alcoholics talk about their drinking. Television watching, computer games such as ‘Space Invaders’, and a variety of recreational activities might appear to take on a similarly obsessional character for some people at 1east.To try to confine one’s concept of addiction to activities which produce psychological effects of certain kinds does not help very much when so little is known about the

24

J Richard Eiser

psychophysiological effects of these other activities. Likewise, to reserve the concept for activities which are clearly damaging to health may be misleading, since even within the fields of drugs, some substances might be damaging without being addictive and others addictive without being damaging. In fact, it could be argued that cigarettes illustrate this point - put crudely, the nicotine is what makes cigarettes addictive, but the tar is what leads to lung cancer (though the nicotine may be associated with other health risks). On the other hand if no bounds are set on the applicability of the concept, it ceases to explain anything at all. As I argued earlier, calling something an addiction has a variety of political and other implications, for example for legislative control or prohibition designed to ‘protect people from themselves’. Addiction is not a label which should be lightly attached to any activity one regards as a problem. I personally would be happy with a definition according to which drug-related compulsive behaviors were considered as addictions in a literal sense and other compulsions as addictions only in a metaphorical sense. This is still arbitrary, since, from a psychological point of view, what matters is the pattern of reinforcement contingencies associated with the behavior, rather than the particular kind of substance involved. Of course, without an understanding of the psychopharmacological effects of different substances one could never, for instance, explain why smoking tobacco can be a pleasure but smoking lettuce leaves is at best a penance. Such reinforcement contingencies rely largely (though maybe not totally) on the psychopharmacological effects. Although in principle any behavior might be addictive in either the literal or metaphorical sense, in fact there may be very definite constraints on what behaviors can become so. Little, if anything, is known about such constraints, but the literature on biological constraints on learning (Hinde and Stevenson-Hinde, 1973; Seligman and Hager, 1972) points to definite limitations to the kinds of behaviors that may be learned with particular kinds of reinforcements. It would be the height of naikety to assert, on the basis of current knowledge, that there is nothing special about drugs as reinforcers or about drug-taking as learned behavior.

Smoking, addiction and decision-making

25

Conclusions

Cigarette smoking thus provides both an important and instructive example of where a behavioral medicine approach can achieve an integration of biomedical and behavioral science perspectives. For too long, biomedically oriented research has investigated addictions to specific substances as specific physical conditions, and an impression may have been created that, by describing the pharmacological (and even pathological) effects of a drug, one has said all that needs to be said about the psychological and behavioral effects. This emphasis has contributed to a view of addiction as a physical state in which the individual is completely unable to help himself - a view which may not only be at variance with the facts but damaging to prospects of prevention or treatment. The argument put forward in this paper - that smoking is still, in a particular sense, under volition control and the product of a kind of decision-making, hopefully may provide some counterbalance. For this argument to be informative, however, it has been necessary to examine what kind of decision-making may be involved, and how various cognitive processes and biases may lead smokers to make the decisions which they do. Also, to say that smokers decide to smoke is not to say that they could easily decide to do otherwise. On the contrary, it would seem to be extremely difficult for people to escape from the trap of short-term reinforcement contingencies - to avoid letting their decisions be dominated by immediate appetites, cravings or anxieties, and to start to aim for a long-term goal with little certainty of success. Escape from addiction is not simply a matter of detoxification or endurance of withdrawal symptoms - it requires a cognitive shift from short to long-range thinking, even if the ‘long-range’ goal is no more than merely abstinence for one day at a time. Such a cognitive view, however, may itself lead to misconceptions if the biomedical evidence is ignored. It is simply untrue that any kind of cognitive or attitudinal manipulation, designed, say, to change people’s perceptions of the consequences of smoking will be as effective as any other in producing behavior change. Quite apart from questions of credibility, individuals may be highly selective in the kinds of factors they take into consideration when making decisions, and this selectivity is not always taken full account of even in some of the more successful psychological models of attitudes and decisions. Just as the biomedical approach cannot afford to ignore

J Richard Eiser

26

similarities between addictions and other learned behaviors, so too must the cognitive approach avoid the pitfall of denying that drugs may have special effects, and these effects may have special implications for decision-making.

References Aronson. E. (1972). The social animal. San Francisco: Freeman. Coombs, C.H. and Huang, L.C. (1976). ‘Tests of the betweenness property of expected utility’, Journal of Mathematical Psychology, 13, 323-337. Eiser, J.R. (1978). ‘Discrepancy, dissonance and the “dissonant” smoker’, Internarional Journal of the Addictions, 13, 1295-1305. Eiser, J.R. ( I 980). Cognitive socialpsychology: A guidebook to fheory and research. London: McGraw-Hill. Eiser, J.R. (1981). ‘Addiction as attribution: cognitive processes in giving up smoking’. In J.R. Eiser (ed.) Socialpsychology ond behavioral medicine. Chichester: Wiley. Eiser, J.R. and Sutton, S.R. (1977). ‘Smoking as a subjectively rational choice’, Addictive Behaviors, 2, 129-134. Eiser, J.R., Sutton, S.R. and Wober, M. (1977). ‘Smokers, non-smokers, and the attribution of addiction’, Brifish Journal of Social and Clinical Psychology, 16, 329-336. Eiser, J.R., Sutton, S.R. and Wober, M. (1978) ‘ “Consonant” and “dissonant” smokers and the self-attribution of addiction’, Addictive Behaviors, 3, 99-106. Eiser, J.R., Sutton, S.R. and Wober, M. (1979). ‘Smoking, seat-belts, and beliefs about health’, Addictive Behaviors, 4, 331-338. Evans, R.I. (1981) ‘Training social psychologists in behavioral medicine research’. In J.R. Eiser (ed.) Social psychology and behavioral medicine. Chichester: Wiley. Fishbein, M. (1981). ‘Social psychological analysis of smoking behavior’. in J.R. Eiser (ed.), Social psychobgy and behavioral medicine. Chichester: Wiley. Fishbein, M. and Ajzen, 1. (1975). Belief, attitude. infention and behavior: An infroducrion to theory and research. Reading, Mass.: Addison-Wesley. Gentry, W.D. (1981). ‘What is behavioral medicine?’ in J.R. Eiser (ed.) Social psychology and behavioral medicine. Chichester: Wiley. Hammond, K.R., Stewart, T.R., Brehmer. B. and Steinmann, D.O. (1975). ‘Social-judgment theory’. In M.F. Kaplan and S. Schwartz (eds) Human judgment and decision processes. New York: Academic Press. Harding, C.M.,Eiser, J.R., and Kristiansen, C.M. (1981). ‘The representation of mortality statistics and the perceived importance of causes of death’.

SmokinR, addiction and decision-making

27

Manuscript submitted for publication. Hinde, R.A. and Stevenson-Hinde, J. (eds) (1973). Consfrainfson learning. London: Academic Press. Hunt, W.A. and Matarazzo, J.D. (1973). ‘Three years later: Recent developments in the modification of smoking behavior’, Journal of Abnormal Psychology, 81, 107- 1 14. Jaffe, J.H. (1977). ‘Tobacco use as a mental disorder: The rediscovery of a medical problem.’ In M.E. Jarvis et al. (eds), Research on smoking behavior. Rockville, Md., Dept of Health Education and Welfare: National Institute on Drug Abuse. Kahneman, D. and Tversky, A. (1979). ‘Prospect theory: an analysis of decision under risk’, Econometrica, 47, 263-291. Kelley, H.H. and Michela, J.L. (1980). ‘Attribution theory and research’, Annual Review of Psychology, 3 1, 457-501. Mausner, B. and Platt, E.S. (1971). Smoking: a behavioral analysis. New York: Pergamon. McKennell, A.C. and Thomas, R.K. (1967). Adults’andadolescents’smoking habits and attitudes. Government Social Survey, London, HMSO. Nisbett, R.E. and Ross, L. (1 980). Human inference: Strategies andshortcomings of social judgemenf. Englewood Cliffs, N.J., Prentice-Hall. Parsons, T. (1951). The social system. Chicago: Free Press. Platt, J. (1973). ‘Social traps’, American Psychologisf, 28, 641-651. Pruitt, D.G. and Kimmel, M.J. (1977). ‘Twenty years of experimental gaming: Critique, synthesis, and suggestions for the future’, Annual Review of PSyChOlOgy, 28, 363-392. Raw, M. (1978). ‘The treatment of cigarette dependence’. In Y. Israel et al. (eds) Research advances in alcohol and drug problems. Vol. 4, New York: Plenum. Robins, L.N., Davis, D.H. and Goodwin, D.W. (1974). ‘Drug use by U.S.Army enlisted men in Vietnam: A follow-up on their return home’, American Journal of Epidemiology, 99, 235-249. Robinson, D. (1 972) ‘The alcohologist’s addiction: Some implication of having lost control over the disease concept of alcoholism’, Quarferly Journal of Srudies on Alcohol, 33, 1028-1042. Russell, M.A.H. (1971). ‘Cigarette smoking: Natural history of a dependence disorder’, British Journal of Medical Psychology, 44. 1-16. Russell, M.A.H. (1977).‘Smoking problems: An overview’. In M.E. Jarvik et al. (eds) Research on smoking behavior. NIDA Research Monograph 17. Rockville, Md., DHEW. Schachter, S., Silverstein, B., Kozlowski, L.T., Perlick, D., Herman, C.P., and Liebling, B. (1977). ‘Studies of the interaction of psychological and pharmacological determinants of smoking’, Journal of Experimental Psychology: General, 106, 3-40. Seligman, M.E.P. and Hager, J.L. (eds) (1972). Biological boundaries of learning. New York: Appleton-Century-Crofts. Sutton, S.R. (1979). ‘Can subjective expected utility (SEU) theory explain smokers’ decisions to try to stop smoking?’ In D.J. Oborne, M.M. Gruneberg, and J.R. Eiser (eds) Research in psychology and medicine: Vol. 2. Social aspecfs: attifudes, communication, care and training, pp. 94-101. London: Academic Press.

28

J Richard Eiser

Tversky, A. and Kahneman, D. (1974). ‘Judgment under uncertainty: Heuristics and biases’, Science, 185, 1124-1131.

Furner, depmdance et prise de dkision

Le comportement des fumeurs est examine a la lumiere des recherches faites en psychologie sociale cognitive. L’idte difendue est que les fumeurs adultes continuent B fumer probablement parce que c’est la conskquence d’un processus de dtcision plut6t qu’un fait indtpendant de tout contr6le volontaire de I’individu. Les implications du concept de dtpendance sont discuttes ici particulittrement en relation avec les concepts de rationalitt et de renforcement. La conclusion est la suivante: pour comprendre le comportement de fumeur, i! faut inttgrer les approches biomtdicales et psychologiques.