INDULGENCE, EXCESS, AND RESTRAINT:
PERSPECTIVES ON CONSUMMATORY BEHAVIOR IN EVERYDAY LIFE

C. Peter Herman* Lynn T. Kozlowski*

A regulatory model of consumption is outlined, with an emphasis on upper and lower regulatory limits which govern consumption. We argue that aversive physiological symptoms counteract failure to maintain consumption within regulatory limits, but that consumption within these limits is affected by various situational and cognitive factors. This model is applied to the consumption of food, alcohol, tobacco, and caffeine, with particular attention to the distinctive features of regulation that pertain to each. Implications are derived regarding the multiple meanings of "excessive " and 'insufficient " consumption in this framework. We also discuss the relative (perceived) contributions of physiological and "personal "mechanisms of regulatory control and ensuing effects of self-perception and attribution. Concluding remarks concern the complexities of multiple substance use.

*C. Peter Herman is Associate Professor of Psychology, University of Toronto. Scientist in the Clinical Institute, Addiction Research Foundation of Ontario. Associate Editor of the Journal of Personality, his research interests including eating, dieting, and physique.

* Lynn T. Kozlowski is Assistant Professor of Psychology, Wesleyan University, Middletown, Connecticut. As of July, 1979, Scientist in the Clinical Institute, Addiction Research Foundation of Ontario. A contributor to the U.S. Surgeon-General's 1979 Report on Smoking and Health, his research interests include use of common drugs, particularly nicotine and caffeine.

Perusal of the current best-seller list will convince the reader that weight control is a problem in our society; perusal of back issues will convince the reader that it's a chronic problem. Similarly, we are increasingly aware that excessive use (i.e. abuse) of tobacco, alcohol, and even caffeine is a continuing challenge to society as a whole and to the individual abusers in particular. However, without denying the costs of excessive consumption, we must nevertheless recognize that our notions of excess are themselves problematical. The purpose of this essay is to explore the concept of excess, along with some related concepts (e.g. sufficiency and deficiency) as they apply to substance use/abuse in everyday life.

In general, we shall attempt to explore the questions of too little /enough /too much in a comprehensive framework of regulation. That is, we shall assume that for each sort of substance, the individual ordinarily maintains consumption within certain regulatory bounds. It should be noted at this point that while some theorists have conceptualized regulation to occur at a particular level (e.g., the heroin addict strives to maintain a particular concentration of the drug in his veins), we are arguing that regulation occurs within the limits imposed by two regulatory levels (upper and lower), and that within the range of regulation the degree of regulatory control is often very imprecise. Rather than propose that the cigarette smoker, say, attempts to achieve a precise blood-nicotine level (as Jarvik, 1973, and Schachter and his associates, 1977, have done), we propose that the smoker attempts to achieve a blood-nicotine level that falls between the lower and upper regulatory bounds. Within the range of regulation, however, the blood-nicotine level may vary freely, without inducing the sort of constantly corrective regulatory behavior that characterizes Schachter et al.'s (1977) "machine "-like smoker. The immediate advantage of our "range" assumption over the traditional "level" assumption is that it is much more flexible, conceptually, and more accurate, empirically. Theorists such as Schachter have been plagued by the intra-individual variability of consumption; adoption of a "range" concept allows for the evident failure of drug users to titrate their doses precisely.

While smokers (and drinkers and eaters) may not regulate their intake with ,)recision within the regulatory range, they do show dramatic regulatory (i.e. corrective) behavior when the limits of this range are reached-or breached. If intake falls below the amount needed to surpass the lower bound of the range of regulation, then a true insufficiency exists. Such a state of deprivation will engender attempts-sometimes frantic-to provide enough of the appropriate substance so as to eliminate the aversive effects (e.g., withdrawal symptoms) that prompt drug-taking in this range of insufficiency (or lower range of aversive control).

On the other hand, if the user manages to consume so much that he surpasses the upper regulatory bound, his excessive consumption will expose him to the corrective forces that apply in the range of excess (upper range of aversive control). Nausea is probably the most familiar symptom of this range; we shall delay a more detailed explication until we deal with specific substances. The important points, however, are:

(a) that consumption outside the range of regulation is inappropriate, and usually-but not necessarily-creates aversive symptoms that induce corrective consummatory behavior.

(b) within the limits of regulation, consummatory fluctuations do not ordinarily involve strong aversive control (negative feedback).

(c) the location of the upper and lower regulatory bounds is an empirical question, and a major exploratory challenge for psychopharmacology.

With regard to point (c), we recognize a vast array of issues, including whether upper and lower regulatory bounds invariably exist; how such bounds are acquired, if they are not "built in"; whether the bounds shift, and if so, when and how; whether the bounds can shift independently of one another, and what factors determine the "distance" between upper and lower bounds; whether there are individual differences in the size of the range of regulation, and what factors are responsible for such individual differences.

In the brief space allowed us-and admitting our ignorance with regard to many of these questions-we cannot hope to do more than indicate how our framework helps to organize what we do know and to indicate what we need to learn. Let us see how the regulatory framework applies to particular areas of substance use and abuse.

The Regulation of Food Intake:

As perhaps befits this most essential consummatory activity, the regulation of food intake is both powerfully motivated and intricate. Complexities derive from the fact that there are most likely different, overlapping regulatory structures, corresponding to short-term and long-term nutritional needs and capacities. An individual who has been starving for a number of days may thus be deficient in the long-term regulatory system; his short-term regulatory system, however, is unlikely to permit an immediate correction for a long-term deficit. It is possible, then, for such an individual, after eating, to be simultaneously hungry (below the lower limit of long-term regulation) and sated (at or above the upper limit of short-term regulation). These different regulatory systems, of course, exert some mutual influence; for instance, a chronically starved organism is likely to develop an exaggerated positive response to good-tasting, high calorie foods (Nisbett, 1972), with the result that short-term "appetite" is increased and the animal will continue to approach the upper limit of short-term regulation during meals. The net effect, of course, is to provide a gradual correction for the long-term deficit, through an accumulating series of positive energy balances at meals. As the long-term deficit is eliminated, the relative attractiveness of good-tasting foods declines to "normal" levels, and the frequency of eating to capacity at meals is reduced. The system, then, is homeostatic, in a complex way.

But doesn't one always eat to capacity at meals? Or, at least, don't animals given ad lib access to palatable food stop eating only when satiety mechanisms are activated? These questions point to further subtleties in the regulatory process. It appears to be the case, for instance, that satiety does not reflect a precise state or level of intake, but rather a somewhat flexible process operating within a narrow range (band) at the upper end of the range of regulation. The activation of satiety mechanisms is conditioned by a number of factors beyond simple caloric intake; put another way, the caloric input needed to trigger satiety processes shifts with circumstance. The well-known Coolidge effect, for instance, testifies to the ability of anovel sexual stimulus to elicit sexual activity in apparently sated animals (Bermant & Davidson, 1974). It is likely that the palatability of food as well declines as a function of repetition, and that novelty may well serve as the spice of life in a literal sense, increasing palatability and, as a consequence, raising the threshold of satiety. The exact mechanisms of such satiety shifts remain largely unknown.

The importance of the foregoing examples is to illustrate the advantages of conceptualizing regulatory processes in terms of regulatory ranges and boundaries. It is often noted that despite relatively wild fluctuations in daily intake, individuals tend to regulate their body weight with great precision (Beller, 1977). Our conceptual scheme does not "account" for this alleged anomaly, but it helps to organize our thinking. First, it seems clear that the width of the range of regulation is much broader for short-term than for long-term consumption. Perhaps a better formulation is that what is being regulated in the short- and long-terms is quite different: in the former case, we regulate caloric intake, or some correlate thereof; in the latter case, we regulate weight, or a correlate.

The apparent imprecision of caloric regulation in the short run is attributable to a number of factors. At the lower end, we are capable of going without food for a considerable period of time, and the degree of "hunger" required to initiate feeding is quite variable (i.e. responsive to circumstances). Blood sugar level, for instance, depends not simply on the caloric content of the most recent meal, and the time elapsed since that meal, but also on a variety of processes tied to learning and perception. Powley (1977) has recently described the spectrum of cephalic phase (anticipatory ingestive) responses that precede eating; these physiological "preparations" can in some sense produce hunger (as in hyperphagic animals with ventromedial hypothalamic lesions, the effect of which, according to Powley, is to exaggerate cephalic phase responses, and thus to actually increase hunger). One interesting implication of Powley's model is that mere perception of appetizing food, by stimulating anticipatory reactions (e.g., insulin release) can render the organism hungry; in short, the availability or salience of food may not only permit eating in a hungry animal, but may induce hunger-based eating in an animal not otherwise hungry. This sort of anticipatory hunger, furthermore, is probably conditionable, so that regular eating patterns, and consequent regular anticipations, become self-fulfilling and self-sustaining.

Food differs from other substances subject to abuse, of course, in that there is an innate lower bound of regulation. Hunger appears to be one of the first sensations that the neonate experiences, whereas the demand for other abuse-prone substances ordinarily must be acquired. Truisms such as the foregoing, however, occasionally justify their appearance by suggesting possible limits to their accuracy. The recent explosion of work on the endorphins, for instance, (Snyder, 1977), and the attendant suggestion that morphine-induced endorphin depletion may be responsible for the ravages of withdrawal from heroin, is premised on the notion, not always articulated, that a certain minimal level of endorphin is required naturally. Chemical analogues of other drugs of addiction, if they are discovered, will perhaps force us to consider the possibility that drug addictions are acquired only in the sense that we acquire a dependence on external rather than internal sources of the drug.

On the other hand, the adolescent girls who suffer from anorexia nervosa may present a picture of a range of food regulation that has no lower bound (and a very low upper bound). While it is frequently argued that the absence of a lower bound (i.e. hunger) is due to a neuroendocrinological insult of some sort, most-though not all--of the evidence indicates that the physiological correlates of anorexia are probably interpretable as effects rather than causes of starvation (Bemis, 1978). Other theorists (e.g. Bruch, 1978) argue that the lower bound is in fact present in anorexics, but that it is simply denied; other motives become more important than hunger, but the hunger is there nevertheless. Whatever may be the truth in this matter, it is clear that hunger itself is not a simple construct. Nisbett (1972), for instance, has argued that actuarially obese humans who constantly insist that they are "legitimately" hungry may well be correct. Nisbett suggests that the lower regulatory bound in the longterm hunger system is "set" by various physiological parameters (e.g. adipocyte number) that vary markedly within the population. Persons with a high set-point (lower bound) for weight may thus be below their long-term lower bound, even though they are statistically obese. Such deprivation in the long-term system may condition the short-term system, raising its lower bound so that the individual almost always experiences hunger. Nisbett has pointed to the similarities between the obese person described here and the starving individual described at the beginning of this section. To explain why any individual would suppress his or her weight down below the lower regulatory bound, thereby incurring the aversive symptoms associated with chronic nutritional deprivation, requires that we invoke forces that can override the aversive consequences of poor regulation. In the case of obese individuals, one can identify pressures to diet-based on arbitrary cultural norms of fashion as much as on superficial medical advice-that operate on the obese as pervasively (though not always as successfully) as they do on anorexics.

To return to our general conceptual scheme, we can identify a number of questions (and a smaller number of answers) that pertain to eating, food, and weight regulation. First, there is a generally acknowledged lower bound of regulation, below which one experiences aversive symptoms. For short-term regulation, the symptoms include most prominently hunger, and also fatigue, irritability, and other negative states quickly alleviated by eating. Between the elimination of hunger and the onset of satiety, there is a fairly wide latitude-the width depending on constitutional (reflexive) and conditioned factors--in which "non -regulatory" factors may obtain. The quality (appetizingness) of the food available, for example, will to a large extent determine whether a minimal (hunger-stopping) or maximal (satiety starting) amount is consumed. At the upper bound, consumption is inhibited by satiety mechanisms (bloating, temporary aversions) that become more powerful as one consumes "beyond" the upper bound, into the range of excess. (The absence or weakness of such satiety mechanisms in certain individuals is often regarded as a prime etiologic factor in obesity; as we have seen above, though, long-term hunger, even in the obese, may raise the threshold of excess). Inhibitions, however, do not appear only at the upper bound of the range of regulation. Individuals may restrict their intake in a deliberate fashion well before satiety is reached. Such dietary restraint (see Herman, 1978, for a review) is evident in normal weight dieters, the obese, and most dramatically, anorexics, who may act "sated" even while below the lower regulatory bound.

Such self-imposed restraints on eating bear an interesting relation to the "automatic" restraints represented by satiety mechanisms (inhibitions activated by approaching or achieving excess); the issues here extend well beyond regulation per se into the realm of self-perception and attribution. It is not always clear, for instance, whether any particular failure to indulge is attributable to self-control of the colloquial variety (i.e. the sort of self-control that merits personal praise, or occasionally blame) or whether the lack of indulgence is mandated physiologically. As we have seen, the various complexities associated with the boundaries of deprivation and excess (including individual differences) create a correspondingly complex problem of causal attribution; are the "decisions" to eat or not to eat decisions of the mindwith implications for perception and self-perception of "character"-or are they decisions of the body, with implications for neither admiration or deprecation? La Rouchefoucauld cynically maintains that "If we resist our passions, it is due more to their weakness than to our strength", suggesting that we give ourselves personal credit for our physiological accomplishments. The issue, however, is not a clear-cut one, and the interplay of deliberate and automatic controls remains a fascinating and crucial. question for psychology. The question of control becomes paramount in anorexia, where it is often the case that patients claim that they can't eat, while therapists are typically convinced that they "simply" won't eat. The patient, in short, denies the personal exertion of deliberate control, while the therapist sees such exertion of control as the central clinical issue, with the control of food intake (and weight) becoming both so strict and so crucial that it becomes, in effect, all-consuming and out of control. In sum, the complexities of the regulatory system are no more intricate than are those of the personalities in which they operate.

Our discussion of anorexia, of course, raises novel questions about the notion of excess. What is excessive, when it comes to eating? Obviously, excess is that level of consumption that surpasses a certain limit; but there are a whole spectrum of limits. For the anorexic, excess may be anything more than the amount (or weight) necessary to sustain life; and for some, even that is too much. Excess may be anything beyond the amount needed to assuage hunger; or it may be that amount that triggers satiety mechanisms; or it may be an intermediate amount dictated by "diet calculations" of greater or lesser arbitrariness. Schachter (1971) has argued that obese individuals may simply not have salient limits (of either excess or deprivation), whereas normals tend to respond principally in terms of lower (hunger) and upper (satiety) regulatory limits. Others (such as Nisbett) have argued that the limits are there, but set at aberrant levels. The specification of excess, in any case, is not a simple matter, and raises a host of empirical questions that are at present not only unanswered but rarely even articulated.

The Regulation of Alcohol Intake:

Alcohol, like food, is a substance subject to abuse; on balance, however, the differences between alcohol and food regulation are at least as striking as the similarities. Most prominent among the differences is the evident absence of a lowerregulatory bound for most drinkers; there is no inherent lower limit to consumption, below which aversive (withdrawal) symptoms appear. Alcoholics, while not diagnosed solely on the basis of the acquisition of a lower limit (i.e. physical dependence), are principally distinguishable from normal or non-drinkers in that they must maintain a certain minimum level of alcohol in order to prevent negative physiological repercussions. But even the most "predispositional" theories of alcoholism (c.f. Goodwin, 1976) insist that the lower regulatory boundary is set on the basis of some exposure to alcohol, in contrast to the innate lower limit for food. Another point of contrast is that it seems unlikely that there are specifically longterm regulatory processes for alcohol. The regulation of alcohol, to the extent that it appears at all, appears as a unitary process.

The regulation of alcohol, though, while perhaps unitary, is certainly not uniform. The upper regulatory bounds, for instance, are remarkable for their variability. Individuals appear to differ initially in their capacity for alcohol; and the same individual, depending on his experience with alcohol, may or may not develop tolerance (i.e. an elevated upper and/or lower bound). For most drinkers, there is clearly a wide latitude between initiation of drinking and onset of satiety (impending or actual excess) cues. The factors that determine the amount of alcohol consumed up to the point of satiety or excess, are at least as variable as those that apply to food: availability / opportunity, cost/effort, palatability, stress, direct positive effects, enabling effects, and so on.Enough alcohol is usually defined in terms of the individual's subjective state, the euphoria or relaxation that prompted drinking initially.Excess consumption has a number of different meanings, unusually well -articulated owing to alcohol's status as a recognized public health problem. First, there are the short-term physical effects that signify too much: loss of motor control, dysfluencies of speech, gastric distress, stupor, or worse. The hangover may be regarded as a somewhat delayed member of this category. Next, too much can be defined more behaviorally in terms of its consequences involving other persons and activities: impaired driving and interpersonal belligerence are familiar examples of the consequences of excess so defined. It should be noted, however, that the drinker may exhibit physical and social impairments without encountering feedback from an internal regulatory sensor; in short, the effect of "excessive consumption" may appear before satiety is reached. This consideration applies even more strongly with respect to a third sort of excess: namely, excess defined in terms of long-term medical sequelae of alcohol ingestion. It appears that the effects of drinking on the liver, brain, and heart are essentially cumulative, and vary as a function of total consumption, whether that consumption is achieved by chronic mild drinking (not excessive by the previous standards) or by more infrequent, "excessive" bouts of drinking (Lelbach, 1974).

We cannot hope to do justice, of course, to all the complexities of alcohol regulation and excess. Indeed, for most drinkers, the notion of regulation is somewhat inapplicable, except with respect to upper limits. A similar problem of substance abuse, however, is that of tobacco. Like alcohol, tobacco must be experienced before a lower bound of regulation emerges. Further, it is arguable that all tobacco use-like all alcohol use--constitutes abuse. Unlike alcohol, however, tobacco seems to engender physical dependence in a majority of its users. Let us examine the intricacies of tobacco regulation.

The Regulation of Tobacco Intake:

Tobacco is a complex substance, composed of many agents that no doubt contribute to its special appeal, but the drug that is most likely at the root of the pharmacological attractions of tobacco is nicotine. Indeed, it has been argued that cigarette smoking for most smokers is supported by an outright addiction to nicotine (Russell, 1976).

Virtually everyone who has smoked even a single cigarette has experienced the rigors of the upper boundary of nicotine regulation. Nicotine--a deadly "nerve" poison favored by mystery writers-is highly toxic. Thus, fledgling smokers who take a few "extra" puffs on a strong cigarette may find themselves dizzy, sweating, nauseous, and generally feeling as if they were at death's door. As is the case with alcohol, individuals differ initially in their capacity and later in their tolerance (and rate of acquiring tolerance) to nicotine. Indeed, differences in nicotine sensitivityin conjunction with social pressures to become smokers-can influence who is recruited successfully to the ranks of smokers (Kozlowski and Harford, 1976). Veteran smokers may be describing what it is like to encounter the upper boundary of regulation when they complain of being "smoked out" and are unable to smoke comfortably for a few minutes (or hours). Party-going smokers seem to be as likely to suffer from over-smoking the night before as from over-drinking. In fact, it may be the case that normal brakes on consumption other than nicotine levels them selves-inhi bi tors of smoking such as harsh taste of tobacco smoke-are to some extent eliminated by alcohol's "soothing" effects oil the oral cavity, with the results that the smoker who drinks to excess is also likely to smoke to excess, although the penalty may not become evident until the morning after.

The lower boundary-and the withdrawal syndrome that defines it-develops after exposure to nicotine, but there is no agreement as to how much exposure is required before the smoker becomes dependent; nor is there agreement as to what factors are likely to determine exactly what the minimal exposure period is. The questions regarding the development of physical dependence take on a special urgency in view of the fact that the majority-by Russell's calculation, three out of four-of people who smoke even a few cigarettes go on to become dependent smokers.

The nicotine withdrawal syndrome itself is far from clear-cut (see Jaffe and Jarvik, 1978, for a recent review). It usually involves craving, sleep disturbances, gastrointestinal upset, irritability, and sometimes headache; these symptoms, however, usually appear only after a substantial period without smoking. Factors that control the severity and timing of withdrawal symptoms are for the most part merely matters of speculation. It seems likely, though, that incipient withdrawal symptoms may be responsible for the sorts of subtle behavioral regulation that occur at the lower regulatory boundary, in smokers who are in no way attempting to quit smoking but who may be trying to keep their smoking down to minimal levels or who may be smoking relatively meager amounts not intentionally but for reasons of opportunity, preoccupation, and so on. Nicotine has a short plasma half-life, less than 30 minutes (Isaac and Rand, 1972) so that a smoker who is attempting to maintain sufficient plasma levels of nicotine has to be smoking much of the time that he is awake.

Much of the research on cigarette dependency has focused on the issue of nicotine regulation. There is not enough space here to consider this literature in detail (see Russell, 1976, and Schachter, 1978, for reviews), but the application of the present regulation model to some of the perplexities of this work is in order. Though nicotine is the drug that we have most in mind, we shall frame the discussion more generally. Those drugs (or individuals) that show only a small gap between the upper and lower boundaries of biologically -based drug regulation will show superb titration. Those with large gaps between the boundaries may or may not show well-defined regulatory effects-depending on the fluctuations in intake reflecting the influence of the many non-biological factors operating within the gap (the range of indifference). The range of indifference, then, probably ought to be designated the range of indifference to the pharmacological properties of the drug. Consumption within this range is not necessarily random or unmotivated; rather, it is influenced by psychological and situational factors that are not especially linked to the biochemistry of the drug being used. Such factors are evident in the case of the individual who is just beginning to smoke, who is not fearful of the health consequences of smoking, and who is trying to smoke at a rate consistent with that of his heavily smoking friends; such a person would be likely to be close to the upper boundary of regulation much of the time. On the other hand, countervailing social and psychological pressures can lead a smoker to try to "diet," to restrain his intake as much as possible, and, as a result, to maintain a level of consumption that is quite close to the lower boundary of regulation.

If a drug is sought when intake drops below some lower boundary (deprivation) and is avoided when some upper boundary is surpassed (excess), then a biologically based regulation of drug intake can be said to occur. Of course, all drugs can be given in toxic doses and therefore in some sense have an upper boundary to their use; but the upper boundaries that are of interest to us are those that pertain to the ordinary use of the drug and that normally operate to limit consumption by the habitual drug taker. To avoid a logical circularity regarding the influence of upper and lower boundaries on drug regulation; indeed, to determine if the pattern of drug intake that is observed is truly regulatory with respect to such boundaries, it is necessary to determine relatively independently whether a given individual has an upper and lower boundary of intake and to determine the size of the gap (zone of indifference) between them. The question of the distance between the boundaries is essentially an issue of the proper metric with which to evaluate someone's drug consumption. For example, it is possible that for one smoker, 20 cigarettes worth of nicotine (distributed appropriately over the waking day) will prevent withdrawal, and up to 40 cigarettes can be taken without short-term aversive consequences; another smoker, however, may have a range extending from 20 to 30 cigarettes. It seems quite likely that the biological significance of a 510 cigarette variation in consumption will be considerably different for these two individuals.

An additional complexity of measurement stems from the nonequivalence of the units "cigarettes smoked" and "blood nicotine level," even after correcting for the variable nicotine content of the available cigarettes. It must be kept in mind that it is the precise parameters of puffing (puff volume, duration, and number) along with characteristics of the cigarette (especially its nicotine content) that determine the ultimate nicotine level in the bloodstream. The number of cigarettes smoked per day can be an extremely inadequate measure of nicotine intake. Ashton and Watson (1970) have shown that under some circumstances smokers can compensate exquisitely for the variation in filter efficiency in cigarettes, and hence maintain a constant nicotine dose. Further, interacting with behavioral (puffing) variations are a range of essentially constitutional factors, both acute and chronic. For instance, urinary pH is known to influence the excretion rate of nicotine from the bloodstream. Alterations in urinary pH, then, can affect either the upper or lower boundary of regulation. In lighter smokers, for whom tolerance is not as fully developed, an increase in nicotine excretion rate may be giving the smoker the opportunity to consume more nicotine while still remaining below the upper boundary. At the lower boundary, of course, increased nicotine excretion induced by pH alterations may result in the onset of preliminary withdrawal symptoms which promote smoking (Silverstein, 1976).

As is the case with other drugs, variations in the "setting" of upper and lower boundaries between and within individuals are often responsible for psychological "achievements" that are perhaps more suitably regarded as physiological. For instance, one of the implications of regulation at the lower boundary is that the smoker who is restraining his intake to minimal levels (i.e., the smoker who may feel that he has successfully "cut down" in a remarkable feat of self-control) may be merely limiting "his" intake to his lower boundary. Others, with lower boundaries set at a higher level, may regard our will-powerful smoker with envy or admiration. But this commendable "self-control", so evident at or above the lower regulatory boundary, is often observed to desert our smoker when he is forced to contend with the more urgent pressures to smoke encountered below the lower regulatory bound.

The Regulation of Coffee Intake:

Coffee drinking has only recently come to be viewed as drug usage; consequently, there are only a few empirical studies on the behavioral pharmacology of caffeine use in man (see Gilbert, 1976, for a thorough review of the research on caffeine and coffee drinking). For purposes of this discussion, we shall focus on caffeine regulation in coffee drinking, while acknowledging that there are clearly other factors involved in the control of coffee intake.

Most people who regularly drink at least four cups of coffee per day are susceptible to caffeine withdrawal headaches and fatigue if they switch to decaffeinated coffee (Goldstein, Kaiser, & Whitby, 1969). To our knowledge, there is only one study on the effects of caffeineconsumption on coffee drinking in people (Kozlowski, 1976a). Withdrawal effects (headache, fatigue) were evident when drinkers were given "quarter-caffeine" coffee (i.e. 25 rather than the usual 100 mg. per cup). "Half-caffeine" doses (50 mg.) did not produce these withdrawal effects, and subjects showed signs of trying to regulate their caffeine dose by drinking more (5 cups) of this coffee than of the regular coffee (4.1 cups). This admittedly preliminary finding suggests that many coffee drinkers may ordinarily be supplying themselves with needless surpluses of caffeine: in this instance, the 160 mg. difference between 5 cups of 50 mg. coffee and 4.1 cups of 100 mg. coffee may be both superfluous in terms of regulation demands, and excessive in terms of stimulant stress on the system. This example indicates how knowledge of the relevant lower regulatory boundary and the range of drug indifference may have great practical significance in the design of safer drug preparations that will still provide users with the biological and social rewards of drug use while minimizing the dangers. Such dangers, of course, are evident at the upper regulatory bound; in the case of caffeine, the most obvious problem is that of excessive stimulation, the "coffee nerves" or "jitteriness" attendant upon overindulgence. These effects, in fact, are not dissimilar to those produced by excessive nicotine intake, though it must be remembered that nicotine is a much more toxic drug than is caffeine. In the case of any drug where a demonstrated or suspected excessive level of intake is a consideration-and in the case of some drugs, like alcohol and nicotine, the excessive dose may be anything greater than abstinence--it would be wise to try to limit the user's normal consumption to a level that will preclude withdrawal effects but not approach the upper limit of regulation. Whether such adjustments of intake to appropriate levels can be engineered by altering the drug concentration itself (as with the "half-caffeine" coffee) or whether alterations in consummatory behavior must be implemented directly remains a challenging question for clinical research.

Concluding comments on multiple drug use:

The consideration of regulatory boundaries as they apply to specific substances, complex as they may be in and of themselves, must not obscure the fundamental fact that individuals who use one drug are very likely to use another (e.g. Groves, 1974); when we include food as a regulated substance, all drug users become multiple substance users by definition. Of course, it is conceivable, in principle, that the regulatory systems that pertain to any given substance operate independently of any other systems; in short, there need not, in principle, be any necessary interaction effects involving the substances under consideration. This logical possibility, however, appears to be regrettably incorrect. It often seems to be the case that not only does each substance's regulatory mechanism interact with the mechanisms governing the intake of other substances, but that the type of interaction varies with the particular substances that are interacting. Second-order interactions are at this point too complex for us to contemplate, though they probably do account for some of the variance in drug consumption.

The possible effects of substance interactions are as extensive as the parameters that can vary. Effects on the lower boundary of regulation, for instance, might involve a lowering of the boundary of regulation for one substance as a consequence of the ingestion of another. An example might be that a smaller amount of nicotine is required to stave off withdrawal effects if the smoker has some caffeine in his system (Kozlowski, 1976b). Such an effect amounts to a form of substitutability of one drug for another. Such substitutability, it should be added, poses a number of interpretive and methodological problems for psychopharmacological researchers. For instance, if caffeine deprivation has the effect of increasing cigarette smoking, one might conclude that the individual's 11 need" for nicotine has somehow been augmented; it is also quite possible, however, that the perceived "need" for nicotine in this case is really nothing other than a misguided "need" for caffeine. In short, because the effects of these drugs (and of withdrawal from these drugs) are often quite similar, it is reasonable to expect their users to experience some confusion in interpreting their own physiological states and in assessing what exactly it is that they "need." Such confusion, it appears, is likely to characterize the researcher as much as it does his subject. If one sets out to discover the, precise upper and lower boundaries that pertain to the nicotine regulation of a particular smoker, one immediately must face the issue of how to "correct for" the effects of caffeine. If one assesses nicotine regulation in the smoker who has continued to ingest his normal quota of caffeine, one is quite possibly observing lower levels of nicotine consumption than would be evident if nicotine regulation could be observed in the absence of caffeine. On the other hand, if one were to ensure the absence of caffeine, by asking the smoker to, abstain from it for a number of hours, one would be observing nicotine regulation not only in the absence of caffeine, but also quite possibly in the presence of caffeine withdrawal symptoms, which might well elevate the level of nicotine intake above its "true" level. There is no definitive solution to this problem; and the problem, it must be remembered, is not simply a potential one, given the very strong likelihood that a cigarette smoker is also a heavy user of caffeine. Needless to say, the same sorts of confoundings apply to the interaction of the other substances under discussion.

Another example, presumably operating through a somewhat different mechanism, though with equivalent results, is the effect of nicotine on short-term hunger: smoking is considered to be an anorectic activity, reducing the need for food, or at least the level of deprivation below which hunger is experienced. It is to some extent an open question as to whether interactions that effectively lower the lower regulatory boundary also have the effect of lowering the upper boundary in a parallel fashion. Does coffee drinking reduce the amount of nicotine that can be consumed without experiencing the symptoms of excess? Does nicotine intake simply affect the hunger mechanism, as suggested above, or does it perhaps also affect (i.e. lower the "setting" for) the satiety mechanism, with the result that smoking not only reduces the need for food but also the amount of food that can be tolerated? There does seem to be some accumulating evidence that smoking has such effects on long-term weight regulation (Beller, 1977). One of the clearest and most reliable effects of smoking cessation is an increase in weight; and while some theorists argue that the increase in weight simply reflects increased eating as a substitute for the oral activity of smoking, it is equally possible that the effect of smoking cessation on weight change is mediated metabolically and does not require additional food intake in order to be manifested. In any case, one must ask why it is that smokers regulate their weight at a lower level before quitting than after.

The effects of alcohol on regulatory thresholds are also pronounced. Alcohol has a remarkable effect on cigarette smoking: acute doses of alcohol can double or even triple smoking rate in ongoing smokers (Herman, Ozyurt, & MacArthur, unpublished). In the absence of secure knowledge as to what exactly causes this effect, one may ask, in our regulatory terms, whether the effect is one at the lower boundary (i.e. alcohol somehow increases the effective lower boundary, and thus increases the demand for nicotine), or whether the effect is an upper boundary one (.e.g alcohol somehow increases the individual's tolerance for nicotine, and his ability to smoke a great deal.) At this point, either possibility is viable; but our analysis, by explicitly focussing on boundary phenomena, forces the researcher to generate specific hypotheses that pertain, perhaps, to different sorts of smokers.

In a similar vein, it appears that the effects of alcohol on eating are different in different sorts of individuals. While alcohol consumption decreases eating for nondieters, dieters react to alcohol intoxication by increasing their consumption of (attractive) food (Polivy & Herman, 1976). This sort of differential effect of one substance on another forces us to confront the issue of regulatory mechanisms in a correspondingly complex way. The fact that dieters and nondieters respond differently to alcohol demands that we employ a multi-faceted regulatory model in order to explain such behavior fully. It is imperative that we distinguish conceptually between upper and lower regulatory limits; and such a discrimination will almost inevitably force us to consider the possibility that pressures to increase (or decrease) consumption will be different at the lower regulatory bound than at the upper regulatory bound. Strictly pharmacological /physiological forces will predominate outside the range of regulation, whereas psychological /situational factors will be more evident within the range of regulation. Thus, at different levels of consumption, the sorts of prevailing pressures that apply, and the degree of force with which they operate, will vary significantly. The exploration of theinteraction of physiological and psychological factors in substance use, and the differentialinteraction of these forces as a function of one's current "location" relative to the upper or lower boundaries of regulation, will provide a continuing challenge to psychopharmacology, and will perhaps provide a response to that challenge as well.

REFERENCES

Ashton, H. & Watson, D. W. 1970 Puffing frequency and nicotine intake in cigarette smokers. British Medical Journal, 2, 679-681.

Beller, Anne Scott 1977 Fat and Then: A Natural History of Obesity. New York: Farrar, Strauss, Giroux.

Bemis, K. 1978 Current approaches to the etiology and treatment of anorexia nervosa.Psychological Bulletin, 85, 593-617.

Bermant, G. & Davidson, J. M. 1974 Biological Bases of Sexual Behavior. New York: Harper & Row.

Bruch, H. 1978 The Golden Cage. Cambridge, MA: Harvard University.

Gilbert, R., M. 1976 Caffeine as a drug of abuse. In Gibbins, R. et al (Eds.) Research Adavnces in Alcohol and Drug Problems. New York: Wiley.

Goldstein, A., Kaiser, S., & Whitby, 0. 1969 Psychotropic effects of caffeine in man. IV. Quantitative and qualitative differences associated with habituation to coffee. Clinical Pharmacology and Therapeutics, 10, 488-497.

Goodwin, D. 1976 Is Alcoholism Hereditary? New York: Oxford University Press.

Groves, W. E. 1974 Patterns of college student drug use and lifestyles. In Josephson, B. & Carroll, E. E. (Eds.) Drug Use: Epidemiological and Sociological Approaches. Washington, D.C.: Halstead.

Herman, C. P. 1978 Restrained eating. In Stunkard, A. J. Psychiatric Clinics of North America: Symposiumon Obesity. Philadelphia: Saunders.

Herman, C. P., Ozyurt, Y., & MacArthur, R. D. Effects of alcohol on smoking rate in light and heavy smokers. Unpublished manuscript. University of Toronto.

Isaac, P. F. & Rand, M. J. 1972 Cigarette smoking and plasma levels of nicotine. Nature, 236, 308-310.

Jaffe, J. H. & Jarvik, M. E. 1978 Tobacco use and tobacco use disorder. In Lipton, M. A., Di Mascio, A., & Killam, K. F. (Eds.) Psychopharmacology: A Generation of Progress. New York: Raven Press.

Jarvik, M. E. 1973 Further observations on nicotine as the reinforcing agent in smoking. In Dunn, W. L.Smoking Behavior: Motives and Incentives. Washington, D.C.: Winston.

Kozlowski, L. T. 1976 Effect of caffeine on coffee drinking. Nature, 264, 354-355 (a).

Kozlowski, L. T. 1976 Effects of caffeine consumption on nicotine consumption. Psychopharmacology, 47, 165-168.

Kozlowski, L. T. & Harford, M. A. 1976 On the significance of never using a drug: An example from cigarette smoking.Journal of Abnormal Psychology, 85, 433-434.

Lelbach, W. K. 1974 Organic pathology related to volume and pattern of alcohol use. In Gibbins, R. J. et al. (Eds.) Research Advances in Alcohol and Drug Problems. Volume 1. New York: Wiley.

Nishett, R. E. 1972 Hunger, obesity, and the ventromedial hypothalamus. Psychological Review, 79, 433453.

Polivy, J. & Herman, C. P. 1976 Effects of alcohol on eating behavior: Influences of mood and perceived intoxication.Journal of Abnormal Psychology, 85, 601-606.

Powley, T. L. 1977 The ventromedial hypothalamic syndrome, satiety, and a cephalic phase hypothesis.Psychological Review, 84, 89-126.

Russell, M. A. H. 1976 Tobacco smoking and nicotine dependence. In Gibbins, R. et al (Eds.) Research Advances in Alcohol and Drug Problems. New York: Wiley.

Schachter, S. 1971 Some extraordinary facts about obese humans and rats. American Psychologist, 26, 129-144.

Schachter, S., Silverstein, B., Kozlowski, L. T., Perlick, D., Herman, C. P. & Liebling, B. 1977 Studies of the interaction of psychological and pharmacological determinants of smoking. Journal of Experimental Psychology: General, 106, 3-40.

Schachter, S. 1978 Pharmacological and psychological determinants of smoking. Annals of Internal Medicine, 88, 104-114.

Silverstein, B. 1976 An addiction explanation of cigarette-induced relaxation. Unpublished doctoral dissertation, Columbia University.

Snyder, S. H. 1977 Opiate receptors and internal opiates. Scientific American, 236, 44-56.