It has become almost trite to say that drug abuse is the result of many interrelated factors. There is no one single sufficient cause. We are hearing about various factors today. Because there are many contributing factors there are therefore many ways of combating abuse. We should be looking for ways to understand as many of the different contributing factors as possible and learning how to use our knowledge to attack abuse from as many different angles as possible.

 

The phenomena I am going to talk about today are not the prime factors in initiation of drug abuse, though they may be more important than most people realize in the maintenance of abuse. They are of interest here, because by understanding them we may learn ways to weaken abuse patterns. They also illuminate some hitherto puzzling features of abuses: by seeing a little of what is going on, we can avoid following false trails.

 

Let me start with an example of a puzzling phenomenon, one that is at the heart of abuse. To any impartial observer the game is not worth the candle. Why does a heroin addict spend so much of his time chasing a fix of heroin when the effects are so trivial and evanescent and the withdrawal syndrome no worse than a dose of flu, indeed much less so for most street "addicts"? Why do alcoholics make themselves dizzy and nauseated and ruin their lives? It has been traditional to assume that with heroin there is a "euphoria" so intense that it overcomes all rational considerations, and that with alcohol there is a desperation to escape which is sufficient to make the distress acceptable. These explanations are ad hoc and there is really no independent evidence adequate to sustain them. What I am going to describe are experiments in animals that show that animals will work, just like addicts, for seemingly inadequate reasons, indeed for consequences that are not merely biologically useless but actually painful.

 

The only way I can describe the results is by a more or less historical narrative of a series of experiments, mostly performed by my colleagues, W.H. Morse and R.T. Kelleher, over a period of many years. No technical knowledge is necessary to follow the description.

 

Most of the experiments have been performed in squirrel monkeys which are small primates weighing between 1 and 2 pounds. The animals sit comfortably in so-called restraining chairs, and they have a small lever in front of them, which they can easily press with their hands. Not surprisingly, if they are kept short of food it is relatively easy to get them to press the lever if food is delivered when they do so, even only occasionally. Indeed they will press hundreds of times at rates greater than 1/sec for hours for occasional deliveries of food. Each lever press is called a response. The actual pattern in time of responding depends on the program that determines when food will follow a response; this program is called the schedule, the schedule of reinforcement, and this is the meaning of the term in my title.

 

An example of a schedule, one that will be important in subsequent discussion, is FI 10 min. It says that a stimulus such as a light comes on, and when it has been on for 600 sec (but only then) a response will produce food. Of course, the subject can respond during the 10 min interval, and indeed that is just what the subject does. When the subject is exposed-10 this schedule consistently over and over again, several times each day, day after day, a very characteristic pattern of responding develops. At the beginning of the 10 min the subject scarely responds, but when no more than a minute or two have elapsed, responding starts regularly, slowly at first, but increasing in rate smoothly until for the last few minutes of the 600 sec the animal is responding regularly at a rate that may be as high as 1/sec.

 

This particular pattern of responding is highly characteristic of the schedule. It is by no means just phenomenon of monkeys pressing a lever but has been seen in monkeys pushing a panel or pressing a lever or pulling a bar; dogs pressing a lever or nosing a key or barking; cats pressing a button or meowing; rats pressing a lever, licking or running in a running wheel; mice pressing a lever or breaking a beam of light with their noses or running around a circular track; and pigeons or quail pecking a key. The list is not exhaustive. Nor does the consequence have to be food. Consequences that have been studied include: food, fluid, electric shock, light, dark, warmth, and drug injections; and again the list is not exhaustive. Extensive review is not possible here.

 

Suffice it to say that for essentially all the pairs of response and consequences that have been studied, in all the species, many aspects of the pattern of responding over time depend on the schedule rather than on the response or the reinforcer. The commonality is amazing. Further, especially with regard to schedules like F1, the pattern of responding in time shows consistency as the parameter value is changed. The pattern that was described for Fl 10 min in the monkey is also seen in a pigeon under FI 100,000 sec, which is over 24 hours. For a few hours, the pigeon does not respond, then responding starts, slowly at first then increasing in rate and then occurring steadily over the last several hours of the cycle (Dews, 1965). One feature of schedules that resembles drug abuse is apparent already: a great deal of behavior over long periods of time is controlled by relatively trivial consequences.

 

We will return to FI but first we must consider some entirely independent lines of experiments that were started almost 45 years ago to try to learn something experimentally about anxiety. The problem of a functional characterization of anxiety was addressed. It was argued that the most important aspect of anxiety may not be the outward manifestation or inward feelings of anxiety but rather the interference with normal behavioral activities.

 

The following series of experiments in rats were devised. The rats pressed a lever under an FI 4 min schedule of food deliveries. When regular responding was occurring through the 1 hour daily session, a noise was sounded in the middle of the session for 5 min. At the end of the 5-minute noise each day an electric shock was delivered to the feet of the rat through the grid on which it stood. After a number of daily presentations of noise concluded by shock, responding slowed down or ceased during the period of noise, only to resume as soon as the shock had been delivered. The interpretation was that the rat was "anxious" during the noise, anticipating the shock, and the anxiety suppressed normal responding for food during the noise.

 

Some years later in an entirely different series of experiments, it was shown that a rat would press a lever if that response postponed electric shocks that would otherwide have occurred repeatedly. If, for example, each response postponed the shock for 20 or 30 sec, the rat responded regularly at an average rate much higher than once per 20 or 30 sec, so shocks were received only infrequently. Monkeys performed similarly but so that they even more infrequently received a shock.

 

What do you suppose would happen if the "anxiety" paradigm is superimposed on responding maintained by shock postponement rather than food delivery? What happened, probably to the surprise of the first workers to make the experiment, was that on the background of postponement respondings the rate increased during the noise preceding inevitable shock rather than decreasing as on the background of responding maintained by food. In subsequent series of experiments it was shown that when no shocks were given except at the end of the noise, so that responding was no longer postponing shock, then respondings ceased except during the noise. During the noise the monkey continued to respond until the delivery of the shock, although, of course, the responding had no influence on the occurrence of the shock or any other programmed event. Thus the animal was responding only when an imminent shock was signaled; so we could say that the shock was maintaining responding.

 

We now switch to an independent line of experiments. During the 1950s it was shown that the effects of a drug were greatly influenced by the pattern of responding in time on which the drug effect was superimposed. On different rates and patterns, for example, amphetamine could cause either an increase or decrease in rate and the same was true for barbiturates, although different patterns are increased and decreased by the two classes of drugs. Of course, the schedule of reinforcement is the prime determinant of the rate and pattern of responding in time. As described earlier, Fl schedules give rise to a smoothly increasing rate of responding through the FI X sec interval, a characteristic pattern consistent across species. FI responding thus provides a whole series of rates in a consistent sequence which can be exposed to the influence of a drug. For many drugs, the effects are different on the different rates. The question was posed: would the drug effects be similar if the drug effect were superimposed on a similar sequence of rates of responding but in reversed order, that is, with the rate at a maximum at the beginning of each cycle and then falling progressively to zero towards the end of cycle?

 

To try to produce such a pattern, two squirrel monkeys were trained under a VI schedule for food. VI is like FI except that instead of responding during a fixed period of time being concluded by a reinforced response, responding during variable periods of time are concluded by a reinforced response. Typically the periods vary from a few seconds to a few minutes. Such a schedule characteristically maintains a steady, constant rate of responding. On this pattern of sustained responding was superimposed the additional feature that at the end of 10 minutes, for a 1-minute period, every response was followed by a noxious electric shock to the tail. It was anticipated that each cycle might start with responding occurring at full VI rate and then slow down and cease as time approached when a response would produce a shock. In the event, responding in one subject was greatly reduced, especially as the shock became imminent. In the other subject, on the contrary, the frequency of responding during the 10-minute period actually accelerated, so that the subject came to be responding faster and faster as the time approached when a response would produce a shock. The subject came to be responding several times faster just before a response was followed by a shock than at any time under the earlier program when there were no shocks at all.

 

As soon as the first shock was delivered, however, responding practically ceased until the 1-minute period when every response would be followed by a shock was ended. The pattern of increasing rate of responding up to the time when a shock was delivered brought to mind the previous experiments in which responding that would not otherwise have occurred was maintained by an inevitable shock. Can one go one step further and maintain responding by noxious electric shocks that are a direct consequence of the response, and are thus not inevitable, that is, would not occur if the subject did not press the lever? Can responding be maintained when the only consequence of the response is the occasional delivery of a shock?

 

Accordingly, in the experiments just described, the delivery of food under the VI schedule was discontinued. The subjects now never received food on pressing the lever; all they could get were electric shocks to the tail at each response in the eleventh minute. The subjects continued to respond through each 10-minute period, however, showing the typical accelerating pattern of FI until a shock was delivered. On delivery of the shock, the monkeys vocalize and thrash around as though the shock were just as noxious and aversive as the same shock is known to be under other circumstances. They then virtually stopped responding for the rest of the 1-minute period when additional shocks would have been delivered, and through a time-out period, only to start again and increase the rate through the next 10-minute period. Thus, we have a subject responding repeatedly, showing a definite FI pattern of rates, when the only consequence of responding is the occasional delivery of a severe, apparently noxious and aversive electric shock under an FI schedule. The shock was highly effective in suppressing responding during the remainder of the 1-minute period of shock availability following the initial shock, so that the shock had not been magically transformed into a different kind of stimulus.

 

Continued responding when the only consequence of responding is a severe electric shock is not a transient phenomenon. One of the subjects in the original experiments was exposed to some 170 sessions over a period approaching a year, making an average of over 4,000 responses per session or a total of some 700,000 responses, maintained only by occasional shocks. Shock-producing responding was reduced when shock was reduced in intensity and disappeared when shock was removed, only to return when shock was reinstated. In other experiments rhesus monkeys worked steadily, day and night, for 7 days to postpone the delivery of a lesser shock to their tails than the shock the squirrel monkeys were self-administering.

 

Two points deserve emphasis. One is that the monkeys in the last series of experiments described had never pressed a lever to postpone a shock. It is not, therefore, possible to attribute the maintenance of responding by shocks to spurious shock-postponement responding, as there was no history of postponement. Second, in the later phases of the experiments the monkeys were fully fed, so that responding was not dependent on food

deprivation.

 

Shock-produced patterns of responding have also been generated by yet other different training procedures. Shock-maintained responding has been developed from shock-postponing responding by arranging that the length of time each response postponed the shock became progressively shorter with each succeeding response until it became 0 sec, that is the shock occurred when the last response was made. The original postponement time was then restored and the progressive reduction cycle repeated. Under this procedure, a progressive increase in rate of responding came to occur, ending with the shocked response. Removal of the interim postponing schedule left an FI schedule of shock-presentation to a response. Responding was maintained and the pattern of responding was that characteristic of FI.

 

The final way of developing shock-maintained responding that will be described involves no independent training. A species may have a high tendency to engage in a particular behavioral activity following an electric shock, for example, attacking and biting other members of the species. If two rats are put together in an enclosure and both given electric shock, the two rats typically rear on their hind legs and bite or attempt to bite one another. If only one subject is present, an inanimate object may be vigorously bitten. A rhesus monkey will bite a drinking spout when an electric shock is delivered to its tail. Such behaviors may occur the first time the shock is delivered. No training is required, and the activity may be similar in all members of a species that are studied and even homologous over a variety of species. Now, squirrel monkeys in laboratories commonly wear collars to which a chain may be attached for ease of handling, for example for taking the monkey from its cage and putting it in a restraining chair. If the chain is left attached to the collar when the monkey is in the chair and a shock is delivered to the tail, the monkey vigorously pulls on the chain. Pulling is seen the first time the shock is delivered, no training is required, and it is a consistent characteristic of squirrel monkeys. If the other end of the chain is attached to a switch activated by pulling the chain then activation of the switch can be made to lead to delivery of a shock, according, for example, to an FI schedule. When electric shocks were delivered the reciprocal processes of shocks producing chain-pulling and chain-pulling producing shocks came into play. Chain-pulling as an elicited response came to be maintained by shock as a consequence.

 

When shocks were made to occur only on a response, responding was maintained. Significantly, however, the pattern of chain-pulling changed. Initially, each shock delivery was followed by a paroxysm of chain-pulling that subsided into desultory pulling until the next shock started another bout of rapid pulling. As the procedure was continued, the bouts of pulling immediately after the shock became attenuated and pulling increased in frequency through the interval to the next shock just as we would expect for a response controlled by an FI X sec schedule. Thus, in this example, as in the previous ones, it was the schedule that determined the pattern of shock-maintained responding.

 

Responding maintained by electric shock as the consequence has been shown under FR (a schedule under which the number of responses made determines shock deliveries) as well as FI and the phenomenon has been shown in cats. Because these experiments effectively contradict strongly held prejudices of determinants of behavior, there has been great resistance to accepting them at face value and people have devised absurdly contrived anthropomorphic "explanations" to avoid facing the implications. In retrospect, one can find other instances in the literature of paradoxical effects of noxious stimuli, in which the noxious stimuli increased responding they had been expected to reduce.

 

We asked at the beginning: as the harm and distress caused by abuses seem so very much greater than the gratification they provide, why do abuses persist? We may not have a final answer, but at least we can now see the phenomenon in a broader context. "Abuse" is not a peculiarity of people and drugs. Perfectly normal experimental animals "abuse" themselves by giving themselves noxious electric shocks as well as by giving themselves drugs. The schedules engender the "abuse" in the animals. Perhaps then schedule type factors play a role in perpetuating drug abuse in humans. The work on schedules has shown that even an unpleasant event, properly scheduled, can maintain a lot of behavior. We don't therefore have to find a predominantly positive effect of a drug to account for it being abused: a bad effect, properly scheduled, could maintain a habit.

 

Practical consequence follow this line of reasoning. One outstanding feature of schedule controlled behavior is that while it can be very strong and well maintained almost indefinitely—in pharmacological studies we have had monkeys work for hours a day for years; they were doing more or less the same at the end as at the beginning—yet it continues to remain easily modified when the schedule is changed.

 

Thus, to the extent that schedule factors play a role in human drug abuse, that part, at least, is fully remediable. What sort of a schedule is a heroin addict influenced by? Well, it is a very regular schedule of hustling, making contact to get a fix and then the ritual of taking the drug, pursued so religiously that it may leave little time for other pursuits. How can we interrupt the operation of the schedule? One obvious way is to make the abuser take and keep a job with regular hours. Vaillant in his studies here in New York found in his follow-up of ex denizens of Lexington that by far their best drug-free periods (when they were not institutionalized) was when they had a job that they had to keep (because of a parole requirement that made a condition of freedom the holding of a regular job.)

 

We know very little about schedule effects in human subjects and it is a very difficult area of research. But research in human subjects will have to be done, before we can really apply laboratory findings to therapy. I believe the benefits, when we learn to apply lab findings to humans, will be great enough to justify the cost and frustrations of doing the work.

 

ACKNOWLEDGMENT

 

Original work in this laboratory was supported by U.S. Public Health Service grants MH02094, MH07658, DA00499, and DA02658.