B. F. SKINNER, the Harvard psychologist, once conducted an experiment to see if two cats, left to themselves, would share a task essential to their joint survival. His apparatus consisted of a cage with a small pedal at one end; pushing the pedal caused a pellet of food to drop into a cup at the other end. The apparatus, which he had originally designed for an earlier experiment, has come to be known as a Skinner box.

The cats were taught separately to feed themselves by operating the pedal. Skinner next deprived the animals of food for 24 hours and then put them in the cage. One of the two cats spent its time alternately working the pedal and vainly running to the empty cup. The reason the cup was empty was that the other cat simply stood by the cup and ate the pellets as they came out. When it had eaten its fill, it curled up for a nap. The first cat then fed itself.

Scores of other behavior patterns have since been investigated by both specialists and amateurs using modified Skinner boxes. For example, a Skinner box made with lighted windows of various hues can disclose the range of color that an animal can perceive [see "The Amateur Scientist, SCIENTIFIC AMFRICAN, October, 1970]. The learning that results from these procedures is called operant conditioning because the animal actively operates some aspect of its environment. Skinner boxes of the appropriate proportions can disclose variations in the sense perceptions and social behavior of animals ranging from mice and birds to dogs and pigs.

The first steps in doing such an experiment are acquiring a Skinner box and mastering the simple art of teaching the animal to use it. The box can be built by anyone reasonably skilled in the use of hand tools. Alternatively, inexpensive Skinner boxes designed for easy modification and use with animals the size of mice and rats are commercially available in the form of kits from Life Science Associates (P.O. Box 500, 1 Fenimore Road, Bayport, N.Y. 11705). The company also distributes a paperback manual titled Experiments in Operant Conditioning, by John M. Christiano, which presents in detail the technique of teaching animals to use the apparatus and of evaluating the results. Christiano also designed the apparatus distributed by the company.

The experimenter must provide at least one easily cleaned container for housing the animals. The containers are kept at room temperature between tests. To ensure that the experimenter is in compliance with the Expanded Animal Welfare Act of 1910 and with the policies of the National Institutes of Health on the humane treatment of animals in the laboratory, experimenters are urged to obtain a copy of the booklet Guide for the Care and Use of Laboratory Animals. It is available for 70 cents from the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402.

The cost of the experiments varies with the size of the experimental animals, but the results of experiments with small animals are no less interesting than those of experiments with large ones. Moreover, teaching a mouse to perform an operation it would not undertake naturally can be just as satisfying as teaching a similar "trick" to a dog. Suppose you decide to work with white mice.

Live mice that are known to be in good health can be bought from companies that specialize in biological supplies, although most of these establishments do not fill small orders. The beginner should therefore ask for advice on procurement from an experienced person, perhaps a biology teacher in a local school or someone who makes a hobby of breeding mice. Not all pet shops are reliable sources of healthy animals. Mice bought from a pet shop should be checked by a local veterinarian for a few days. Frequently one can obtain a breeding pair from a local educational institution. With a pair or two from such a source you can soon raise a small laboratory colony from stock known to be healthy.

If breeding is undertaken, arrange adequate housing to keep the sexes and even the individuals apart as desired. A single pair of mice, if left to themselves, can produce 100 offspring per year. Each generation can begin to reproduce itself in from five to eight weeks. The unchecked colony could multiply within a year to more than 50,000 individuals. Humane considerations require that the size of the colony be limited to the minimum number of animals needed for the experiments. Mice, like other animals raised in captivity, lose many of their defenses as well as their food-gathering ability. Any such mice released into the natural environment are almost certainly doomed to starvation or to death by predation.

Christiano's manual on operant conditioning urges the experimenter to handle each animal briefly every day for a few days before teaching it to operate the Skinner box. Never chase the animal around its home cage with your hand, an action that reinforces the creature's natural tendency to run. Instead dangle your hand in the cage. Have patience. The animal will soon venture near, often with its tail conveniently outstretched. Pick it up by the middle of the tail.

Christiano also suggests that you let the mouse walk on your bare hand. You should wear a sleeved shirt that can be easily washed. Pick up the mouse by its tail. Lower it onto your sleeve. By gentle nudging induce it to walk from your forearm onto your hand and thence onto the cage floor. Repeat the procedure several times. Finally, put your hand flat on the cage floor. Wait patiently until the mouse climbs onto your hand. Keep your fingertips buried in the bedding of the cage. This stratagem will help you to avoid getting a slight nip. You will be nipped only if the animal is frightened. If you are bitten, wash the small wound with soap and water and consult your physician. Laboratory mice are unlikely to carry a disease. They are more likely to catch something from you.

Having learned to work with mice, familiarize yourself with the Skinner box. The device consists of two essential elements: a housing that more or less confines the animal and a mechanism that presents simultaneously to the animal a reward such as food or water and a signal such as a sound or a light. The experimenter trains the animal by giving it a reward every time the animal happens to behave in the desired way or makes a move that tends to lead to the desired behavior.

The reward, called a reinforcement by psychologists, is most effective only if it is presented simultaneously with the desired behavior or with some tendency toward that-behavior. A delay of even a second between the behavior and the presentation of reinforcement can destroy much of the desired effect. In practice it may be difficult to synchronize the behavior and the reinforcement. For example, the animal may be in a remote part of the Skinner box and facing away from the source of food or water.

Operant-conditioning experimenters solve that problem by applying a principle from I. P. Pavlov's classic experiments with dogs. Pavlov demonstrated that when a neutral stimulus such as a bell is presented at the same time as food, the bell acquires some of the conditioning properties of the food. In a Skinner box that presents water as a reinforcement the water dipper makes a characteristic sound as it comes up. This sound becomes a secondary reinforcement and can be used to instantly reinforce the desired behavior. These techniques, based on the experiments of Pavlov and Skinner, enable an experimenter to condition in a matter of minutes a pattern of behavior that might take a professional animal trainer using traditional methods days or weeks to teach.

A simple but highly effective Skinner box has been designed for mouse experiments with water as the reinforcement [see illustration at right]. The box has a front and back of transparent plastic; the ends, the floor and a hinged top are opaque. A mechanism for presenting the water and the signal simultaneously runs the length of the box just under the middle of the floor. It includes a slender metal shaft that is free to turn in holes serving as bearings in the ends of the box.

One end of the shaft supports a counterbalancing lever arm. Clamped to the other end is a similar lever that serves to turn the shaft. This lever can be coupled to a smaller one by a piece of stiff wire. The smaller lever extends through the end of the box. The inner end of the smaller lever carries a pedal that a mouse sitting on its haunches can depress with its front paws.

The shaft also carries another lever, the outer end of which terminates in a horizontal disk about eight millimeters in diameter. When the shaft is turned to one extreme of its excursion, the disk is submerged in a small container of water under the floor. The counterweight is adjusted to the position where it automatically rotates the shaft to its opposite extreme, thus lifting the wet disk through the hole in the floor to a position that is accessible to the thirsty mouse. The upward excursion is stopped when the supporting lever strikes the bottom of the hole in the floor. The resulting sound, a distinct click, is heard at the instant the wet disk becomes available to the mouse.

A mouse that has been deprived of water for 24 hours will usually lick the film of water off the disk. More water could of course be delivered during each operation of the lever by substituting a shallow cup for the disk. Then, however, the animal might quench its thirst before the experimenter could complete a series of conditioning exercises.

Why not use food as the reinforcing agent? Food works as well as water for certain investigations and better than water for others. On the other hand, it is not needed for an experiment intended primarily to introduce the experimenter to the technique of teaching an animal how to use the Skinner box.

If food is used, the experimenter must first establish how much the animal normally eats and weighs when it is given unrestricted access to food and water. When these data have been compiled over a period of time, the food supply is reduced by 25 percent until the animal's weight has been reduced by 20 percent. The experimenter can then work out a feeding schedule that maintains a healthy but hungry animal.

Set up the initial experiment in a quiet, normally lighted room. Put the Skinner box on a table. Elevate the wet dipper and put in the box a mouse that is accustomed to being handled and that 3 has been deprived of water for about 24 hours. (The precise period of time depends on the relative humidity and the moisture content of the animal's food.)

The mouse will soon begin to explore its surroundings. Put your hand on the control lever of the dipper assembly but do not make any further movements or otherwise distract the mouse. Soon, usually within a minute or less, the mouse will approach the disk and lick off the water.

Gently but quickly lower the disk into the water container and immediately let it rise to make a click. (Time the action to avoid hitting the nose of the animal with the dipper) The mouse will again lick the disk. Repeat the procedure about six times.

Lower the dipper into the container and let it stay there. Wait until the animal turns its head away. The mouse will have begun to associate the click with the presentation of water. Depress and release the lever. At the sound of the click the mouse will turn its head to the n dipper and again take water. Repeat the routine several times. Then wait until the mouse turns its body partly or fully away from the dipper before repeating the exercise.

The training routine requires some patience on the part of both the animal and the experimenter. Never touch the mouse, nudge it, speak to it or otherwise influence its behavior except by manipulating the apparatus. Limit the initial training period to 10 minutes. Return the animal to its home cage with a normal supply of food and water for 24 hours. Then deprive it of water for 24 hours and proceed with the second 10 minute training session.

Begin the second session by placing the mouse in the box with the dipper immersed in the water. After the animal has explored the box for a few minutes release the lever so that the dipper rises and sounds a click. The probability is high that the conditioned mouse will immediately go to the dipper and lick off the water. If it does not, let it find the dipper by exploration. Then rewet the dipper only once before the mouse turns away. This single operation will be sufficient to restore its memory.

The next step is to reinforce the desired behavior only when the mouse turns in the direction of the small lever that has the pedal on its inner end. After this behavior has been reinforced three or four times withhold the reinforcement until the mouse takes a step in the direction of the pedal. Then, having repeated the exercise two or three times, reinforce only after the mouse has taken two steps. By extending the routine you can teach the mouse to walk (even scurry) from the dipper to a position immediately under the pedal. You should be able to achieve this result by the end of the second training session.

Again allow the animal to eat and drink normally for 24 hours in its home cage. Then, following another 24 hours when the animal is deprived of water, undertake the third training session. Begin it exactly as you did the second, reminding the mouse of what it has learned as may be necessary. Thereafter when the mouse arrives at a position under the pedal, reinforce only an upward movement of its head. Soon it will hurry to the dipper and lift its head high.

When you have "shaped" this pattern of behavior, reinforce only after the animal has begun to sit upright on its haunches. By now, if you have been careful to reinforce only desired behavior, you should be able to induce the mouse to assume the upright position within a maximum period of five minutes. The remainder of the exercise is obvious. Shape the behavior of the mouse so that it puts one paw or both paws on the pedal and pushes it down.

Connect the wire link between the small lever and the large lever that you have been depressing. Thereafter the mouse will operate the mechanism until it is no longer thirsty. It might be interesting to repeat Skinner's investigation of work sharing by substituting a pair of trained and thirsty mice for his hungry cats.

The range of investigations that can be undertaken with the Skinner box is limited primarily by the ingenuity of the experimenter. Remember, however, that (as Christiano's manual points out) your timing in applying reinforcement is crucial to the success of all behavior shaping. Otherwise you may reinforce not the behavior you wanted but whatever the animal was doing at the exact moment the click sounded. Tail chewing, nose scratching, toe picking, circle turning, head bobbing and any other conceivable behavior might actually have been reinforced. Be sure to reinforce only the exact behavior that you want repeated.

It is possible to extinguish a pattern of reinforced behavior. Put the conditioned animal in the Skinner box for the usual conditioning period but remove the wire link between the pedal and the mechanism that operates the dipper. The animal will of course operate the pedal in an attempt to provide itself with water. In the absence of reinforcement the number of pedal presses per minute will decrease with time. Return the animal to its home cage for an hour or so (without water) and repeat the procedure. A smaller number of pedal presses will be observed, although usually an initial spurt of responding known as "spontaneous recovery" will occur. If this routine is followed for a few-cycles, the animal's initial conditioning will be extinguished.

You may observe other patterns of behavior during extinction. For example, the mouse may strike the pedal sharply, bite it, resort to jumping or quick jerky motions, defecate, urinate or wash its face excessively. It is useful and interesting during any conditioning experiment to keep a detailed log of all behavior of every animal, including the number of times it pushes the pedal during each conditioning session.

The mice can be marked for identification by placing a tiny spot of distinctively colored lacquer, or coded combinations of colored spots, on the back of every animal. Coat only the tips of a few hairs. Keep and chart records of all responses.

Mice, like other animals, can learn to respond to more than one signal. For example, the Skinner box can be fitted with a miniature electric lamp placed, say, near the pedal. The experimenter can turn the lamp on and off by an external switch. An animal that has been conditioned to receive water at the sound of a click can then be conditioned to respond only when the lamp is on or, alternatively, when it is off.

The conditioning procedure is simple in concept. First the animal is placed in the Skinner box with the wet dipper and the lighted lamp. Remove the wire link and operate the dipper by hand. After the mouse has pressed the pedal perhaps six times turn off the lamp. Turn it on again just before the mouse operates the pedal. Gradually lengthen the time that the lamp is off, but do not make the mistake of letting the mouse successfully operate the mechanism and receive water while the lamp is off. Eventually you will succeed in extinguishing the established behavior pattern when the lamp is off. The animal will learn to operate the pedal only when the lamp is on, not when it is off.

Established patterns can be selectively alternative techniques. For example, you can let the animal operate the pedal successfully for a stated period of, say, 30 seconds with the lamp lighted. Then turn the lamp off and simultaneously disable the apparatus for the same interval. After a sufficient number of conditioning cycles the animal will operate the pedal less frequently when the lamp is off, and finally it will not operate the pedal at all. Whatever technique you employ, the objective is to establish a sequential link (as reflected by the behavior of the experimental animal) between the two signals, the click and the lamp.

Can an animal be conditioned with a Skinner box to perform a prescribed set of operations serially? The answer is yes if the experimenter masters the art of linking (as evidenced by the animal's behavior) a set of sequential signals and responses. The series of responses can be quite complex.

Some years ago a commercial device based on that principle was displayed in the Chinatown section of New York. A sign on a wire cage invited the passerby to drop a coin in a mechanized collection box to see an "educated chicken" perform. When the coin dropped, a bell rang. A chicken in a corner of the cage immediately ran to the center and pulled a chain suspended from the top of the cage. That caused a mechanism to deliver a picture postcard of the chicken to the customer. The chicken then did a pirouette and pecked a pedal. The pedal caused a kernel of corn to drop into a cup. After snatching up the corn the chicken ended the show by retiring to its home corner.

Christiano suggests that behavior patterns of this complexity be initiated with a mouse that has been thoroughly conditioned to operate the pedal only when the light is on. Hang a looped key chain on the panel near the lamp and pedal. Put the conditioned mouse in the box with the lamp off. At first reinforce any casual or accidental touch of the chain by immediately turning on the light, which has by now become a secondary reinforcer. Maintain the lighted condition until the animal presses the pedal and obtains the primary reinforcer (the water). Turn off the lamp in exact synchrony with the presentation of the primary reinforcement Thereafter, as in the other experiments, shape the chain-touching behavior by the method of successive approximations, always withholding the secondary reinforcement (the light) until the sequential pattern of behavior that you are attempting to shape has been observed.

Bibliography

SCHEDULES OF REINFORCEMENT. C. B. Serster and B. F. Skinner. Appleton Century Crofts, 1957.

THE ANALYSIS OF BEHAVIOR. James Holland and B. F. Skinner. McGraw-Hill Book Company, 1961.

The Society for Amateur Scientists (SAS) is a nonprofit research and educational organization dedicated to helping people enrich their lives by following their passion to take part in scientific adventures of all kinds.