of Behavior, Desensitization
There are many skilled trainers out there in cyberspace who know all, or most, of what I am going to present in this message. To those old-hands, my apologies for covering familiar ground in such detail. I hope you will agree that, to those who have not been exposed to the demands of achieving reliable behavior, these are principles and practices that deserve discussion.
As some of you know, Animal Behavior Enterprises, our company for almost 5 decades, was a for-profit organization dedicated primarily (but not wholly) to the study and training of animals. Therefore, time was money, and efficiency was vitally important. However, we learned very early where we could not, should not, and did not cut corners.
We were very
patient (as my kids said frequently, "like the chipmunks")
when establishing STIMULUS CONTROL and BEHAVIORAL FLUENCY (doing the
behavior quickly and doing it
We also spent what some would consider an inordinate amount of time analyzing behavior, both the behavior we wanted and the (natural) behavior emitted by the animal (what we had to work with). We did this before anyone picked up a clicker (or other device). We then determined the best way to get from the animal's natural repertoire to what we wanted. This often took a bit longer to do than it does to say. And, we were not always right in the path we chose, although, after more than 15,000 animals and over 140 different species, we got pretty good at this. To an outsider, it might have looked that we were not at all concerned with getting on with the job - again, WRONG!
When we found we had made mistakes, and we made some really good (read huge) ones, we had no hesitation about cutting our losses, going back to square one, and starting over. We knew the penalties for continuing to reinforce unwanted behavior - LOOOONNNGGERRR periods of extinction; and that costs MONEY.
While we are believers in operant conditioning, we know that ours is not a perfect technology. We also know that human error is a major factor in failed programs. Human error in program design. Human error in the craft of training. We constantly reviewed our progress. If we were stalled, we tried to ascertain why, and then moved on. But, again, we took as long as necessary to achieve a goal, and we seldom reduced our criteria to give the appearance of success so that we could move on to the next step. Moving on prematurely almost always returns to haunt a training program, sometimes sooner, sometimes later. So, we backed up or started over when necessary, and we did not move on until the animal and the trainer were ready. In contrast, we have found many trainers (for reasons not always their fault) keep pushing ahead, accepting mediocrity in early stages of training, and possibly hoping to polish up behaviors later down the road - still again, WRONG!
What I have pointed out here is the importance of the EARLY stages of training. Consider a novice trainer, just learning the craft, and making lots of errors - understandable and forgivable errors, but errors nonetheless. The animal is subjected to many instances of grossly misreinforced behaviors - sins of omission and of commission, all of which build a repertoire of reoccurring undesirable behaviors.
What about the old hand? Well, this is his or her 40th dog, and he or she is bored to death. Laboring under the false premise that the really important training does not really begin until much later, he or she sloughs off the early training, making only a half hearted effort to precisely shape behavior. Perhaps a pro might even turning it over to an apprentice, unsupervised, of course. Much the same as with the amateur, poor behavior is rewarded, good behavior goes begging. In any event, the critical foundation of introduction of stimulus control, including the introduction of the vital secondary reinforcer, and the development of what we might call, for want of a better term, a work ethic, is left to the whims of fate. Need I say - WRONG!
Of course, I recognize that a pet owner must learn training sometime and somewhere and with some animal, most likely his or her own pet. Also, trainers, professional or otherwise must get experience somehow. Also, we don't want trainers to be paralyzed with fear about doing the wrong thing. What we must strive for is the balance between concern for applying the proper techniques and methods, and the recognition that ours is an imperfect world, and we must live with that.
terms, this means recognizing and accepting that there will be mistakes.
It also means that we should be more realistic about how long it takes
to train a behavior, or a set of behaviors. If you are a pro, and your
reinforcements are fairly well selected and timed about half the time
(a very good level of performance, by the way), you should allow yourself
I suggest to trainers that they pay proper attention to an animal's early experiences with the training process. No more saying that early training doesnt much matter; that it will all sort itself out later on during the REAL training. When you pick up the clicker for the first time - YOU ARE TRAINING, and be aware that you are training. Behavior (yours) has consequences (the animal's behavior).
This thought for novice and pro trainers alike, don't try to NOT MAKE MISTAKES. Dont let the thought of mistakes enter your head. Don't think much about what you should NOT do. Instead, concentrate on what you know you SHOULD do. One reason not to concern yourself excessively about mistakes is that you will make them anyway, so don't worry about them. You will make them. I will make them. We all will make them. If we try veryhard to shape our behavior towards perfection, we will get closer to that trainer's Holy Grail - the PERFECT REINFORCEMENT. And, of course, once we give an animal that perfect reinforcement, everything would be wonderful - right? WRONG again.
What will follow that perfect reinforcement will be another reinforcement that might be OK, but maybe not. Just think about it - the effect of that perfect reinforcement will be buried under all of those imperfect rewards. That is the trainer's lot, always another reinforcement to give - and another, and another. Some will be very good, many will be less good. Some will be downright bad. Keep trying for the good, and let the bad ones take care of themselves - don't agonize or punish yourself or the animal. The message: do your best to give the sharpest, most well-timed click you can, but, keep on truckin', regardless.
All of this may seem long winded, but, there is a purpose to this discourse. Here is my very simple philosophy of training: THE MOST GOOD REINFORCEMENTS POSSIBLE FOR THE MOST ACCURATE AND FASTEST BEHAVIORS IN THE TIME AVAILABLE.
Say what? as my navy son would say. Putting it another way: You want to give as many trials and well selected and timed rewards as possible to shape the sharpest and most precise behavior of your choosing, within the time that is available. Notice that I said nothing about getting the behavior as fast as possible, or in the fewest trials. It will take as many trials as it takes, and not one trial more or less. I did say the most accurate, or best, behavior possible within the time available.
Remember the commercial THE QUALITY GOES IN BEFORE THE LABEL GOES ON? Let that commercial be your guide to quality assurance or control. Now, did I say that there is an unlimited time for a behavior? No. Not in real life. You train within the time constraints set by yourself or your client. But, DO NOT BUILD AN ARTIFICIAL AND UNNECESSARY TIME REQUIREMENT INTO YOUR TRAINING PROGRAM.
How often have you said that you would train a behavior within a certain period of time? Maybe it was necessary for you to do this. But, maybe it wasn't. Maybe you said it just to indicate your prowess and facility with the craft (or, face it, to feed your own ego. No? Well, I have. And, in the end, it hurt to look at a finished behavior that I knew I could have shaped to be more accurate.). I suggest that it would be best to use the most time available to get the best product, rather than find yourself cutting corners just to meet a self-inflicted deadline.
But, what about a pet owner. A pet owner is not often worried about deadlines, or difficult or even life-or-death training situations. Sometimes a pet owner has the most difficult situation of all; poor environmental control (wife needs the kitchen during training time, or kids screaming in the living room), erratic schedules (what to do on vacations, or long weekends), or inconsistent reinforcement (you cant believe how creative a 6 year-old can be with a clicker, and not all creative behavior is good). And, there are some truly challenging behavioral problems for dog and cat owners - including the animals being uncontrollable at certain times, such as while chasing other animals or not responding properly to cues.
All of this has been building towards the process of desensitization of animals to environmental stimuli and establishing stimulus control. Animal Behavior Enterprises (ABE) spent many years pushing the limits of desensitization and stimulus control. Our animals, whether used in entertainment, commerce, or for the military or other purposes, all had to face unusual situations, unfamiliar people, and some difficult, sometimes hazardous, environmental conditions. In spite of all of these stressful conditions, ABEs animals were trained to perform their tasks, sometimes for several hours without reinforcement.
As I mentioned
earlier, we have trained over 15,000 animals. We, ABE, used about 4,000
or 5,000 or so for our own shows in Hot Springs, Arkansas, or in our
touring shows that traveled all over the USA and Canada. The remainder
of the animals, about 10,000 were sent out to clients and customers
all over the world. Each and every ABE animal had a rather lengthy period
of environmental adaptation prior to and during training and before
being shipped out to do its job. Such adaptation might include something
as simple as a bunch of helium filled balloons attached to a continuously
moving cable and marched up and down in front of hundreds of chickens,
rabbits and ducks during various stages of training. The erratic behavior
of floating balloons have a most devastating effect on some
However, most trainers find the performance of our cats and dogs most remarkable. Indeed, the cats and dogs did demonstrate the high degree of stimulus control that is possible with straight forward operant and respondent conditioning techniques. I wish to emphasize here that neither Marian nor I claim to have any secret method for establishing such operant and respondent control. Many have done what we have done, including Pavlov and Skinner. What we did do, perhaps, is put it all together, and push it further than most. We know that we stood on the shoulders of others, and we are most grateful for the technology handed us.
I will talk mostly about cats. We began with both cats and dogs. However, we soon found that cats were much tougher than dogs, in many ways and for many reasons. To save time, we opted to reduce the effort on dogs and concentrate on cats. We believed (and it appears to be true, within reason) what we did do with cats, we could do with dogs, except we could do it quicker with dogs. Also, for the most part, there are fewer environmental restrictions (temperature, humidity, etc) with dogs than with cats.
PATIENCE, or the waiting game.
a number of systems using cats. The requirement of some of these systems
was that the cat must sit, or lie down, for long periods of time, up
to 20 to 30 minutes (longer in some systems), depending on circumstances.
As important as the duration of the STAY is that the cat must be perfectly
quiet and not move, regardless of the surrounding
It all began
during the selection process. First, cats had to pass a hearing test,
and then a simple tractability test. Cats that passed the tractability
test were placed in cages in a high traffic location at our training
farm. Constantly during the day, people would pass by cages.
Next, in the same general high traffic area, we taught the cats to sit on command on a special sensor equiped platform, and to remain seated while the SIT signal was heard. The SIT signal could have been almost anything (and we tried out lots of different signals), but we settled on a slowly warbling 9 kHz (or kcps, in the old days) tone. If the cat sat for the duration of the SIT signal, the cat was rewarded automatically by a special feeder, which dropped the food into a nearby pan. The cat got up and ate. Within a minute or so, the next trial began. If the cat got up prematurely, the signal stopped, the time reset, and for several seconds, there was a time-out. After a short time, the signal to sit began again, the cat sat down, and the clock was running. We gradually extended the time using a variable interval (extended the WAIT duration during the signal) schedule. We had several training cages equipped with small platforms for WAIT training, so we could train several cats at once. Over a period of many days, and hundreds of trials, each cat learned to sit quietly while people and animals of various descriptions went by. Each cat had two sessions per day. At first the sessions were a half hour in length. Later, the time in the cage during each session was extended to two hours. We did extend some sessions to more than 8 hours, but just to collect data on the effects of longer sessions. We wanted to test the limits of the system, but not unduly stress the cats.
The behavior of the cats during this early WAIT training was interesting. The cats quickly learned to keep one foot on platform while moving the rest of the body about the cage. This was humorous to watch, but cats behaving like a base-runner in baseball was not the objective of the training. We increased the pressure required to trigger the switches during the sit. This kept the cat closer to home, firmly seated on the wait-platform. We then began to shrink the size of the platform. Soon, the cat had to sit on a small platform. The cat could not lie down. The cat had to maintain a more or less standard cat sitting position in order to trigger the switch. We ran the SIT time requirement up to 10 to 20 minutes, depending on the objectives for individual cats. All cats had to do 10 minutes. Selected cats had to do 20 minutes, or more. The cats had no trouble with the SIT criteria. Within a few months we had more than a dozen cats sitting still on command.
As the time was being extended, there was some increase in the cats' environmental complexity. Once the cats would sit reasonably still, and we achieved our time goals, we REALLY began to increase the frequency, duration, and intensity of environmental noise. Noise, used here, means more than sounds. Noise in this context includes lights of various colors and intensities, both steady and flashing; vibrations, from slight to heavy; winds (big fans), light to heavy; unfamiliar objects moving by, etc. We even dragged furry objects that looked like mice past the cages. What was constant was the SIT signal. If there was a training break, and there was no signal, there were no distractions presented. Later on, there were some additional signals used for guidance and as a RELEASE that told the cat it was on its own time - it was no longer necessary to sit; it could BUG OUT, if it wanted to. There was also a RECALL SIGNAL that was a VERY SPECIAL signal that said COME HOME - BUT THESE SIGNALS HAD TRAINING PROGRAMS OF THEIR OWN, INDEPENDENT OF WAIT AND DESENSITIZATION. Sometimes these other signals were being trained concurrently with the wait-training, but a complete discussion of the systems development program is beyond the scope of this discussion. Of course, the integrated system of cat, behaviors, trainer-handler, and equipment, had to function as a unit. However, that is another story.
All of this training took place inside of cages. Cats in cages are of limited use. It was time to take the cats out of their cages. We began by removing the sides of the cages, first the back, then the right and left sides, and then the top, and then the front. WE SPENT 4 MONTHS, SOMETIMES MORE, DOING THIS! We were constantly introducing environmental distractions. The cat had to sit still as long as the sit command was being given, regardless of environmental distractions. At no time were any of the cats physically abused. Certain objects fell close by, but never touched the cats. Sounds became more intense, and sharper. Objects moved all around and over the cats. Ideally, to determine the cats level of anxiety during this desensitization process, we should have constantly monitored medical data such as heart rate, blood pressure, blood chemistry, body temperature, pupillary activity, etc. Such information would have alerted us to the extent and type of Pavlovian conditioning that was taking place. However, all of this was going on in the mid to late 1960s and well into the 1970s, and there were not the wonderful medical gadgetry we have today. We occasionally monitored medical data, but it was such a hassle, and it was unreliable anyway. We did observe the cats eyes (pupillary activity) as much as possible. When the cat becomes excited, there are rapid changes and often extreme dilation of the pupils. It is this WILD EYE, accompanied by laid-back ears that are good indicators of when a catis about to panic. We made efforts to avoid such levels of excitement. Most of the time, our cats had the look of boredom, curiosity, or bemusement that seem to be the hallmarks of well known cat expressions. One might say we cultivated such blase appearing behavior. All in all, wait-training, and environmental desensitization, as a combined entity, lasted from 6 to 9 months. Of course, environmental desensitization continued along with field training, and wait-training became incorporated into the system, so desensitization and wait-training continued throughout the program.
Over a period of about 60 days, the wait-training and desensitization programs moved outside. During part of that time, we reintroduced the cage, both to make training more efficient, and to assure the animals reliable performance and safety. Over several weeks we removed the cage, all of the time looking for signs of anxiety while introducing new and ever more intrusive stimuli, including nearby large-caliber gunfire. We also began to take the cat to new territories, where there were large crowds of people engaged in various activities. At first the cat was in a standard cat transporter. Later, the cat was released and held by the trainer, or the cat was given a sit command at the feet of the trainer. The trainer would slowly back away, leaving the cat somewhat exposed to the passage of the surrounding crowds. The cats were also taken into parks and into woods where there was abundant wildlife. The cats would be reinforced frequently for sitting still in the face of all of the various distractions. Again, concurrent with this desensitization were other training programs. Each cat was kept very busy during the workday. All indications were that the cats slept very well at night, in spite of their nocturnal nature.
When we began open field training in meadows in the Spring, we were faced with perhaps the greatest challenge we had with our cats - grasshoppers. The fields were thick with them, and they not only jumped, but they flew, and with a nice buzzing sound that clearly drove the cats crazy. Usually, while we were actively guiding the cats, or had them performing some other labor intensive task, there were no significant problems with these insect distractions. But, during the long wait periods, and especially when the hot sun was beating down, and a grasshopper would buzz right near a cats head -- off the cat would go, trying to get the crunchy morsel. We tried an experiment of occasionally giving a release if a cat had performed well for a sufficient period of time. However, there was not sufficient data to establish if this was really effective in reducing breaks in discipline; it did make the trainers feel better though.
Rain was another environmental condition that shook the cats concentration, and the heavier the rain, the greater the distraction. We did not have to worry much about very heavy rains because that interfered with the control signals anyway, which allowed the trials to be terminated. In light to moderate rains, the cats would sit there, sometimes licking its fur, especially its flanks and paws. Some cats developed the habit in the rain of closing its eyes. That looked very strange, a cat sitting there with its eyes closed. We must admit that we often avoided rainy day trials, primarily because the trainers did not like working in the rain. Working in the snow was not fun, but it was funny. The cats would walk about, frequently shaking their paws. Fortunately for all concerned, we dont get much snow in the Hot Springs area.
We did a little experimentation with SIT targets; small pieces of cloth or rubber that the cats could home in on, and sit on. These SIT targets were another method of maintaining stimulus control. We did not use these very much. We found them unnecessary for our purposes. However, they MAY BE just the ticket for pet owners. The SIT target would probably speed desensitization (because they are familiar), and they would offer another mode of control (sit on this until I tell you not to, or some such). The SIT target was a spin-off of the platform wait-training. We covered each platform with a heavy piece of muslin. When we moved some of the cats outside, we took along the muslin and not the platform just to see what the animals response would be. The cats responded well to the patch of muslin, though we did not do enough tests to prove its worth.
The pet owner could teach the dog or cat to sit on a large piece of heavy cloth. The cloth could be made smaller and smaller as the training progressed. Of course, the cloth could also be moved around a lot, thus requiring the animal to look for, locate, and properly use the cloth. Hmm! Does this sound familiar? Yep, it is a target, except a target for the rear end rather than the front end - but it works nonetheless. We have used such rear-end targets to tell different animals where to sit on stage, or elsewhere. When the on-stage handler said GO HOME, each animal would look for its unique piece of cloth to sit or stand on. Goats and pigs can be a problem because they sometimes try to eat their rear-end targets which is awkward for the showperson during a performance (HA HA).
The piece of cloth, or whatever is chosen, offers a form of sanctuary, IMO, in a constantly changing world. If a pet owner carried such an item about with him or her, and used it appropriately, I think their dogs or cats would be less prone to be a bother to others, and would be less frightened of new situations. I would suggest that those having problems controlling their dogs when in strange environments consider using such a device during their desensitization program. This device, or any device, is NO SUBSTITUTE for a well planned and executed desensitization program - they are only aids to establishing and maintaining control.
Now, why all of the explanation about what we did, how long it took, and the results, if all I wanted to do is to tell how to desensitize. Saying something, and doing it are two different things. I have described just a few of the many steps that we took to desensitize na#ve cats. When you begin to desensitize your animal, it will probably already have many behaviors that you want to extinguish, or replace with other behaviors. You may find you need many more steps to the training then I have described. Use as many steps as is necessary to assure that the animal will be successful. Be prepared for LOTS of trials - and I mean hundreds and hundreds of trials, maybe thousands of trials. During all of those trials, each reinforcement is important. You cant slough-off a single reinforcement. And, remember what I said earlier, every reinforcement wont be right. It is your job to make as many as close to being right as possible. If your problem is a dog chasing squirrels, you must establish your control by a combination of desensitization and stimulus control (doing what you want, when you want) training. If the dog has been chasing squirrels for the last 5 years, how long do you think you must work on the problem to overcome it? You might ask the question, is it even worth the effort? What is wrong with a dog chasing squirrel in the great scheme of things? <G> Seriously, pick your battles carefully. Dont waste your time by killing ants with howitzers. If you want to keep your dog from chasing other dogs, or just running off every time there is the slightest distraction, there may be good reason for a concerted desensitization and stimulus control program. BE PREPARED FOR A MAJOR EFFORT. If your animal has had a bad behavior for years and years, dont expect it to go away overnight, or even in a few months. Be prepared to learn how to train well, and to train consistently and persistently. None of this once or twice a week session stuff. If your dog simply does not pay attention to you (attention, by the way, is usually earned. If you dont get any, you probably dont deserve it, in the eyes of your pet), you must learn how to make that animals world revolve around yours (not the other way around).
If you have not target trained yet, that is one of the best ways to start. You can get one of Gary Wilkes target sticks, or use a yard stick or anything else similar, and get the animal to touch one end with its nose. I don't mean to sound like I am talking down to anyone, but PAY ATTENTION TO WHEN YOU SHOULD, AND DO, SOUND THE CLICKER. And, folks, please use a clicker. I know there are all sorts of very good trainers who say it is OK to use a verbal GOOD, YES, OK, etc. However, with all of the lack of precision already in the system, including most peoples lack of experience, the inherent sloppiness of words just compounds the problem. When you, and your dog, become experienced with OC, then consider the more relaxed verbal bridging stimuli. Your attitude should be, and your behavior should reflect, HOW CAN I BECOME MORE PRECISE, HOW CAN I BECOME MORE ACCURATE, HOW CAN I REWARD MORE, rather than to just search for easier ways.
Once you have the target stick where you can take your pet anywhere with it, try the tail-end version. Teach your dog or cat to sit on a large piece of cloth. As the animal, and you, get better at this, gradually shrink it down in size. Do this training in familiar areas indoors. Gradually (VERY gradually) move elsewhere indoors. Then move outdoors, around the yard, and then all over the neighborhood. Finally, after a long while of doing this, you can use shopping at the market as the opportunity to lay out your target cloth and sit your pet down, and walk away. This may take 6 months or more to get this. DONT BE IN A HURRY. If you fail the first time you try a new area, you were a LONG WAY from being ready. And, remember, it is the number of reinforcements that mean more than elapsed time. If it were otherwise, we would train once a day, and reinforce only one time, or maybe even not train at all. We would just wait for time to pass and the animal would suddenly do what we wanted. <G>
Training is a proactive exercise - it is the trainer who makes good (and bad) things happen. To get certain behaviors from an animal, the trainer must give certain behaviors. If a trainer gives random behaviors, the trainer will be paid in kind by random behaviors. If a trainer attempts to coerce behaviors, the returned behaviors may or may not be what are desired. This is especially true when the coercion begins to excessively stimulate respondent behaviors - which, of course, is the exact opposite of the desensitization process I have just been describing. If your methods include a great deal of punishment, do not expect these desensitization procedures to work very well - the cats eyes will be wild, and you havent even started yet. Before you can expect your pets to be calm (as calm as a hyper BC can be, anyway), you must calm yourself, and get your own act in order. Then, and only then, can you allay the animals fears, and gain the animals trust, confidence, and be in control.
Enough for this post. We spent more than 10 years on projects specifically related to stimulus control and environmental desensitization. At various time we used all of the tools made available by OC, P+, P-, R+, R-, separately and in combinations. In general, during early phases of training, IF THE ANIMAL IS CONFINED, there seems to be wide latitude in the type of OC procedures that are effective in establishing control. However, it is in the outside world, under stressful conditions, where the power of R+, and, to a lesser extent, R- and P-, seem most powerful. Almost all of our final systems used only R+. These training methods, without any special complicated gadgetry, can be useful to other trainers, even the pet owner. However, to be truly effective, the aspiring trainer should be willing to follow the rule of THE QUALITY GOES IN BEFORE THE LABEL (that says <finished>) GOES ON. To get quality, the trainer must be prepared to reinforce lots of good behavior. By now, the reader should have at least an idea of what I mean by LOTS of good behavior. We were rewarded by lots of good behavior. I believe you would be so rewarded too.
Copyright of all posts is the property of the original author. Please obtain permission from the original author before copying, quoting, or forwarding.
List and Site Owner: Melissa Alexander, mca @ clickersolutions.com