Post date: Aug 20, 2019 1:06:37 PM
Don’t Shoot the Dog by Karen Pryor
(03) Stimulus Control: Cooperation without Coercion
Anything that causes some kind of behavioral response is called stimulus. Sounds, lights and scents are called unconditioned or primary stimuli. Other stimuli are learned by association with a reinforced behavior. Cues or signals are learned when the behavior associated is reinforced. If the response is delayed, the response is not under stimulus control. Positive reinforcement must be established for the cue as well as the response to that cue.
Establishing a Cue. Shape the behavior before the cue. Once the behavior is secure, shape the behavior just before or right after some stimulus. The cue becomes a sort of green light.
The cue may be given as the behavior is starting to reinforce the completion of the behavior. Then repeat this sequence at different times and in different locations, gradually backing up the cue time until the cue comes before the behavior starts. By and by the learner will identify the cue as the cue for that particular behavior continues to be reinforced.
A second method is to alternate between the cue and no cue. Say “sit” and reinforce it with a click. Let a ‘sit’ or two go by. Then say “sit’ and reinforce sit. In the same session we are reinforcing on cue sits and extinguishing off-cue sits. Once the learner understands the rules, new cues can be attached to new behaviors practically instantly.
A third way to add a cue is to shape response to the cue as if it were the behavior itself. When the dog is ‘throwing sits’ is the perfect time to introduce the cue.
Get out the clicker and treats. Say sit as the rump lowers, toss the food so the dog has to get up to eat.
Then, ‘sit’ and click before the sit is complete.
Next, click when the rump is all the way on the ground to avoid ½ sits.
The next step is to intersperse some well learned behavior in between bouts of giving and reinforcing the new sit cue.
The last step is to shape the behavior of waiting for the cue-½ a second, a full second, 3 seconds--until the dog is visibly attending to you and not offering the behavior until the cue comes. Once this is accomplished all auxiliary cues can be phased out, leaving just the word. At this point the cue response has been developed as an operant behavior, intentionally offered in the hope of gaining reinforcers.
The Rules of Stimulus Control. Bringing behaviour under stimulus control is not accomplished until the behavior is also extinguished in the absence of the conditioned stimulus. In the training or working situation, where discriminative stimuli (cues or signals) are going to be used, the ‘go’ and ‘no go’ aspects must be established if performance is to be reliable. Complete perfect stimulus control is defined by four conditions, each one may have to be approached as a separate task, a separate item in the training recipe. When all four conditions are met, the dog finally understands what ‘sit’ means:
The behavior always occurs upon presentation of the conditioned stimulus (the dog sits when told to)
The behavior never occurs in the absence of the stimulus (during a training or work session the dog never sits spontaneously.)
The behavior never occurs in response to some other stimulus. (If you say “lie down” the dog does not offer the sit instead.)
No other behavior occurs in response to this stimulus. (When you say sit the dog does not respond by leaping up or lying down.)
What Kind of Signal? A learned signal (discriminative stimulus) can be anything that the subject is capable of perceiving. Any signal the subject can sense can be used to cause learned behavior to occur.
Dogs respond to hand signals and voice commands. Humans respond to just about anything. It is useful, in a working situation, to teach all subjects the same cues and signals, so that other people can cue the same behaviors. Signals are mere conventions.
While we don’t want more than one behavior occurring on a single stimulus, it’s perfectly feasible to have several learned signals for one behavior. Establishing a second cue for a learned behavior is called transferring stimulus control. To make a transfer, present the new stimulus, then the old one and reinforce the response. Gradually make the old stimulus less and less obvious while calling attention to the new one by making it very obvious, until the response is given equally well to the new stimulus, even without giving the old one at all. This process usually works out faster as the behavior has already been established.
Signal Magnitude and Fading
Learned cues or signals do not have to be of any particular volume or size to get results. A learned cue simply has to be recognized to get a response. Once a stimulus has been learned, it is possible to make it smaller and smaller, until it is barely perceptible and still get the same results.
One result of establishing stimulus control is that the subject must remain attentive if it wants to get reinforced for responding correctly.
Targeting
A physical target can be a very useful type of discriminative stimulus for all sorts of learners and behaviors. Teaching an animal to touch the end of a stick with it’s nose is an excellent beginner exercise. The behavior can be seen and felt so it’s easy to reinforce, and easy to see how to raise criteria in steps: two inches from the nose, four inches, to the left, to the right, up, down and then forward until the animal is following the stick around. Use targeting with tricks and treats.
Use a target to teach a rambunctious out of control dog to walk nicely in heel position. Use longer and longer stretches of “keep your nose here for a click and a treat.”
Modern trainers are able to accomplish all the traditional behaviors that used to be trained by force using positive reinforcement (a word or a click.) A discriminative stimulus that is a cue for avoiding an aversive can not only reduce any need for physical control or intervention. The stimulus can be completely neutral so long as the association is distasteful. it serves as a conditioned warning signal, which teaches the subject to cease the unwanted behavior prior to any distasteful experience.
Limited Holds
A very useful technique for getting a prompt response to a discriminative stimulus is the limited hold. If the subject has learned to offer a behavior in response to a cue, but there is usually some gap between presentation of the stimulus and the subject's response, this interval can be shaped downward using a limited hold. Start by estimating the normal interval in which the behavior occurs then reinforce only when the behavior occurs during the interval. Any responses outside the interval no longer earn reinforcers. Gradually all responses will fall within that interval. Then, we can tighten the screws again, staying within the range of where the behavior is occurring most of the time. Both animals and people have a very sharp time sense and will respond to limited hld training with dramatic precision.
Use the clock. With humans, don’t discuss what you are doing (no argument) . Just do it and watch it work.
Limited holds are simply the amount of time we are willing to wait for a request or instruction to be carried out.
Anticipation
Once the cue has been learned, the subject is so eager to offer the behavior, it acts before the cue has actually been given. One way to cure anticipation, is to use ‘time-outs.’ Stop all activity when the cue is anticipated and undesirable. Give no cues for one full minute. Every time the subject jumps the gun, reset the clock. Penalize overeagerness by making it the cause of the delay for the chance to work.
Stimuli as Reinforcers vs. Behavior Chains - Once a conditioned stimulus is established, it becomes a reinforcer. A discriminative stimulus signals the opportunity for reinforcements, so it becomes a desirable event. A desirable event in and of itself is a reinforcer. We can reinforce a behavior by presenting the stimulus for another behavior. Long strings of behavior are called behavior chains (behaviors involving many known steps.) Unlike simple long-duration behaviors, these can be maintained comfortably, without deterioration or delayed starts, because each behavior is actually being reinforced by the signal or opportunity to perform the next behavior, until the final reinforcement of a job is completed.
Homogeneous chains are chains in which the behavior is repeated over and over again.
Heterogeneous chains are various different behaviors that are reinforced only when the last behavior is completed.
The pattern of a sequence is not essential to the nature of a chain. What is essential is that the behaviors in the chain follow each other without a time gap, they are governed by cues, either from the trainer or the environment, and that the primary reinforcer occurs at the end of the chain. What makes behavior chains work is that each behavior has a history of reinforcement, and each behavior is under stimulus control or on cue. A discriminative stimulus signals the opportunity for reinforcement, so it becomes a desirable event. The desirable event in and of itself is a reinforcer. We can reinforce a behavior by presenting the stimulus for another behavior. The learned cues maintain behavior within the chain. The cues can be given by a handler. The cues can also be provided by the environment. Sometimes the cue for the next behavior consists of the previous behavior. Many things we do every day are behavior chains. Behavior chains are useful and powerful.
Misbehavior is often just the result of a chain breaking down. Incorrect events are the result of the trainer’s failure to build or maintain a behavior chain. Behavior chains break down and behavior goes to pieces if there are unlearned behaviors in the chain or behaviors that have not yet been brought under stimulus control. We can’t reinforce the subject with a cue if it doesn’t recognize the cue or if it cannot accomplish what the cue indicates.
Behavior chains must be trained backwards. Start with the last behavior in a chain. Make sure that it has been learned and the signal to begin it has been recognized. Then train the next to last one, etc. Treating memorization tasks as behavior chains not only shortens the needed memorization time considerably, it also makes the whole experience more pleasant. When making a cake, the frosting goes on last, but to teach a child to enjoy making a cake, start by asking for help with the frosting.
Teaching a Dog to Play Frisbee - Frisbee allows the owner to stand in one place and still run the dog’s legs off. There are two training problems in this game: the distance the dog goes after the Frisbee must be shaped. The second is that the game is a behavior chain: first the dog chases the Frisbee, then the dog catches the Frisbee, then the dog brings the Frisbee back for another throw. Each behavior must be trained separately, the last behavior in the chain, retrieving, must be trained first.
Retrieving can be taught on very short distances, even indoors. Most dogs will drop the item almost spontaneously but those dogs who prefer tug o’ war may need to be shaped to release it. When the dog begins to carry things to you and giving them up on cue, s/he is ready to learn frisbee.
First, get the dog excited about the frisbee by waving it in his face. Let her take it and bring it back, praising her madly for returning it. Hold it in the air, letting her have it when she leaps for it. Then toss it momentarily in the air, making a big fuss when she catches it. Once she has the idea, toss it up and out a few feet so she has to chase it to catch it. As the distance grows longer, the dog needs to learn to watch the frisbee and place herself well for the catch. Once fatigue has set in, as evidenced by dropping the frisbee halfway or returning slowly, it's time to end the session.
Generalized Stimulus Control - After learning 3 or 4 cued behaviors, most subjects seem to recognize that certain events are signals, each signal means a different behavior and acquiring reinforcers depends upon recognizing and responding correctly to signals. From then on, establishment of learned stimuli is easy. The subject already has the picture, and all the subject has to do is learn to identify new signals and associate them with the right behaviors.
As trainers, we help by making it very clear, so training can go much faster than the initial laborious steps. As a teacher, use a variety of reinforcers besides good grades and teacher approval, to include games, peer approval, early dismissal and free gum. As long as the learner knows that we mean what we say (keep promises) and that it pays off to what we want, he/sher generally becomes responsive and attentive.
Pre Learning Dips and Tantrums - The pre learning temper tantrums, dips or altogether dips in performance - In humans, the pre learning temper tantrums often seem to take place when long held beliefs are challenged and the subject knows deep inside that there is some truth to the new information. The pre learning temper tantrum is a strong indicator that real learning is actually finally about to take place.
The Uses of Stimulus Control - Most of the time there is no need to boss the world around. Let the children dawdle if not in a hurry. Employees who are already working hard don’t need more orders and instructions. Unnecessary rules and regulations only breed restrictions. Stimulus control is involved in producing cooperative children, obedient pets and reliable staff members. There is a certain amount of satisfaction involved in responding to elaborate sets of learned signals. Once one has mastered all behaviors and signals, executing the responses brings a lot of reinforcement. It is fun to perform.
Approach perfection by rewarding improvement in that direction. Real, elegant, stimulus control established through the use of shaping and reinforcers may produce something interpreted as discipline in the subject. The reality is that the trainer is the one who has to become disciplined. Knowing how to get stimulus control without resorting to uproar or coercion, makes life a lot easier for trainers and subjects alike. Learn to establish stimulus control without unnecessary escalation.
A disciplined understanding of stimulus control means avoiding giving needless instructions or orders that can’t be obeyed. This means you don’t make requests you are not prepared to follow through on and you always know exactly what to expect. You don’t fly off the handle at a poor response. You don’t nag, scold, whine, coerce, beg or threaten to get your way, because you don’t need to. And if you ask them to do something and they say ‘yes,’ they do it. When the whole household, organization etc. is working on the basis of real stimulus control--when all the people keep their agreements, say what they need and do what they say--it is perfectly amazing how much gets done, how few orders need to be given and how fast trust builds. Good stimulus control is nothing more than true communication--honest, fair communication. It is the most complex, difficult and elegant aspect of training with positive reinforcement.