Negative reinforcement is perhaps the most misunderstood of the methods of operant conditioning. Although negative is used in the term, it does not mean “bad.” Instead, negative means “negating” or taking something away. Reinforcement on the other hand refers to increasing behavior. Again this neither means good or bad but just means that the behavior will recur with greater frequency.
In the game of animal training, increasing desired behavior is generally the goal so reinforcement is to be used. But the question remains as to whether it ought to be positive or negative?
When it comes to whether an animal “wants” to do something, it should make little difference whether the reinforcement be positive or negative. Avoiding something aversive may be as, if not more, desirable as gaining something nice. Pretty much any behavior that relates to safety is going to be based on negative reinforcement. There is nothing bad about being safe and in fact it is a good thing.
So why does negative reinforcement carry a bad rep? Part of it is the misunderstanding of the word negative, where in this context it does not mean doing something bad. Many people confused negative reinforcement with positive punishment. Based on common language it would seem that negative reinforcement should be the opposite of positive but based on psychological terms that is not the case. Let's not get carried away with technical definitions and focus more on the meanings.
The main reason negative reinforcement is perceived badly is because professional trainers (or tamers as they used to be called) would intentionally introduce aversives, pressure, or pain to animals for the sake of being able to stop causing it as a reward for desirable behavior. This is how horses, donkeys, camels, elephants, and many other animals have been trained for thousands of years. Basically it would involve beating the animal a lot and letting it learn that if it would do what it was commanded, then the hurt would stop.
If a trainer walks around with a stick, hook, or whip, its very presence symbolizes negative reinforcement and that if the animal stops doing what it is supposed to, then it will be used. What's bad about this? If the animal always does what it is supposed to, the instrument won't even be used. The problem is that the animal is acting simply out of fear and not out of a genuine desire. This does not lead to a great relationship between animal and trainer. If the trainer were to stop carrying around the negative reinforcement instrument, extinction would begin to occur. Whenever the animal would slip up and not receive the normally expected beating, the animal would realize that it no longer has to do as the trainer commands. This is why this type of negative reinforcement is neither truly effective nor nice.
But just because professionals misuse negative reinforcement, does not mean that it is all bad. The professional does not seek to have a personal relationship with the animal like a pet owner would. The professional needs little more than for the animal to do its tricks on command in front of spectators. On the other hand, the exotic pet parrot owner seeks desirable behavior throughout the day and a fantastic relationship to go with it. Thus clearly a relationship based on fear cannot be the solution.
Negative reinforcement is frequently misused in the parrot community although most are unaware they are even applying it. “Stick training” where a dowel is shoved into a parrot's belly until it steps up is an example of the misuse of negative reinforcement that is neither ideal training nor good for the relationship. The parrot learns to step on a stick before it gets nudged. But what happens when a hand is substituted for the stick? The parrot may opt to bite the hand instead because unlike the stick, it isn't inanimate. The parrot doesn't learn to always step up but instead learns to step when a negative-reinforcement instrument is used.
If you aren't scared of a bite and let your arm be like an inanimate object, the same effect can be achieved and the parrot will even learn that biting is futile. But the arm becomes just as much the negative-reinforcement tool as it a part of your body. This is conveniently tempting because unlike treats, you can have this (your arm) with you all the time. The problem is that if the parrot has any alternative to reluctantly complying (such as flying away, clinging to the drapes, running under the cage, etc), it may choose to do those instead because they are less intimidating/painful than the arm into belly shove. The arm is both the thing you want the parrot to trust/like and the instrument of aversion which creates a dilemma. The parrot learns to step up onto the arm to avoid the arm. Thus this example of the classic step up routine, really is a parallel to the circus days of animal training.
If a parrot's wings are not clipped, it is difficult for the trainer to be able to apply negative reinforcement because the bird will sooner fly away than put up with threats to do a behavior. The old school approach of clipping, flooding, and trainer induced negative reinforcement fail to drive cooperation when a parrot is no longer clipped which makes the owner choose to clip the wings again. Genuine success cannot be achieved this way.
I'm not going to get into the more successful positive reinforcement based method to training, you can find the details of this in my book, The Parrot Wizard's Guide to Well-Behaved Parrots. Instead, I'd like to write a bit about the positive side of negative reinforcement. How can negative reinforcement be a good thing?
Well, negative reinforcement is the increase of behavior through reduction in aversives. Thus any means of providing safety is essentially negative reinforcement. Safety is the reduction of danger and thereby a sought thing. There are naturally present dangers that drive fear in a captive parrot's life. By providing safety from these naturally present aversives, we can both provide comfort to our parrot as well as receive some training benefit out of it.
I do not believe that parrot owners should be intentionally causing harm to their pets in order to get to rescue it. But if the aversive already exists, then why not take some training advantage out of it? In fact, capitalizing on these negative reinforcement opportunities should help strengthen the bond rather than harm it. If the parrot realizes that it can receive protection from its fears through you, that will actually improve your relationship. This works as long as those aversives are not created (or even perceived to be created) by you.
For example if there is a loud disturbance going on outside, by stepping up for you, you bring your parrot to a quieter room, the parrot would be negatively reinforced for stepping up. The scary disturbance would be reduced by action you took to protect the parrot and the parrot would be more likely to step up for you in the future.
Here are some other examples from my experience. I take Kili & Truman to the park regularly to fly on harnesses or free fly. Sometimes the kids become too bothersome and I can tell the bird is getting a bit flustered. I will offer the bird a chance to flight recall to me and stay on me to avoid further confrontation. I don't have to give a treat for this type of flight recall because the bird gets reinforced negatively in the process.
Another recent example was when I took Santina to the vet for the first time. I was still in the early stages of gaining trust with her and using a lot of positive reinforcement to encourage her to step up. After Santina came back from anesthesia, Lorelei put Santina down on the floor. From the other end of the room I put my arm down. I did not go over to get Santina but instead she walked over to come to me. I was not giving out treats or head scratches. Santina came to me for safety and thus negative reinforcement for coming. I would not torment her with a towel (and I'm sure the vet wouldn't either) but since it was medically necessary and going to happen regardless, I might as well take the opportunity to take credit for providing safety. This is a case of a good application of negative reinforcement. The parrot learned to come to me for safety in moments of panic rather than to flee or worse yet that the danger was caused by me in the first place.
One of the reasons negative reinforcement tends to be ineffective is because of desensitization with time. As the animal begins to get used to an aversive or pain, it may loose its effect. Especially if the aversive turns out to be physically harmless, the parrot will realize and stop responding to it. For this reason either a stronger dose of aversive is necessary or the behavior is lead toward extinction. Since I don't mind my parrots to stop fearing the naturally present aversives they encounter, I am not worried if the negative reinforcement will lose effect. If my parrots stop fearing these things, I will be just as happy because I don't want them to be fearful. I only use occasional natural negative reinforcement on behaviors that are already trained through positive reinforcement. Since I do not cause the aversives intentionally myself (and disassociate with any aversives that could be perceived as caused by me), the intensity of the aversive should not change. Further, these types of scenarios are fairly infrequent anyway. But since they happen from time to time, I capitalize on the situation and play the role of savior for my parrot.
Let me illustrate with human example. Let's say you are on a long cross country trip with a friend driving. You get an upset stomach and really have to get to a bathroom. So your friend is very understanding and gets you to one promptly to one in return for asking persuasively. We can agree that you gain nothing from going to the bathroom but you do get relief so this is negative reinforcement. You would then be grateful to your friend for helping you find relief from something uncomfortable that isn't your friend's fault. Now what if your friend put something in your lunch that caused this? Even though your friend saves the day by getting you to a bathroom, they were still the cause of your discomfort. Would you be happy and grateful to them? Probably not. You'd be more angry that they caused you discomfort in the first place. This is the difference between giving negative reinforcement to a natural/unrelated aversive vs being the cause of the aversive. Negative reinforcement can only work well when it is providing relief from something you have not caused or associated with.
So am I encouraging you to go and use negative reinforcement with your parrot? Not necessarily. What I am encouraging you to do is to give further thought into whether or not you are rewarding by giving something (positive reinforcement) or rewarding by taking something away (negative reinforcement)? Are you threatening in some way that until the parrot does something you want, you will do something? Why should your parrot comply with your requests and what will the consequence be if it does not? Being more aware and intentional in your training means will ensure that you are making the most effective application but also help you to ensure a good relationship. Making sure that you are not causing aversives but helping to reduce naturally occurring ones will work in your favor.
My book almost exclusively focuses on training using positive reinforcement because this needs to be the basis of any parrot relationship. However, this article is a supplement for those who already use positive reinforcement. This is for those who want to take their training to an even further level and learn to apply the good kind of negative reinforcement responsibly to get even more out of their training.
Avoid using threats (whether it be with a stick, your arm, a squirt bottle, or going back in the cage) and do things so the parrot would want to engage in the behavior without coercion. But on the other hand, when there are opportunities to save your parrot from uncontrolled environmental factors, take training advantage by having the parrot do something for it. Stepping up, coming to you, flying to you, etc are all important behaviors and ones that the parrot can learn to do more readily when it feels scared. By teaching your parrot to come to you rather than away, you can ensure that in times of panic, your parrot is more likely to return to you and that your relationship can be so good that you would be your parrot's means for safety. Negative reinforcement is often misused but it is not always bad. Focus on the good stuff with your parrot and your relationship will be better than ever.
This is written for people who are wondering how to punish their parrot for biting or why punishment is not recommended for training parrots. First allow me to define what punishment is in relation to training. Punishment is a consequence that reduces a specific behavior. There are two types of punishment: positive punishment and negative punishment. Positive punishment involves doing or increasing something which results in a reduction of behavior. On the other hand negative punishment involves not doing something or taking something already existent away which results in a reduction of behavior.
As a psychology definition pertaining to operant conditioning, punishment can only be defined in terms of its effect on behavior (which is the reduction of it). The trainer's attitude and perception of what is punishment is not relevant if it does not reduce behavior. Furthermore punishment is not necessarily a bad thing. It is merely the process of reducing undesired behavior in the parrot.
Let me continue and explain what punishment is not. Yelling, squirting, and saying "no" are not necessarily forms of punishment. They could be, but only if they reduce behavior. Considering that some parrots like getting wet, yelling, and attention, these things could actually turn out to be positive reinforcers that encourage behavior (and in that case would not be punishment). Retaliation, dominance, sadism, and abuse are not punishment either. Although these are terrible things and quite likely harmful to the parrot, they are not forms of punishment. These will most likely ruin any trust relationship between owner and parrot but they almost certainly will not solve any behavioral problems that lead the owner to using these methods.
It seems to me that there is a very common misunderstanding of the concept of punishment of parrots in training. The problem likely stems from the owners own self gratification of establishing a sense of justice rather than a methodical training methodology. An example can be where the parrot poops on the floor so the owner yells at the parrot. In this scenario, the parrot pooped on the floor merely because it is a parrot and parrots have the whole forest to poop over in the wild. On the other hand, the owner is aggravated that he or she will have to clean the floor and yells at the parrot. The problem is that the parrot does not understand this and is no less likely to poop on the floor yet again. Besides the fact that the parrot might even enjoy a screaming contest with the owner, the parrot also probably does not realize that this has absolutely anything to do with pooping. These leads us to reason #1 why punishment is often ineffective with parrots:
The perceived punishment is not properly coupled with the behavior and therefore does not help eliminate it.
Positive reinforcement training proves more effective because the behavior can be practiced multiple times and the connection between the behavior and the reward be more sufficiently established (for example clicker bridge). A more effective way to make the parrot stop pooping on the floor might be to put it down on a regular basis so that it could do its business on a perch over a newspaper. Further, the parrot can be positively reinforced for pooping on its designated potty perch by picking it up and giving it attention once it has pooped there instead of the floor. This way the parrot will seek to poop on the perch rather than floor whenever possible. See potty training guide for more information about the use of positive reinforcement to train parrot to poop in a designated place.
Another reason why punishment often fails is because of unintended consequences. While the trainer thinks he is punishing one behavior, in reality the parrot learns to avoid doing something else. Let's say for example the owner decides to use a timeout as negative punishment whenever the parrot starts chewing on the windowsill. The owner grabs runs over and grabs the parrot when it starts chewing the sill and then locks it in the cage for a while. The owner is content thinking that the parrot will stop chewing the sill. However, the parrot obviously finds it reinforcing to chew the sill so the parrot has lost no desire in doing it again. Instead, the parrot learns that "getting caught" is what gets it punished. Now the parrot will either chew the sill when the owner isn't watching or the parrot will just fly avoid getting picked up by the owner when the owner comes to punish the parrot. Perhaps in the beginning, the owner was able to grab or ask the parrot to step up so that the owner could put the parrot away in the cage. Now the parrot realizes that when it is on the windowsill, if the owner comes over, it will surely be punished. If the parrot is flighted it will most likely fly away from the owner and avoid the owner. If the parrot is clipped it will bite the hand in order to avoid getting picked up to be punished. This brings us to reason #2 why punishment should not be used:
Punishment can lead to unintended consequences where there is a difference of opinion between pet and owner of what is being punished.
It is easy for a punishment to get out of hand and go too far. This can happen where it goes to the point of classically conditioning a phobic response. A classically conditioned phobia can be much stronger and more difficult to eliminate than some operant learned behavior. This can happen if a punishment is traumatic. Let's say the owner decided to punish a parrot for biting by blowing an air horn (or making some other terrifying noise) the moment the parrot bites. This could arouse a phobia of the owner in the parrot. The parrot learns to associate the owner with the feeling of terrible and evasive fright. The parrot has no idea that this one time the owner happened to be holding an air horn and that it was because it bit. The parrot just develops a strong feeling of anxiety at the site of the owner. These phobias become self reinforcing as time goes on. The owner walks over to take the parrot out of the cage, the parrot is terrified of the owner and starts trying to fly in the cage. Since the space in the cage is restricted, the parrot keeps crashing into the bars and gets hurt. Now it has yet another reason to be fearful of the owner because in the sight of the owner it received more pain. This can become an endless cycle. Therefore it is very important never to create a situation (purposefully or accidentally) where the parrot can become phobic. Are you phobic of anything? Is there something makes your heart race and makes you feel sick at the mere sight of it? Imagine your parrot feeling this way at the sight of you.
Excessively traumatizing punishment can develop a classically conditioned phobia which makes the parrot feel fright at the sight of the trainer.
Punishment makes it much more likely that the parrot will avoid the owner all together rather than the unwanted behavior. Think about it. The parrot gets food, water, and toys in its cage as it is. It already has what it needs to survive. Spending time with the owner, getting attention, petting, etc are just bonuses. However, if the parrot views the owner as someone that does more harm than good to it, it will try to avoid the person all together. It amazes me that owners often expect the parrot to come to them to accept its punishment. It simply doesn't work that way. Parrots are prey animals and in the wild must avoid creatures trying to consume them. If the parrot is not getting good things from the human but only bad, why should it possibly tolerate the human's presence? Why shouldn't it bite? Why shouldn't it fly away?
Unlike dogs which are pack animals with strong requirements for social hierarchy and acceptance, parrots do not hold such bonds. Flocks are bound by common interest in protection by numbers rather than loyalty. The prospect of getting hurt (or perhaps in their mind eaten) by the owner far outweighs a couple nice things like treats and scratches. Since the rewards we use are just nice but punishment potentially life threatening, the punishment far outweighs the reward. The parrot is simply better off avoiding the human than risking further punishment.
The threat of further punishment can easily outweigh the benefit of treats and other positive reinforces commonly used in training which will make the parrot avoid human contact all together.
The only ways a parrot can avoid further punishment from the owner is flying away or biting. A clipped parrot realizes that it cannot fly away and will predominantly resort to biting. I doubt the risk of increasing aggression toward the owner is worth it in order to try to get the parrot to be quiet, not poop on the floor, or something else like that. Trying to reduce most kinds of biting through punishment will prove ineffective and just create more biting. The parrot is biting to avoid being drawn into a situation which will result in punishment again. This is why it is important to create strictly positive encounters in order to develop trust with the companion parrot rather than continuing to drive the fear which causes much of the biting in the first place. If the parrot has no benefit of being with the owner (or worse yet punishment to fear receiving from them), then surely the parrot will attempt to defend itself.
It is easy to believe that an action such as saying "no," distracting it with a toy, or locking it away in the cage punished the behavior because in the short term indeed this stopped the behavior. However, this would not necessarily be punishment but merely a temporary distraction from the bite. It can only be considered to be effective punishment if it reduces or eliminates the behavior in the long run. If you find yourself needing to punish the parrot more and more frequently (rather than progressively less), then you did not punish the parrot. In fact you may have even positively reinforced it which brings the behavior back even more frequently than it naturally occurred. As parrot trainers, our goal is the long term elimination of unwanted behavior rather than temporary cessation or vengeful reparations.
Improperly executed punishment can end up strengthening the unwanted behavior and only making it worse. Here's an example. The parrot is sitting on the owner's shoulder while the owner is watching tv. The owner is not paying attention to the parrot while focusing on the tv program. The parrot is bored and nips the owners ear for attention. The owner turns and looks at the parrot and says "No, don't bite." The owner is thrilled because the parrot let go of the ear and stopped for a few seconds while the owner believed to be punishing the parrot. However, this is not at all punishment as there is no primary punishment coupled to the secondary which is merely the words. In fact, this turns out to be positive reinforcement because the parrot learns that tugging on the owner's ear will get the owner to stop what he is doing just to vocalize with the bird together. This will only lead the parrot to tugging on ears more and more and will never make the behavior go away. Grabbing the bird and putting it away in the cage will not work either because by the time it is grabbed, carried, and put away, the parrot will forget for what it is being punished. Although shaking the shoulder to unbalance the bird might serve as an effective punishment, it can also cause the bird to become more aggressive. For all of these reasons, punishment proves to be quite ineffective and very risky. Punishment can easily ruin the parrots trust and the damage is quite difficult to undo.
It is also important to realize that not all punishment is necessarily intentional, yet the ramifications are no less than if they were intended. Reflexively dropping your parrot after biting, tripping over something and make a loud noise in front of your parrot, or putting the parrot away in the cage because you have to leave could all be unintended forms of punishment that could lead to the same fears or aggression as mentioned in earlier examples. If a parrot trusts its owner and steps up willingly, putting it away in the cage immediately after stepping up could result in punishing step up behavior and cause the parrot to resist stepping up. Even though this is not intended as punishment, it could still have the same effect. For this reason it is important to keep stepping up a positive experience the majority of the time so that it would be worthwhile for the parrot to cooperate rather than exclusively using step up when you want to return the parrot to the cage. Furthermore, you can use returning to the cage as a positive experience by providing food or a new toy. While this may be the most common situation where owners inadvertently punish their parrot, there are many other times this could happen. So try to think if what you are doing is going to make the parrot avoid doing the right thing in the future.
I am not saying that punishment is absolutely ineffective or can never be used. But I am definitely suggesting that it is very easily misunderstood and misused. It is virtually impossible to misuse positive reinforcement to the point that the parrot is scared of you, but this can very easily happen with punishment.
Since punishment can hurt the owner-parrot relationship, my best suggestion for dealing with undesired behavior is to do absolutely nothing in response to it. Do not say "no," do not yell, do not try to reason it out, do not give your parrot a dirty look, do not make a loud distracting noise, do not give the parrot a toy to chew, do not drop your hand, do not hold your parrots beak, do not strike back, do not lock it away in the cage, do not even walk away. Anything you do can (and probably will) work against you. Some of these milder forms of "punishment" are not even punishment at all. They do not hurt or displease the parrot enough to make it stop doing what it already wants to do. Others of these things are so drastic that they will make the parrot scared or aggressive toward the owner. Even still, some of these will only encourage the behavior even more (if a parrot is biting to make you go away, and you give up and go away, the parrot learns to bite even more in the future). The simple fact is that finding an effective punishment that balances being aversive enough to discourage behavior and yet not go so far as to ruin the parrot-trainer relationship, is very difficult. Extinction on the other hand is your best resort. If you do not react in any way to the unwanted behavior, you are guaranteed not to be rewarding it with attention and eventually the parrot will get bored of doing it. Along with extinction, focus a lot on prevention: do not leave important documents out, take off jewelry you do not want chewed, provide alternative toys/perches for parrot to enjoy, do not touch your parrot where it doesn't want to be touched, do not try to force a reluctant parrot to step up, etc.
I do not agree with the trainers or so called "experts" that are absolutely against punishment. Some people try to make it into an ethics debate or to anthropomorphize the parrot. Others misunderstand the training concept of punishment and just think its bad because they mistake it with retaliation or sadism. The simple fact is, punishment will happen. Whether purposefully or unintentionally, the parrot will learn not to do certain things and it's fine. The important thing is for us to realize that we cannot think in the same way that parrots would. They respond to punishment far more drastically than we or other pets do. Since we do not yet have a full grasp on the effectiveness of punishment, we as trainers need to be aware of the adverse affects of it. Someone has to be the pioneer of research and experiment with these methods surely, however, they should not be used by most parrot owners. Particularly, punishment should not be recommended to beginners because they neither have a solid relationship to build on and nor do they understand the full ramifications of using these methods. Punishment is far more likely to damage a new parrot relationship rather than one with a very positive history. Positive reinforcement based training should be the primary way to go (while ignoring unwanted behavior) and perhaps effective punishment can be applied just to clean up the remaining bits of undesirable behavior when all other methods fail.