Someone messaged me on youtube strongly criticizing my methods of training (namely the use of positive reinforcement). They suggested that using treats as a reward is bad because it leads to excessive starvation of the animal and that it is ineffective because the animal will not perform if it isn't hungry or if treats are absent. The person went on to boast that they tamed two, let's just say very small, birds without the use of positive reinforcement at all and suggested I see their videos. Without getting into it too much, basically they exclusively relied on negative reinforcement and flooding.
I was shocked that someone would advocate training solely based on forceful means so I want on to write a very long response detailing the merits of positive reinforcement training as well as analyzing different training methods in general. I figured it would be worthwhile to share this answer with my forum members when I realized it turned out so detailed so in my response I asked the person for permission to use their question in conjunction with my response publicly. They absolutely refused and said that I am not allowed to use anything they said or even quote them. So I'm just going to roughly paraphrase what they said merely for the sake of being able to lay out my arguments here for you because I think they are somewhat profound and thorough. When the person responded to my explanation they basically just said that they think I am wrong, that they could write a dissertation justifying their training method, and that they no longer wish to discuss this subject with me. Must have really put a nail in the coffin that they could not provide a single counter argument except that some experts thought their birds were tame.
I read this by phone to a psychology professor I know and at first he thought I was being too harsh but by the end of the piece he complemented me on being very detailed and thought it presented the case for positive reinforcement very well. So without any more blabbering, with only a few alterations from the original, here is the case for positive reinforcement training:
You are mistaken. I understand that you are trying to be nice and sympathetic to the animal, however, you understand little about psychology. Look at people. Most people are starved for something they want. For many, it is not food but other material things they want. This is why they go to work, so they could earn money for food, clothes, house, car, education, computers, etc. Watch someone win the lottery and they quit their job because suddenly they can satisfy all their wants.
A parrot wants relatively little. Food is a big one. Social interaction, petting, attention, vocalization, and mating are some others. Many of those (mating for instance) we cannot provide to them as reinforcement. The other things are all subjective and qualitative. Sometimes they want interaction and sometimes they don't.
There is also punishment. You can punish an animal for not doing what you want. In the long run this doesn't work for two reasons:
1) Extensive use of punishment extinguishes exploratory behavior which is necessary for teaching new behaviors
2) They become desensitized to the punishment and you constantly need to use more till the point that they are injured, dead, or fight back
Furthermore, with parrots, they have a natural instinct to flee punishment so it would be impossible to be able to get close enough to a parrot to train it if it is constantly fleeing. That rules out punishment. Negative reinforcement is also very limited as a training tool for animals because (except for a few naturally existing aversives) you have to create an aversive to give yourself an opportunity to remove it. Imagine someone with arachnophobia. So you throw them in a room with spiders so you could be a hero and save them once they do something you want. The problem is (not only that it's unpleasant) is that eventually they'd be cured of the phobia so being saved would no longer be so desirable. Punishment and negative reinforcement wear out very quickly.
This only leaves positive reinforcement and extinction as training tools. Extinction doesn't create new behavior so that is out as well. With just positive reinforcement remaining as a training tool, you can only give an animal what it wants in return for what you want from it as a trainer. Toys, attention, talking, petting, etc are only desired on occasion and in small quantities. They are unreliable and not absolute. Food is the only consistent, quantifiable, and continuously desired reward you can give to an animal. You know they want it today and will continue to want it tomorrow. It's the one thing that does not wear off. It may wear off temporarily once the hunger is satisfied, that is certainly true. But in a few hours or by the next day, hunger is guaranteed to come back. Desire for attention, a new toy, some petting, etc can be sporadic. There isn't even a guarantee that it will want any of those from you at all but even if it does, there is no way of knowing when it will want those again. These are completely unreliable for training.
As for me, I weigh my parrot daily. Scratch that, I weigh her every time she comes in and out of the cage. I have done this so much that I don't need to weigh her anymore and can guess what weight she is at with reasonable accuracy. I don't weigh her food out anymore, I can tell at a glance how much food to feed her so that she would be at the desired weight come next training session. I have basically found her "empty weight." I never take her down to or below that weight. I take her to just about the point when she is close to empty. Then she is obviously most motivated to "search for food." In the wild this would mean flying to a place where food is available, for you it would mean going to the fridge to get something to eat, to her it means doing cued behaviors. Every living thing does something to make itself available for eating when it needs food. Parrots, dogs, humans, and other social animals are capable of advanced learning and perform complex behaviors that achieve satisfaction of hunger. While dumber animals might only eat the way they know how to eat and no other way, parrots are natural problem solvers and will try whatever it takes to get food out of a complex situation. I just manipulate the "puzzle" to get the food out of as cued trick behaviors. In the wild they might have to chew through a certain type of branch, then twist their beak to get a nut out of a tight hole, then turn it a certain way in their beak, crack it, then catch the nut while dropping the shell, and then chew it up. With me the rules of the game are changed to whatever behavior I am trying to teach. I'm merely adapting the environment so that the parrot would use its food seeking intelligence to performing behaviors that I want to get that food. So instead of foraging in a tree, my parrot knows to lift her foot when told to wave, to say hello when told hello, etc. Basically the concept is completely natural and the behavior is contrived. However, this ability makes them very adaptable and capable of survival in the wild. If a certain tree developed a certain mutation from preventing its seeds to be eaten, you can bet that the parrots would be quick to figure out how to get to them anyway. Meanwhile the specialized birds with beaks evolved for only a certain kind of food might go extinct if they don't evolve the necessary adaptation. However, unlike genetic evolution (which only applies to future generations), learning applies to the existing members of a species as well. Furthermore greater amounts of accumulated knowledge can be passed through learning which cannot be inherited through hereditary means. If parrot A learns to get nut out of a certain tree and no other parrot knows how, they can watch parrot A do that and be able to do it too. Meanwhile parrot B learned to get a fruit from a different kind of treat. Parrot A can learn to eat the fruit, B can learn the nut. Thus both A and B would be capable of having a more varied/available diet. While I am fully aware that I am manipulating the "terms" of the game and making the parrot do cute behaviors rather than forage in a tree, I believe it is triggering the same natural learning mechanisms they use in the wild. In fact, I believe that people who DO NOT train and/or provide foraging challenges for their parrot are damaging their parrot psychologically.
In fact, from taking a glance at your videos (and I apologize because I can't have sound at the moment and I only glanced through a few of them, so if i missed something contrary to what I say I apologize in advance) I noticed that you primarily rely on flooding through negative reinforcement. Aside from punishment, this is one of the worst ways to train an animal. The reason the bird does something is because it learns that there is no use putting a fight because you will eventually hold it anyway. It also learns that you are not a threat. However, it doesn't get anything good out of it. It only does it because it doesn't have a choice. On the contrary, when I use food based positive reinforcement, my parrot CHOOSES to do the behavior. She WANTS to do the behavior because there is something in it for her. If she is hungry, she will fly over to me and start doing tricks to get my attention because she knows that doing those behaviors makes food come.
For most of my taming and long term "good behavior" so to say, I use alternative methods. For instance I taught my parrot to say "hello" to get my attention rather than screaming. It's about the only thing she knows how to say so I stuck with that. In the evening when she finished her meal she would screech and scream to get my attention to come and cover her for the night. I don't come when she does that but I come when she says hello. So now she says hello to get my attention rather than screaming. This is nice and it works well. However, this isn't a complicated or energy intensive behavior. I challenge you to teach an animal to complete a puzzle NOT for food. I doubt an animal would find it worthwhile (especially the weeks of practice until it starts getting it right) for nothing but attention.
You asked if I am tempted to starve my parrot. I am not sure what you would mean by starve so I cannot say in that context. What I can say is that I have established 113g as her most motivated training weight and 110g as completely empty crop weight. I always set my goal to have her at 112-116g for training. When her weight is below 113g, I noticed that she is overly eager to complete behaviors and she doesn't pay enough attention. She will rush the behaviors or do all behaviors just in case rather than paying attention and performing the behavior deliberately. So I try to get her hungry but not too hungry or she does badly. Think of people. When they are hungry, they can think of what they want to eat, prepare it, cook it, etc. Someone who's too hungry will just grab the first thing they can get from the fridge or cupboard that doesn't require cooking and just eat it. I always try to get her to that hungry but thinking rationally stage. I have no interest in making her desperate and it doesn't help the training anyway.
Let me address some of your points specifically:If training using treats the animal will not perform if it is full or not hungry. (Their quote replaced with my paraphrasing)
The animal will not perform when you run out of other things it wants too. My parrot will often go through 10-30 treats per training session. She might only desire petting/praise/attention for 5 times max. Think of a back massage. You can enjoy one, maybe even two. Getting 10 in a row for 3 hours and you'd hate it. I train right before her meals when her hunger is at its peak. I can train her for up to her entire meal's worth of calories before she is full. I do not do that because I don't want to be training her for hours and because I want her to eat more nutritious foods for her meal. I don't run out of hunger when it comes to training. It's much quicker to run out of non-food reinforcements though.
Furthermore, my parrot will still perform behaviors even if she is not seeking food. I may not be able to get the same number of responses or accuracy of responses as when she is hungry but she does offer many of the behaviors (particularly the easy ones) when she isn't even hungry. The difficult behaviors usually require a bit more motivation.People will get the impression that providing progressively less food will improve performance (Their quote replaced with my paraphrasing and written in much clearer English not to mention )
Actually no. If you simply wait to train at the time the animal is hungry, then it will work for food. It is a matter of time rather than quantity. The only reason I measure out how much food I have is to get the timing to work every day. If I gave her an unlimited portion, she'd eat enough food to hold her over for 2+ days and that would defeat my goal of daily training. However, if I let her stuff herself one meal, if I wait long enough she'd come back to hungry enough to want to train.I don't know anything about clicker training but wonder if it can eventually replace food treats (Their quote replaced with my paraphrasing)
No, the clicker is merely meant to serve as a bridge between the behavior and the reinforcement. I go further and use it as a secondary reinforcer. When I click, it's just to tell the parrot that she did what I wanted regardless if she gets food.We never give our bird food rewards but he does what some people would consider to be tricks (Their quote replaced with my paraphrasing)
I wouldn't call those tricks. The bird is just playing and in social species play is a normal way of practicing useful skills, bonding, and performing exploratory behavior. The things that I call tricks in reference to training are consistently reproduce-able learned behaviors that exert energy and that can be triggered by a particular stimulus. This is how I differentiate taming. Taming I would consider to be learned behavior that does not require the animal to exert energy. Allowing itself to be grabbed for instance is a matter of taming. It merely has to NOT resist rather than make an effort to make something happen. If you cannot make the bird do the same behavior in the same manner several times in a row and for different days/sessions, then I would not consider it a trick. Also I would not consider self learned or innate behavior to be a trick. I would only consider a trick to be a natural or unnatural behavior that has an unnatural stimulus that consistently triggers the behavior. Parrots like to stretch their wings (one leg and one wing) once in a while. That is not a trick. If you could invoke it to do that on demand, then I would consider that to be a trick or a "trained behavior." Then you can take it to another level and teach unnatural behaviors with unnatural stimuli like playing dead.
The last thing I want to say about food based reinforcement is that it is objective. Anyone can cue my parrot to do a trick and she will do it for them for some food. If I could no longer care for my parrot and someone inherited her today, I am fairly confident that if they would follow my methodology, they would be able to get the parrot to respond to them, come to them, do the tricks, and do everything that she knows. However, non-food based reinforcements are shotty and are unlikely to be reproducible by other people. Can a stranger (to your bird) come to your house and just pick the bird up? I highly doubt that. However, with my parrot, as she is trained to respond to certain cues, can be called by anyone. Strangers can call her name and she will fly and land on their hand for a treat. She does not have to see the treat before hand. She is not being lured onto their hand for food. She is just familiar with the cues and can generalize them to other people. However, by the methods with which you tame your birds, they merely learned to not be scared of you, but that does not necessarily go for other people. Lastly, I use a variable ration of reinforcement schedule so I am able to get my parrot to perform a fair number of behaviors without a reward, merely in anticipation that somewhere along the way one of those behaviors will earn a reward. I can often get her to perform 50 behaviors in a training session for merely 20 treats. I can make that limited hunger span a long way by using an unpredictable reinforcement schedule.
As a trainer, you must ask yourself whenever you want the parrot to do something "what's in it for me" from the parrot's perspective. If they have nothing to gain for it, they will not do it.
I wrote this long response because you opened a can of worms. You made some implications which I strongly believe to be misguided. I have read countless books on parrot training, bird anatomy, human and animal psychology, etc. I studied psychology for 4 years in college and am still in contact with a psych professor discussing the implications of operant conditioning and behavioral studies. My vet thinks my parrot is in excellent shape and just the right weight. She is healthy and strong. Her flying gives her good exercise and mobility. I think you are misguided by ignorant assumptions on the basis that "it's not nice" to deprive someone of food rather than thinking about psychology, motivation, and why we do things. We go to work because we need to earn money and not because we love to shovel dirt or answer phone calls. Certainly many people enjoy their work to some extent but the reason they do it is for food (and other human things they want but food as well). My parrot enjoys training and on occasion she will preform her tricks on a completely full crop. However, for the sake of maintaining a repertoire of over 30+ behaviors and continuously teaching hew new ones, food reinforcement is the most effective way of doing this.
If you don't mind, I'd like to forward this discussion around to some of my followers and possibly post it on my forum for the benefit of other people seeing my response to your points as well because this is an important question. I won't mention you by name (unless you want).
I'm not trying to scold you or make you feel bad by any means. You came on strongly and challenged my (highly contemplated and deliberate) methods so I felt I had to explain them.To close, I wanted to mention that while I do believe that flooding, punishment, and negative reinforcement have a place in parrot training, I absolutely DO NOT think that they should be used as the primary means of training. I'd like to see this person try to work with an aggressive macaw solely by poking it until it steps up rather than boasting about forcefully taming a bird that fits entirely in your hand.