“Positive Only” Training?

Recently, there has been a lot of misinformation spread about dog training methods labeled as “positive only”. First, let’s describe what positive only actually means. Positive only would imply that absolutely nothing negative is involved in the methods. Operant conditioning and clicker training are often lumped into the criticisms concerning “positive only” because of their methods of avoiding aversives. While the clicker training that Karen Pryor has made popular excludes all punishment, it is still a highly effective method for teaching animals. It is often joked that the methods used to train dolphins and wild animals to perform tricks is not applicable to real life scenarios, as without punishment it is like lacking the “cold” part of the “hot/cold” game. However, these are false assumptions that operant conditioning, the very training used by these professionals training dolphins, does not use positive punishment. Properly applied positive reinforcement is so effective and produces animals that are so well behaved that operant trainers use it almost exclusively. It is so effective, that often times, aversives are unnecessary.

 

So how does it work?

 

A reinforcement

is a consequence for a behavior. This could either be giving a treat to encourage a behavior, or withholding a reward if the behavior is not what you have asked for. It can also mean that the small twitch of a smile and a glance in the direction of a dog screaming at you from their crate, is a potential reinforcer. When using reinforcers, it is important that you recognize reinforcement can come from an event within, an external event through the environment, or the trainer. To effectively eliminate a behavior, simply remove the reinforcer. The behavior will lessen and eventually reach extinction over time. If a behavior is no longer reinforced, there is no reason for it to continue; especially if a new behavior is being reinforced instead.

 

  • Positive reinforcement: the adding of an appetitive stimulus to increase a certain behavior or response.
    Example: Father gives candy to his daughter when she picks up her toys. If the frequency of picking up the toys increases or stays the same, the candy is a positive reinforcer.
  • Negative reinforcement: the taking away of an aversive stimulus to increase certain behavior or response.
    Example: Putting ointment on a bug bite to soothe an itch. If using ointment on bug bites increases, the removal of an itch is a negative reinforcer.

Source: http://en.wikipedia.org/wiki/Reinforcement#Positive_and_negative

 

Extinction

is the gradual weakening of a previously reinforced response when it is no longer reinforced. This is often the preferred method for operant conditioning trainers because it is more humane and effective. This is often a frustrating method for new trainers, or those inexperienced in these methods, because of extinction bursts.

 

Extinction bursts

often occur soon after the reinforcement stops. It is the increase of frequency, or intensity, of the original response to the reinforcement. The extinction burst, is the dog’s response to a reinforcer discontinuing. The extinction burst is a very valuable part of training with this method, and can be used to your advantage. It does require waiting through an extinction burst for the behavior to finally diminish.

My favorite example of an extinction burst is this: Imagine a button on a table that would deliver a $20 bill every time you pushed it. You would more than likely continue pushing that button for that great reinforcement of earning money. Now if all of a sudden that button stops working, what do you do? That is where the extinction burst comes in. Humans do it too! Our first response would be to push the button harder, faster, hold it down, etc.: an increased version of the original behavior that offered reinforcement. A great real life example was the video game players with the Nintendo 64′s and the Super Nintendo’s. Kids who played those games quickly found reinforcement through blowing into the game’s connectors to cool them, remove any dust, and voila they worked! As the games got older and left out longer, we would offer bigger extinction bursts; blowing into the games longer, more delicately, pushing the game harder into the console, etc.  A behavior once reinforced but no longer receiving reinforcement, is a powerful tool in even an intelligent species; like ourselves. Why not employ such a powerful learning and behavior modification tool with our animal companions, for a more positive way to teach them?
 
So what about punishing the dog for misbehaving?

A punishment is an aversive that is applied by another animal (human or another dog). Punishment is often the cause for putting the dogs brain into a fight, flight, or avoidance response. Dogs have different ways of handling the delivery of a punishment depending on genetics, nurture, learned behaviors, and stimulus. Oftentimes, negative associations are made with the punishments; and the dog can then out of fear, frustration, and anxiety, create unexpected and unwanted behaviors. The methods of applying a punishment to a behavior can vary greatly between each individual and their tools. Punishment often results in a depleting desire to work, loss of motivation, and sometimes even fear. The consequences of punishment are often unpredictable and uncontrollable; leaving room for error in effective communication, and learning. Punishment can not only weaken a behavior but completely extinguish it; so an untimely correction, or one that completely extinguishes a behavior, could be catastrophic to a dog who has not yet completed the learning phase.

 

  • Positive punishment: the adding of an aversive stimulus to decrease a certain behavior or response.
    Example: Mother yells at a child when running into the street. If the child stops running into the street the yelling is positive punishment.
  • Negative punishment (omission training): the taking away of an appetitive stimulus to decrease a certain behavior.
    Example: A teenager comes home an hour after curfew and the parents take away the teen’s cell phone for two days. If the frequency of coming home after curfew decreases, the removal of the phone is negative punishment.

Source: http://en.wikipedia.org/wiki/Reinforcement#Positive_and_negative

 

 

“Aversives in general, and punishment in particular, may have bad consequences for the dog and trainer. They can produce uncontrollable fear, not only of the trainer, but the entire training situation. Aversives can suppress virtually all behavior. They may also encourage aggressive responses.” -The Dog Trainer’s Resource “The ABC’s of Behavior” by Marian Breland Bailey and Robert E. Bailey

 
So where would the punishments be applied ?

Imagine a dog is in a sit while a stranger approaches, and asks to pet said dog. The handler obliges, and the stranger approaches. In anticipation for the friendly greeting, the dog tries to stand to greet the stranger, which is met by a timely punishment by the handler (in this case a leash correction). The dog, running on a very cause and effect type brain, now anticipates a potential correction for standing; which in turn can potentially create fear and anxiety among other negative responses. Dogs internalize pain and anxiety quite well, but are very deliberate in their body language if you look closely. A subtle yawn, a dart of the eyes, a flick of the tongue, can signal they are anything but calm. They are now behaving out of classical conditioning, and a habit. The habit of remembering not to get up from that position or they will be met with a correction. This method of training is quite often successful, and when done tactfully can produce a reliably trained dog. Operant trainers recognize the value of a punishment in some scenarios; but more often than not, you can shape and mold a behavior much easier without punishment. Punishment leaves the door wide open, and invites new behaviors to surface; which can quickly become problematic in competitive dog training.

 

 

So what are some other myths about this training and why are they not true?

Often mocked for their seeming lack of punishment for bad behaviors, animals trained by operant conditioning are rumored to be unpredictable, unreliable, and poorly trained. If one were to observe a dog that has been trained effectively with operant conditioning, one would come to find they are extremely reliable. Operant conditioning primarily uses positive reinforcement in the learning phase, and adds aversives at a more appropriate time. These dogs have been trained with a very effective set of communication tools, which involve reinforcement instead of punishment. When positive reinforcement is used effectively in a training program that utilizes good mechanical skills, problem solving, and creativity, it will result in a well trained animal. Clear communication is the key with animals when you are trying to teach a completely foreign behavior. The people spreading the misinformation about positive only trainers, immediately assume that these methods fail to have a consequence for unwanted behaviors. The consequence for undesirable behaviors in “positive only” would be withholding the desired reinforcer. So you see, positive only really isn’t all that positive, despite some people mistakenly arguing such. Operant trainers are not permissive, in fact their methods require set standards and limits, which are then reinforced when the dog reaches the desired behaviors. It is also rumored that this type of training is only effective in a sterile environment. I question the legitimacy of any person who does not use distractions in their training program. That is not someone who is employing the methods correctly; and as such, should not be mistakenly thought of as correct positive only training.

 

Another myth often surrounding “positive only” training, is that the dogs cannot sit still, and never shut off their brains, because they are constantly trying to seek things from the handler; which results in raised adrenaline. It is true, that a positive only dog is taught to try new things, and as a result grows confident because they have control over their consequences. However, operant conditioning is far more powerful than what these myths try to play it off as. Marker training and operant conditioning; creates an enthusiastic dog, because the dog expects pleasurable consequences for behaviors. It results in a purposeful behavior on the dog’s part instead of habitual behaviors created through classical conditioning. When the training is done correctly, and finished before being applied to the “real world”; the dog is engaged with the handler. The dog is motivated to work, and finds constant reinforcement through the handler, and the open and positive communication. A dog who cannot sit still, and never shuts off, is more often than not, a result of poor training on the part of the handler; whom has failed to teach the dog how to cap drives. That is a separate thing entirely from the positive only training.

 

 

Through all of the myths and rumors that each training group seems to conjure from either side, it is widely agreed upon that not one method works for every dog.

 

Balanced dog training is not just using punishment to prevent, and rewarding to encourage, and to simplify it to such is insulting. Balanced dog training is employing a vast majority of effective and scientifically proven training methods. This includes modifying methods depending on the individual dog’s needs. It is also important to make note that dog training, and the teaching of behaviors, is vastly different than the communication methods used between pack members. Dogs use punishment as a tool for communication with each other, but it is important to note that they are the same species, and use it in context with a social structure and timely applications. Dogs speak in a very black and white language, while humans create a grey area of confusion. But the use of punishment really has no use, or even benefit when trying to teach an animal a new and foreign behavior.

 

 

I will not deny the effectiveness of punishment in dog training.

If I denied that it worked, I would be flat out wrong, and discredit everything I have written. It is effective, and people use the training methods every day. I still practice it quite often in pet dog applications because people want quick solutions. However, with my personal dogs I want precision. I have the time and skill to work my dog to an engaged, motivated, and happy obedience level. Bond is important to me, and watching my dogs learn how to problem solve , and challenging how intelligent they really are fascinates me. I recently watched a Michael Ellis video explaining what he is interested in learning about dogs now. He said when a dog makes a mistake in a training exercise, he prefers to evaluate the situation and why the dog did not do the act. Is it not yet proofed in enough environments under enough distractions? Is the dog not yet clear on the exercise being performed? Many people are quick to assume the dog understands the exercise at a certain point, and use corrections to then mold the dog to the desired result.

 

I guess all of this is to say, think from a different perspective. If you were in your dog’s position, what would you be more compliant and happy with? Operant conditioning methods, with primarily positive interactions; or punishment methods, that you don’t understand, which can cause anxiety and fear? I want to have a positive companionship with my dogs, so I use a combination of primarily positive methods to effectively communicate and motivate my dogs to work, in obedience, and real life scenarios. Using operant conditioning and clicker training does require more time, precision, problem solving on both ends to correctly motivate, desensitize, and proof a behavior, but the end results are worth it!