Join Us

Join the Animalinfo's online directories. Silver Listings are free. Find out more >

Get a gold plan Get a silver plan
Join >

Positive Reinforcement Training for Dogs

Search the summaries below for articles on many aspects of reward based positive reinforcement training for dogs. Click on 'Read more' for the full article or search for books in our online book resource.

Positive reinforcement training

How Clicker Trainers Correct Behavior - Part II

Category:
Positive Reinforcement Training for Dogs
Author(s):
Cindy Ludwig, Canine Connection LLC

Part II: Correction, the Clicker Training Way

This is part II of this article. For Part I: How Clicker Training Minimizes the Need for Correction - Click here

Introduction

Good clicker trainers who understand the principles of animal learning and clicker training do not often have to correct behaviors, but if they do, they do so by using a clicker to clarify. A clicker is used to tell to a dog, "Yes, that is correct!"  No click means, "Try again".

Oftentimes a dog has learned and repeats an unwanted behavior because someone or something in the environment is rewarding the behavior. For example, a dog may learn to bark or jump for attention because it works! Giving the dog the attention he wants, even if in the form of reprimands or just eye contact makes the behavior stronger and more likely to continue or even increase!

Sometimes a dog may appear "stubborn". In reality, it doesn't do what is expected because it doesn't really know what to do, or how to do it, isn't sufficiently motivated, is frightened or is too distracted. For example, chasing squirrels may be more enjoyable (rewarding) than coming when called. While punishing the dog for not coming when called might seem like a solution, the dog is unlikely to learn the intended lesson. He is more likely to develop a negative association with coming when called and tarry all the more. A better way to build a reliable recall would be to highly reinforce the desired behavior of coming when called with clicker training, and drop the corrections altogether.

Correcting behavior using clicker training involves clarifying with the clicker and being cognizant of what is reinforcing the unwanted behavior. It may also involve determining what is punishing or inhibiting the desired behavior. Clicker trainers correct behavior by teaching animals what to do instead of the unwanted behavior and by setting up the environment so that the desired behaviors are the ones most likely to be reinforced.

Following are some examples of what can go wrong in training and how clicker trainers would go about correcting the situation.

The Problem of "Poisoned Cues"

Clicker training teaches cues rather than commands. Cues are like commands in that they are the names given to specific behaviors such as sit, down or come when called. The difference between cues and commands lies in the intent. Cues represent opportunity whereas commands are orders with an associated threat for non-compliance. While the threat of punishment can be motivating, the motivation to avoid or escape punishment does not generate the same level of enthusiasm for participation and reliability in performance that force-free training generates, and punishment can actually have an inhibitory effect on learning.

Cues, as well as commands, can take on a negative connotation and become ineffective if an owner nags at the dog, says the cue in a harsh way, or pairs it with negative consequences or unpleasant circumstances. For example, if a dog is scolded or physically punished for not coming when called, the cue is likely to be perceived as something unpleasant the dog wishes to avoid. When this happens we say that the cue has been "poisoned", or rendered ineffective.

This situation can be remedied by teaching the dog a new cue that has a positive association or consequence for the dog. For example, to teach a dog to come when called, we can click as the dog is coming and provide tasty food treats such as real meat or cheese when it arrives. We can also avoid calling the dog to us to do anything the dog considers unpleasant. In effect, this would be punishing the dog for coming which in turn would produce an unreliable recall.

Reversing the Effect of Punishment

While most animals learn very quickly using clicker training, animals that have been trained with force and correction tend to be slower to respond and more inhibited. They may be passively obedient, but are reluctant to offer behaviors and may appear "stubborn." Their performance often breaks down in the performance ring.

Dogs that exhibit fearful or anxious behavior due to harsh treatment, or that have developed negative emotions associated with people, other animals, or things in their environment must first learn that they are safe before they can be trained. This is done through a careful process of "behavior modification" which helps the dog develop a less fearful and more positive emotional response to the feared object or situation.

Clicker training facilitates this process of behavior modification by teaching an animal clearly defined alternative responses to feared objects, or situations, without the threat of punishment for incorrect responses. Not only does an animal learn new emotional responses to the feared object, or situation, but he develops confidence as a result of learning new behaviors and successful, non-threatening encounters in the previously fear-inducing situation.

Is Saying No Necessary?

One has only to look at the frequent use of the word, "No!" or "eh-eh" by dog owners to know that these terms are not a useful way to change behavior.

These terms are just two of the more commonly used terms trainers refer to as "no reward markers." No reward markers or verbal corrections are used to indicate to a dog that he has made a wrong choice. At best they temporarily interrupt behavior. They are rarely effective punishers since by definition a "punisher" is something which causes behavior to decrease.

The word, "No!" really provides very little, if any information to a dog. Think about the "hot-cold" game that children play. Would silence provide any less information than the word, "cold?"

Saying "No!" or "eh-eh" can become a crutch and a bad habit of pet owners and trainers who can easily come to rely more and more on the use of these terms. Saying "No!" can provide a temporary feeling of control over a situation and a sense of relief of stress which can be very reinforcing to the frustrated owner or trainer.

For example, yelling at a dog that is barking may provide some sense of relief as an outlet for anger or frustration, but it does little if anything to solve the barking problem. Unless the cause of the barking problem is identified, barking is likely to continue, and so is the yelling which has no effect.

Further, a no reward marker can poison a cue if the cue is followed by a verbal reprimand or correction for failure to perform the desired behavior. A common example is pairing the dog's name with the word, "No!" or even "Bad dog!" The result is that the dog is less likely to respond to his name.

Premature Addition of a Cue

In clicker training, unlike other types of training, the cue is added after the animal has learned the behavior. Not only is it easier for the dog to learn the behavior without the background noise of a word that up to that point lacks meaning, attaching a cue to a behavior that has not yet been learned may result in a behavior that does not look like what the trainer imagined as the goal behavior. Remembering dogs are not verbal creatures like we are.

Should the dog inadvertently learn something other than intended by premature addition of a cue, this problem can be solved by retraining the behavior and giving it a different cue. An example is provided in the video which accompanies this article.

Alternative Incompatible Behaviors

So what if a dog has learned an unintended or unwanted behavior? We can fix this by 1) not rewarding the unwanted behavior, for example, with attention, 1) clarifying with the clicker, and by 3) teaching an alternative behavior that is incompatible with the unwanted behavior.

For example, to stop a dog from jumping I recommend a twofold approach:

1) Avoid giving the dog any attention for jumping, even eye contact. This means no reprimands and no physical contact.

2) Reward sitting with attention, which is precisely what the dog wants when he jumps! The jumping behavior will cease because it is not rewarded, and sitting which is rewarded will take its place.

A clicker can be used to mark the desired behavior. At the moment the dog sits, click, then reward with attention and food.

Once the dog has learned to sit for attention, the food can be phased out. It is important to understand that consistency is key. Occasionally rewarding a behavior such as jumping with attention can actually make the unwanted behavior grow stronger. It is like the draw to gamble after a handful of coins have spilled out of a slot machine, or the lure of buying a lottery ticket because the potential for a very desirable reward is present, even if that potential is statistically very small.

Summary

The way clicker trainers correct behavior is through effective communication aided by a clicker. A clicker is used to clarify exactly which behaviors will be rewarded. Each click is followed by a food treat or other reward so that a dog knows exactly which behaviors to repeat to earn rewards.

Rewarded behaviors are reinforced, or made stronger and unwanted behaviors which are not rewarded are soon replaced by behaviors that work to get the dog what it wants.

When a dog learns what will work to get him what he wants and unwanted behaviors are not reinforced, the behaviors which work are likely to be repeated and the unwanted behaviors are less likely to be repeated. In other words, a dog (or any other animal) will repeat what has been rewarded and not repeat what is not rewarded, making physical correction or reprimands unnecessary. This is the way clicker trainers correct behaviors.

References:

1. Why we learn more from our successes than our failures

http://web.mit.edu/newsoffice/2009/successes-0729.html

2. Rewards and Dopamine; what happens when we clicker train?

http://barnmice.com/group/bodylanguage/forum/topics/rewards-and-dopamine-what

3. American Veterinary Society Position Statement on the Use of Punishment in Behavior Modification in Animals

http://avsabonline.org/uploads/position_statements/Combined_Punishment_Statements.pdf

4. The Poisoned Cue: Positive and Negative Discriminative Stimuli

http://www.clickertraining.com/node/164

5. Poisoned Cues: The Case of the Stubborn Do

http://www.clickertraining.com/node/2849

6. Survey of the use and outcome of confrontational and non-confrontational training methods in client-owned dogs showing undesired behavior

http://vet.osu.edu/assets/pdf/hospital/behavior/trainingArticle.pdf

7. "NRMs" No Reward Markers

http://www.clickertraining.com/node/179

8. Should You Use No Reward Markers? Examining the Debate

http://www.clickertraining.com/node/2848

9. What are the Implications of Using Training Techniques Which Induce Fear or Pain in Dogs?

http://www.dogwelfarecampaign.org/implications-of-punishment.php

  • How Clicker Trainers Correct Behaviors: Teaching Abby to Sit
    How Clicker Trainers Correct Behaviors: Teaching Abby to Sit