Clarity is critical for any learner. Whether we use an errorless learning approach, or we use a method of training that incorporates highlighting errors to our dog, it is still essential that we maximize the likelihood of success through careful set ups. It is also necessary to always be prepared to rapidly change strategies, as soon as we notice our dog is not becoming more confident, and more successful, throughout our training sessions.
Before we dive into the specific issues that can occur in our attempts to create successful and confident complex behavior chains, let's look at the broader area of managing errors.
Methods for Managing Errors
The underlying principle of most modern dog training protocols is to facilitate the performance of the desired behavior, so that this behavior can be reinforced. Then through this reinforcement history, the "correct" behavior becomes more likely.
Even with the best set-ups and training protocols, errors will occasionally still occur. How we respond to these errors can have a significant impact on our dog's emotions, and hence their desire to train with us again in the future. This is of particular relevance when working with less "drivey," or less resilient, dogs.
Some of the most common strategies for responding to errors include:
- Withholding reinforcement – (e.g. giving a treat for a "correctly performed behavior", withholding of the treat for an "incorrectly performed behavior").
- Resets – (e.g. using a treat toss, or a run around a cone, or a step sideways, backwards, or similar, to cause the dog to need to re-perform the behavior).
- No reward marker (NRM) – (i.e. at the end of the behavior using a marker word that lets the dog know the behavior they performed was not "correct", and hence they will not be receiving a reward for that behavior).
- Verbally identifying the error as it occurs (i.e. interrupting at the moment the error occurs, to notify the dog they have made a mistake, or in some cases using a specific interrupter that notifies the dog exactly what error they are making (e.g. they looked away from the "correct" focal point and are notified immediately, this may be combined with information to tell them what the "correct" behavior is.)
- Not highlighting errors – (i.e. not identifying errors to the dog, but instead relying on set ups / props / etc., along with careful building of behaviors, to create accurate responses. Then allowing habit to create reliable and consistent "correct" behaviors.) In this protocol the error does not form part of the learning process (i.e. the dog is not receiving any information about what constitutes an "incorrect behavior".)
In all the above methods, the goal is not to focus for too long on the error, but to rapidly facilitate a "correct behavior," so that it can be rewarded.
When to Change Your Error Management Strategy
Although the modern dog training approaches don't attempt to create errors, or actively "punish" them, we can inadvertently be punishing behaviors in a sensitive dog without even realizing it. This is because "punishment sensitive" dogs do not need physical discomfort or intimidation to feel punished. Simply withholding an anticipated reward, or notifying them that the behavior they just finished performing was "incorrect" can cause stress, confusion, and concern in these dogs. This can then potentially lead to a less resilient dog shutting down, and/or performing escape or avoidance behaviors. It should be noted here though, that all dogs are individuals, and will cope with errors and associated emotions in very different ways. High drive dogs, dogs with good tolerance levels for frustration, and dogs with greater self-confidence and resilience, do not need the same approach as a highly sensitive dog, or a dog with less confidence in their own ability to cope, or a dog that has slow recovery from even brief periods of psychological discomfort (such as experiencing feelings of confusion or concern).
To assess our individual dog's comfort level, we only need to look at their response to errors, and their overall engagement in our training sessions. If errors are being handled in a way our individual dog understands, then we will not see significant increases in emotion or arousal during the training, and we will not see stress signals or unwanted behaviors (i.e. either excess spillage of behaviours – barking, whining, fidgeting, leaping, zoomies, leaving the handler or the activity; or inhibition of behavior – shutting down, slowing, stalling, becoming unresponsive, vacant staring, wandering off sniffing, lying down, or creeping away from the handler or the activity). If we do see these signs of confusion, concern, stress, or frustration, then we need to look closely at our current method/s, and reassess if maybe this individual would be more successful with a different approach.
We also need to ensure that the method we are using for managing errors is not likely to cause confusion when we attempt to reduce reinforcement at a later stage. For example, if our dog has consistently learnt that not being given a reward for a behavior is information that they performed that behavior "incorrectly," then this can cause a sensitive dog to become worried / confused / concerned, as reinforcement is reduced.
Equally, if resets have been utilized in combination with withheld rewards as a method of informing the dog that they performed a behavior "incorrectly," then this can cause confusion when we simply want to repeat an exercise without rewarding in between the repetitions (e.g. in the scent discrimination exercise in UD, where we want to be able to immediately send the dog again for the next article without reward, and without them being concerned that this indicates they made a mistake on the first article).
Remember each dog is an individual, many dogs are very resilient, and will learn successfully and happily, regardless of the strategies used. However, some dogs will struggle to remain engaged, confident, and successful with strategies that may have been very effective with our previous dogs. For this reason, it is worth continually monitoring our training sessions with each of our dogs, to confirm that each dog is keenly engaged and continuing to improve. If instead, we notice that our dog is showing any avoidance behaviors, or showing increased spillage of excess behaviors, then we may need to review the way we are handling errors, as this is a primary reason why we see deterioration in our training.
Errors in Behavior Chains
If we use errors as part of our dog's learning process (i.e. in some way we differentiate to our dog a "correct" repetition, from an "incorrect" repetition), it is essential that our dog knows EXACTLY what the error was. With simple behaviors this may be clear, e.g. the sit either happened on cue or it didn't. However, once the exercise comprises of more than one behavior, this clarity is instantly lost.
Even a simple recall comprises of the stay, the recall at speed, and the sit in front position.
If we require all parts to be correct as a pre-requisite to rewarding the behavior, then when our dog does not receive the anticipated reward, they may become confused. This is because it is unlikely to be clear to our dog what they did wrong. The primary feature for our dog may just be turning up, but then if we withhold reward because they anticipated the cue to "come," or they didn't run fast enough, or because they sat crooked when they arrived, now they may be second guessing whether turning up is the right answer.
As the exercises get more complex, the potential for confusion only becomes worse. The UD scent discrimination exercise is a prime example of this, to complete all elements correctly our dog would need to:
- Leave on cue.
- Leave quickly.
- Search enthusiastically.
- Find the correct item.
- Only mouth the correct item.
- Pick up the correct item.
- Hold the item in the correct manner.
- Not fumble the item in the pick-up or on the return.
- Not chomp or mouth the item during the return.
- Travel directly to the handler.
- Return quickly.
- Sit straight and close.
- Focus on the handler and present the article.
- Hold the article until cued to release.
- Release calmly and cleanly on cue.
- Wait in position calmly and quietly, until the cue is given that the exercise is finished.
Although this exercise clearly has many elements, it is not uncommon to see handlers withhold rewards in the training phase, simply because just one of those pieces was "incorrect" (often the bringing back of the "incorrect" article). The problem is that the dog has absolutely no idea which piece was wrong. A typical robust dog may be emotionally unaffected but the lack of anticipated reward (they may not understand what they did wrong, but they will likely keep trying, and simply through trial and error alone, they will often become fluent at the task – despite the lack of clarity). A sensitive dog however, will become emotional due to the confusion, this may result in excess behaviors (e.g. mouthing and fumbling of articles, ineffective searches, rushing to grab any article rather than making an effort to select, whining, barking, running off, etc.), or reduced behaviors (e.g. avoidance, hesitation, moving slowly, "losing interest" in the activity, wandering off, sniffing but not selecting, second-guessing their selection, dropping articles, not wanting to return to the handler, etc.).
When an error occurs (e.g. our dog brings back the wrong article), it is often best with a sensitive dog to go down the route of rewarding the effort anyway; essentially rewarding all the pieces that were correct. This will ensure our dog is keen to continue working with us, and this gives us the opportunity to help them to learn how to perform the exercise "correctly." Of course, just repeating incorrect sequences is not a good plan (whether rewarding each attempt or not), this is because our dog cannot learn the "correct" behavior, if they are consistently repeating the "incorrect" behavior.
Hence regardless of whether we are using an errorless learning protocol, or a protocol that notifies our dog of errors, we must instantly take apart the exercise, and start working on the "incorrect" component. For example, if the way the article is being held is the issue, then separate out that piece and work on it; if the incorrect item is being selected, then go back to set ups where the scent is indicated only (not picked-up / retrieved), and then fix the issue at the indication level before adding the pick-up / retrieve back in. It all cases it is necessary to isolate the issue down to a single behavior, not even two behaviors (e.g. the selection and pick up of the article), in order to achieve clarity.
Fortunately, most behavior chains even once fully established, can easily be divided into their individual single-behavior components, and each component can be worked on separately at any time to resolve an issue. In exercises where taking apart the components is more challenging, greater care needs to be taken to build the chain carefully in the learning phase, to reduce the need to revisit individual components.
It is also worth remembering that errors in all behavior chains can occur due to distractions, pressure, and excess arousal, not just a lack of understanding of the exercise. If we are in a training session, and an error occurs a second time, we need to quickly determine if this is a lack of understanding (in which case we immediately simplify and break the exercise down into single pieces), or if the error occurred due to the environment or other pressures (in which case we should stop asking for the chain in that setting at that time, to eliminate the ongoing practice of the incorrect behavior). If we determine the issue is something other than lack of understanding of the exercise, then we would need to work on the underlying issue that is causing the error (e.g. undertake further distraction training, or undertake strategies to improve arousal control), rather than working on the exercise itself. Working directly on the exercise in these cases is unlikely to solve the issue (as there is not really an issue with the understanding of the exercise), but it can lead to creating negative feelings about the exercise, and this may worsen future performance of the behavior, rather than improve it.
Influencing Speed in Behavior Chains
Many behavior chains (e.g. weave poles; scent discrimination articles; simple retrieves; directed retrieves; formal recalls; complex rally stations; NW search, locate, and indicate, etc.) share common features, one is that a certain level of speed may be a desirable aspect of creating a "correct" behavior.
The speed at which our dog performs a task is influenced by many factors:
- Motivation.
- Natural tendencies (e.g. size, athleticism, breed / line).
- Skill level within the task (i.e. more fluent / familiar skills will usually be performed faster, than less familiar skills).
- Confidence (i.e. confidence in how to perform the skill, confidence in the environment, and self-confidence).
- Arousal level (i.e. our dog needs to be at an optimal arousal level to achieve peak speed and accuracy. Either too high or too low can result in lack of precision and reduced speed due to a range of underlying factors.)
- Comfort / fatigue / energy levels (i.e. the speed of an exercise will typically be reduced at times of physical or mental fatigue, and when our dog is experiencing any discomfort.)
- Habit (i.e. if the behavior is only practiced at times when our dog is working at their best, and the desired speed is possible, and the behavior is not practiced when our dog is tired or lacking motivation, or the environment or situation is likely to reduce speed, then habit alone will influence the speed at which a task is performed).
As you can see from the above list, speed is not really a "decision" on our dog's part, instead, we can think of speed more as a byproduct of the way a behavior was trained, our dog's understanding of the task, their confidence level in that space, and their comfort and arousal level at that time.
For this reason, it is best to remove speed as a criterion for reward, especially with sensitive and less resilient dogs. Rewarding only when a task is performed accurately AND fast, is likely to produce confusion, hesitation, and slowing in a sensitive dog. They are unlikely to be able to work out why some reps are rewarded, and some aren't, because they are not actively choosing the speed, but rather it is happening to them, and in all other ways they are performing the behavior "correctly". This is particularly true of a dog who is experiencing some physical slowing due to environmental awareness / uncertainty.
Applying pressure, in an effort to increase speed, in a sensitive dog (even when the handler's actions outwardly only appear encouraging / positive), may in fact have the opposite effect, as our dog may become flustered / concerned / confused, and this risks further physical slowing, hesitating, and freezing.
When we are keen to increase the speed of a task, it is best to focus on:
- Increasing our dog's motivation to perform the task
- Increasing their understanding of the task.
- Build their confidence to work in a fully focused way, with specific distractions present.
- Build their ability to fully focus, in a range of different environments.
Managing arousal can also be key. Arousal being too low may result in reduced speed, due to lack of focus on the task, and lack of a primed physiological response. However, arousal that is too high for the task, is also detrimental to speed as it may result in:
- Focus being pulled away from the task. (Increasing arousal narrows focus. It will improve the individual's ability to focus on the most relevant stimuli. If the dog perceives the "most relevant stimuli" to be the task then performance improves, however if the dog perceives there are more relevant stimuli, perhaps due to general environmental awareness, or concern over specific elements in the environment, then speed and performance will decrease with increasing arousal.)
- Reduced ability to perform precise movements. (Excess arousal interferes with precision and coordination, hence excess arousal can cause physical slowing in order to aid with coordination and concentration).