For those of you who have seen me work-- you know that I don't use Obedience Training with dogs. Well that isn't exactly true of course. But it's often the case, and it's especially true when I'm working with a new dog or when I'm teaching a dog not to do something. But Why? If you can get a dog to perform a behavior, isn't that all that matters? If the dog performs the behavior you want, who cares whether or not you use a verbal cue/command, right? Well it does matter, and it matters a lot. Trainers especially are overly focused on performance-- the measure of a physical action or outcome. Rarely it seems does anyone bother to consider what the dog is actually learning during training or from similar interactions. What a dog learns from an interaction or training session is far more important than whether or not a dog performs the action requested. Let's see a few reasons why...
One of the very primary reasons that I don't speak to a dog when I first begin training is that, well, dogs don't speak English. In fact, they don't have a verbal language at all (Note: Vocal communication is not equivalent to having a Verbal Language). Why is this important? Well because dogs do have a language- they are excellent readers and interpreters of body language. And speaking words to a green dog only adds useless information to the situation. What's worse, if I 'talk' to a dog to request a specific behavior, I am expecting, consciously or subconsciously, that the dog should do what I'm saying. I have an expectation for compliance based on the word(s) I've used. The dog on the other hand, is watching what I'm doing (if I'm lucky) and behaving accordingly. So if for example I say the word "SIT" but the dog reads my body language to mean 'Hey let's get excited now that you're paying attention to me!' instead, then we have created contradiction. I expect a SIT; the dog gets over-excited and doesn't SIT. This is because, from the dog's perspective, that's exactly what I've asked him to do!
Of course being the verbal creatures that we are, we resort to one of several options of recourse: we repeat what we just said again, and again; we say what we just said, only louder; or we say the words again, now with more emotion. All the while, the dog is watching us. The dog may eventually SIT (or not) due to the social pressure we are applying, but not because he understands the words we are using. The dog is not refusing to comply, but rather is only following the language that he understands-- our body language, movements, gestures, & energy.
So when I'm working with a new dog, I make sure that I can get the dog to perform all of the essential behaviors that I require without saying a word. This avoids communication errors due to contradiction. Then, once the dog knows how to perform the various required behaviors, I will assign a sound (words are not words to dogs, they are sounds) to denote the performance of each particular behavior. Not only does this avoid the contradiction between what I'm saying verbally (commands) and what I'm doing physically (body language), but it also aligns with how dogs learn associations. So, general rule of thumb-- Always make sure that your dog first knows how to perform the target or desired behavior very well, and then assign a sound (word) to denote the performance of that behavior.
So, I abstain from using verbal OB cues/commands: A) to avoid confusing the dog with what I might otherwise be communicating with my body language; and B) because, in order for a dog to learn the meaning of a verbal command, it must already know how to perform the target/desired behavior. These are relatively minor concerns however. Of greater importance is the overall lesson that the dog may (inadvertently) learn from interactions involving verbal OB cues/commands. Consider...
Obedience Training (and we'll also include reward-only based Compliance Training here) is a form of disclosed communication. That is, the receiver of the information (the dog) is aware of the source of the information (the owner or trainer). The 'information' of course is the verbal cue/command along with any other social cues and body language exhibited by the owner/trainer. This is so very important. If the dog understands that we are the ones issuing the requests 'to do' (and especially 'not to do') certain behaviors, then we may likely create conflict. There are numerous common situations where the use of OB cues/commands will not be effective at addressing behavioral issues, and in some cases the use of OB cues/commands will actually make the situation worse, due to the creation of conflict.
Let's look at a few common scenarios to see why verbal OB cues/commands should not be used to address behavioral issues. First, consider a dog that likes food... a lot. Most dogs, yes? So how do you manage or control your dog at mealtimes? Do you ask him to WAIT or SIT? Does the dog want to perform these behaviors for you? Not really. From the dog's perspective, it looks like this-- The dog wants the food in the bowl, and you are saying he can't have it (via obligatory compliance with your SIT command). This creates internal conflict within the dog. So what happens? Most of the time, the dog just gets very anxious (or what looks to many people like 'over-excitement') with whining, panting, and pacing. Sometimes you'll get protest as well, where the dog will bark back at you, or rush the bowl anyway. And in some situations the dog will comply with the request to wait for the food, but once he takes possession of the bowl.. don't get too close... now the dog vigorously and aggressively defends his food. Anxiety, protest, impatience, and even defensive aggression-- all can result from the (mis-)application of OB cues/commands. That's Strike 1 for using an OB approach to establish resource ownership.
So what about something the dog isn't ever allowed to have, such as waste from the garbage can, or unattended morsels of food from the counter-top? When we catch our dogs attempting to pilfer or steal such goodies, what do we normally do? We command (or yell) LEAVE IT or NO! What happens next? Well if you catch it in time, the dog usually disengages and walks (or runs) away. But what did the dog actually learn from the encounter? Did the dog learn to leave the food alone... that such items are forbidden and not to be touched ever again? Not really. The dog learned a lesson alright-- Do NOT try to enjoy the scraps in the garbage can... whenever there is a human around to yell at you. Congratulations-- you just created what are known as Owner-Absent Behaviors (OABs). Just as the name suggests, these are behaviors that the dog has been conditioned (by a human) to perform only in the absence of humans. Not good. Strike 2 for the use of OB cues/commands here as well.
Finally, what about attention-seeking / soliciting behaviors? Let's look at non-aggressive jumping up during greetings. Many dogs do this, and many owners dislike it, especially when it occurs with guests. When the dog jumps up, owners immediately start talking... 'Rexie, NO!! OFF! DOWN!... DOWN!!!' The dog may in fact get down, but is that what the dog learned, to stay off of the guests? Not really. The dog learned that jumping up onto the owner or guest instantly makes it the center of attention, with everyone moving, turning their backs, touching, grabbing, and most importantly talking to the dog, as soon as the jumping behavior begins. OB cues/commands are a form of communication remember, and communication is attention, or at the very least a form of interaction. So the dog has just been reinforced for jumping up. The dog may comply and get back down, but the next time that same guest comes over... you guessed it, the dog will jump right back up again, only with more enthusiasm or determination this time because the dog knows the behavior works. Strike 3 for the use of OB cues/commands... you're out!
OB training is about performance, about measurable behavioral outcomes. The dog either performs a requested action / assumes a specific body position, or it does not. There is little ambiguity and no need for interpretation. But whether the dog performs a requested action or task may be quite irrelevant in reality when assessing the success of a particular training approach. The lesson learned by the dog is far more important than whether or not the dog performed the requested or required behavior. However, the nature of the lesson learned is not so easily 'measured' as is performance, and therefore requires some additional interpretation and more extensive observation. In all of the 3 previous examples, a dog would likely comply with the verbal cues/commands issued: The dog would sit to wait for its meal, the dog would leave alone scraps of food when told to do so, and the dog would get back down after jumping up on a guest when told to get down. Compliance with the request, as a performance measure, would appear to be excellent. But in fact the outcome would be far from ideal. That same OB approach would have also potentially created anxiety, protest, direct conflict over resources, owner-absent behavior, and reinforcement of an undesired behavior(s).
OB training has its place in dog training, but such an approach alone will not create a well-behaved dog. In addition, it may not be effective in controlling or changing behavior in many instances, and may even make problems worse in others. The few examples above illustrate why I do not rely on, and in some cases altogether avoid, OB cues/commands when addressing many common behavior problems.
© 2017 JRD
Post 131: The Problem with Obedience Training
Post 130: Punishments and Aversives in Training
There always seems to be an ongoing controversy and subsequent debate as to whether or not 'punishment' or aversive/unpleasant consequences can or should be used in training. The debate always gets extra hot when these discussions include dogs and dog training. In fact right at this moment, there is legislature pending in numerous jurisdictions that would ban or outlaw the use of certain types of training equipment, just because these devices might cause a dog discomfort or pain. Let's dive a little deeper to see why this debate continues, and discuss how and when aversives might be used effectively in training...
Every single organism on this planet has evolved the capacity to adapt to and learn from aversive environmental events. Without this ability to learn from unpleasant experiences, an animal would certainly not survive very long, as the real world provides many treacherous and dangerous situations that must be avoided. So what's the issue then with using aversives in training? Well there are a number of factors worth considering before or when you decide to use aversives to train your dog.
But first, let's get past a couple of common myths surrounding the use of aversives in training. The use of aversives will not ruin your dog, or your relationship with your dog, or cause irreparable psychological damage to your dog. Many many dogs have been trained using aversives without any side-effects. In addition, some dogs that have not responded to force-free, positive-only* or reward-based training approaches have subsequently done very well when aversives are introduced into the training program Without some form of aversive intervention, many dogs that would have otherwise been put down (euthanized) would not have survived and thrived.
* Positive-only is often used to describe training approaches where only rewards or reinforcers that the dog enjoys are utilized, while anything the dog doesn't like is excluded from the training. This is hypothetical silliness, as any dog that has ever been on a leash has, by definition, been 'forced' to go in directions other than where it wants to go. The question isn't whether aversives or force should or shouldn't be used; rather the questions that merit consideration regarding aversives should revolve around how, when, where, and to what extent those aversives are utilized.
That being said, aversive use should be approached with considerable caution. The fact that aversives may be warranted or even necessary does not equate to a blank check for utilizing them. Aversive use, especially in certain circumstances or situations, can lead to more problems than it solves.
Before considering aversive use in training, owners and trainers should be familiar with a few tidbits of reality. First, not all behaviors are equal. Some people will try to convince you that 'behavior is just behavior' implying that all behaviors are roughly equivalent and thus equally amenable to modification, shaping, or change. More involved or complex-appearing behaviors they would argue are just a conglomeration or composite of smaller, component behaviors. This is just not the case. The visible action or observable 'behavior' is nearly irrelevant when attempting to understand that behavior. What's more important, especially when considering a training approach or rehabilitation strategy for addressing a behavior, is the underlying drive or motivation that caused that behavior or outward expression of action to occur. We'll talk more about drives, especially emotional drives, in another post.
Second, performing a behavior is not equivalent to not performing (another) behavior. This is a common argument against the use of aversives. The reasoning here is that if you don't want a dog to perform some particular behavior, then reward some alternate behavior, instead of using aversives to address the unwanted behavior. The illogic here hardly needs to be pointed out, but for the sake of clarity, let's consider this argument. If a certain undesirable behavior exists (Behavior A), and you want to teach your dog instead to perform an alternative behavior (Behavior B), even if for no other reason than to keep your dog from performing Behavior A, then you should be able to reward/reinforce Behavior B sufficiently to condition your dog not to perform behavior A. Well this might work, but only if the drive/motivation (see discussion above) to perform Behavior A is really really low, and the drive for the reward and for performing Behavior B is relatively high. In reality, this set of conditions almost never exists, mostly due to the fact the behaviors we want to change (the A Behaviors our dogs have) are frequently high-intensity behaviors. That is the dog's internal motivation to exhibit or perform an A Behavior is high.
You see this frequently, where owners/trainers attempt to address intense, emotionally-driven behaviors such as leash reactivity by asking the dog to SIT and look at them instead of reacting to another dog. This may indeed work with some dogs, but the approach fails many more. Imagine training a human this way, with no consequences for unwanted behavior. If you wanted to train your child for instance to walk down the sidewalk without darting into the middle of the busy street, how would you do it? You could certainly reward your child every so often with a shiny penny and lots of praise. That child would likely walk right next to you... for a while. But when the child gets sated or bored with the routine, or worse yet sees a shiny nickel in the middle of the street... what now? How do address a competing motivator, perhaps one that holds more motivating power than what you are able to bring to bear on the situation? You cannot. There must be a consequence that effectively communicates do not do that specific behavior right now. In the case of teaching your child not to step out into the street, you would also need to communicate the conditional nature of 'do not go into the street' rule. That is, you'd need to teach the child that sometimes it might be ok to step into the street, but other times it's a strict no-no. After all, you might need to cross the street at some point, and would therefore need to step onto the street. In the human world, we'd call this conditional rule-setter a traffic light. The bottom line here is this-- Teaching your child to walk on the sidewalk is not the same as teaching him/her not to walk out into the street. This is true even though the two behaviors are mutually exclusive. And the same is true for training dogs. If you understand this basic notion, then you are ahead of half the trainers out there praticing today.
Post 129: Emotions & Aversives in Training: A Volatile Cocktail
Post 127: So what's wrong with Operant Conditioning (OC)?
The one thing that every dog owner has experienced with their dog is… Emotionality. Dogs are emotional creatures, possessing nearly the same emotional system as we humans do. Dogs get excited, nervous, anxious, afraid, angry, jealous / possessive, happy, confident, sad, lonely, panicked… nearly any emotional or behavioral state that we experience, dogs experience too. Emotionality is of course normal and healthy. However, problems can arise when emotional states become too intense or are expressed at inappropriate times. In fact, nearly every significant behavioral problem that you can think of in dogs is caused or driven by an underlying emotional state. Everything from Separation Distress to Fear-based Aggression; from Social/Competitive Aggression to Food & Bone Guarding; from Territorial Behavior to Pulling on the Leash and Leash Reactivity. All of these behaviors and more are driven by a strong emotional or instinctual state. This is a most critical fact, especially when it comes to addressing the observable behaviors that result from these emotional states. Attempting to correct, address, or re-direct the behaviors we observe but do not like often fails to resolve these issues, and in some cases creates additional behavioral side-effects. Of particular concern is the use of aversives to suppress emotionally driven behaviors. Let’s look at why this practice is so common, even amongst trainers who should know better, and the potential pitfalls of doing so…
If you've heard me speak before, you know I'm not a big fan of the so-called "Operant Conditioning" Paradigm or Learning Principles. But why not? Hasn't OC been around for 100+ years, and more importantly, doesn't everybody use it to explain / describe / justify learning procedures? Perhaps, but that doesn't mean it's right, it may just be a QWERTY phenomenon.
This is a HUGE topic, and an even BIGGER can of worms to open. We're only going to scratch the surface here, but we'll still need to divide this commentary in to 2 parts. If you are interested in learning more about the pitfalls and shortcomings of conceptualizing learning processes through the Operant Conditioning paradigm, be sure to check out the video "Meta-Modern Learning Theory"
Part I of this post will cover Operant Conditioning-- what it is, how it got here, and what it's supposed to represent. Part II will discuss the fundamental problems with the Operant Conditioning construct, how it is misinterpreted by many, and why it's a bad idea to view learning through 'Operant' eyes. So let's get started...
What is Operant Conditioning (OC)? OC is a theory of learning, specifically a set of principles that govern or describe how animals learn and adapt their behavior according to resultant environmental consequences. OC was first formalized by B.F. Skinner, and was an extension of Edward Thorndike's earlier Law of Effect. Thorndike proposed that
Have a comment, question, or idea for a Blog topic?
Contact me by email and let me know!
...And because I hate automatic pop-up windows so much, you won't find any here. However, if you want to be notified when new posts are added to the Blog, click the button below and add your email to the list! I'll send you a quick email to let you know when the next topic is posted! Thank you.