Training Behaviors with Love The Dog treats!
Updated: an hour ago
Training Behaviors: "Operant Conditioning"
In our last post on classical conditioning, we talked about "training" dogs' emotions. Now let's talk about the other process by which dogs learn: operant conditioning. In operant conditioning, dogs do not learn by association, but rather learn by consequences of their behavior. This is one of the key distinguishers between the two types of conditioning: receipt of a reward DOES depend on the dog's behavior in operant conditioning, whereas in classical conditioning, rewards are not conditional on anything the dog does (or does not do).
Operant conditioning is the means by which dogs learn tricks and cues (cues are also known as commands). If a dog learns that a certain behavior causes positive consequences (such as praise, affection, and, most importantly for most dogs, TREATS), this behavior is reinforced in the brain, and the dog becomes more likely to do it again. And remember - classical conditioning is always going on, and you can tell it's really working for Kumara (right), whose enthusiasm for "down" is visible in this photo.
It is easy for us to take advantage of this learning process by simply rewarding the behaviors we like, but with modern training methods, we can enhance it to make teaching dogs EXACTLY what we want FAST!
Here's the process simplified:
(1) Get the dog to perform the desired behavior
Actually, it doesn't matter how you do this, but some ways are more effective than others. Luring dogs to perform a behavior with a treat is preferred to physical manipulation or force (in the photo here, Jinbei is lured into "sit pretty" with a treat). This is in part because it is faster: a large component of many tricks for dogs is similar to learning a dance move for humans - muscle memory and many practice repetitions are needed to respond quickly and smoothly. Shaping is another technique that is used especially for tricky behaviors that are difficult to lure, and involves rewarding the dog for step-by-step approximations until they have understood the full behavior.
The method by which we get a dog to perform a behavior is also critical in other ways. If you want
your dog to perform a behavior quickly and with enthusiasm, such as many agility behaviors, then it is best if you warm up with a rousing game of tug or fetch and thereby condition the appropriate physiological state when you work on it (and vice-versa for calmer behaviors). For example, I am more likely to get a dog to jump quickly and with enthusiasm if I teach them through luring by throwing a toy over the jump and having the dog quickly pursue it. If instead I stand beside the jump and lure my dog over it by hand, I may later need to add in the running approach and gear the dog up to perform the behavior quickly and energetically. In the photo, Gretchen (with no prior jump training) is lured by a toy on a line in an obstacle lure course.
This is one reason why it is critical dogs are not taught behaviors by methods using fear or pain. It is very difficult to communicate criteria like "speed" and "enthusiasm" to a dog through punishment. If you administer a punishment to a dog as they take a jump because they are going slow, it is unlikely the dog will understand you are punishing them for lack of speed and not the myriad other things they are doing at that time -- from glancing at a stranger nearby who is eating a snack, all the way up to and including the jumping behavior itself! Once again: classical conditioning is ALWAYS ongoing, even when humans intend to be teaching behaviors and not emotions. Your dog is constantly updating his or her feelings about people, places, things, and scenarios. In this photo, you can see both processes at work: Stella, the rescue bulldog mix, has been trained by a Karen Pryor Certified Trainer (1) operantly to participate in nail trimming, and (2) classically to do so cheerfully and of her own free will. Modern dog training allows us to train all behaviors without the "fall out" of punishment, so don't risk making your dog hesitant to learn new things, or dislike the training process and potentially, all the people and places associated with it.
(2) Let the dog know when they've done the behavior ("mark" it) and reward!
Although many easy behaviors like a sit can be taught simply by luring our dogs into a sitting position and rewarding them, as behaviors become increasingly complicated, it can get confusing for dogs to know what exactly they are doing that is the right thing. This is true even of "down". To teach down, we lure a dog with a treat until they are laying down, and then reward. However, despite this training, many owners find themselves repeating the cue (or adding "all the way") to dogs that apparently "know" what down means, because the dog starts to lower himself or herself down, but never quite completes the behavior. This dog is not being disobedient: their understanding of the down behavior is the movement of lowering themselves, not the final position. In this case, a trainer would say, "the criteria the dog must meet to get the reward has not been precisely communicated." The dog does not understand that he or she needs to lay all the way down, and in fact most dogs prefer not to since they can maintain proximity to the treat and thereby get it faster if the human would just reward them for doing what, in their minds, worked all the times before!
It is thus critical even for these beginning behaviors that we "mark" with a clicker or a word like "yes" the instant the dog meets our criteria; that is, the instant he or she does what we want. For down, it's the instant their elbows connect with the floor. This also becomes important when you teach a dog distance behaviors: if your dog needs you to be right there delivering the cookie to know it did something right, you will not be able to effectively teach him or her to behave even when standing far away (think about dog sports like obedience and agility). It is also critical to communicate precise understanding of criteria to dogs who will perform advanced behaviors like IPO or FCI heeling. How do you tell a dog that its shoulders may never go past your pants seam, that it must always be physically touching your leg, and that it must maintain eye contact at all times when heeling? Similarly, how do you teach a dog that it must always touch the yellow-painted bottom portion of the A-frame obstacle in agility (known as the "contact"), instead of leaping off to maintain speed? The answer is by marking the behavior so these criteria are communicated with clarity (and a few other tricks when needed)! And remember to always follow up your marker with a treat - or it won't be a marker at all, and you'll have to teach the dog a new marker later to take advantage of this method.
Teaching a dog exactly what they did right by "marking" is superior to alternative methods, such as punishing the dog for wrong criteria. After all, it is a longer, more painful, and more confusing process for a dog to learn the strict criteria for heeling if they are taught it by learning the single correct way of doing things through a painful trial and error process. Since there's usually only one way to do things exactly right, let's speed learning up and keep it fun by telling dogs how to do it right in the first place with a marker, and then reward them so they enjoy training/performing, and especially, us!
(3) Practice this until your dog fully understands the behavior
Ensure your dog understands the criteria of the behavior and, if necessary, is performing with the desired speed (or gentleness) and enthusiasm (or calmness) necessary.
(4) Add a cue
Only once a dog completely understands a behavior (i.e., you expect he or she could communicate the exact criteria for a reward correctly to us if he or she could talk) do we "put it on cue," by which we mean teach the dog the command. If we don't wait until the dog understands the behavior to start naming it, this can be confusing for the dog. After all, it make take many, many trials with luring to get some dogs to down, for example, or to roll over. We do not want them to think "down" means "now is one of those times that I will take a cookie and move it around and if you sort of guess at what I want, sometimes you get it, but you're not quite sure why" in human-speak. We want the association of what behavior "down" means to be crystal clear to the dog from the beginning: "down" means "lower yourself down all the way until your elbows are on the floor".
To add a cue, always say the cue first without moving your body or signaling in any other way, and
then immediately follow up with the lure/shaping you have been using to teach the behavior. After many repetitions, your dog will have made the association between the word and behavior, and will begin to "offer the behavior" even before you finish saying the cue! This is the sign that you may now say the cue, pause and give the dog a chance to perform it without your help, and give extra special praise and maybe 3, 4, or 5 individual little bites of treats in a row to make a big impression that they just did a great thing! (This is called a "jackpot" and may speed up "leaps" in dogs' understanding in training. In this case, it provides strong reinforcement for a behavior closer to the exact criteria we are aiming for, and teaches the dog that doing behaviors on the verbal cue alone really pays off!) We are not sure where Gretchen came up with the criteria for the pictured behavior, nor what to call it.
(5) Practice the behavior in increasingly distracting environments, at distance, and for longer duration
Many owners are dismayed when their dog - who sits perfectly at home - doesn't sit when in the park. Sadly, this sometimes leads to those dogs getting punished and results in frustration for dogs and owners alike when in fact these dogs are not being willfully disobedient. After all, why would a dog who cheerfully complies in the home suddenly be unwilling outside? For dogs, what trainers refer to as "the three D's" (distraction, distance, and duration), need to be proofed until a dog can be reasonably expected to perform a behavior under distraction, at a distance, or for a long duration. This is the same for human children; it may seem perfectly easy to us to focus when out on a hike, but that degree of focus is learned gradually. Any typical, fairly well-trained dog will hold a stay for consecutive seconds in their home, but most dogs will not on a busy street. Punishment will not help; we must help the dog gradually get accustomed to maintaining their focus on us and the behavior at successive levels of distraction, distance, and duration. Thankfully, this learning often "generalizes" to other behaviors and scenarios the more times you proof a new behavior against the three D's, so a dog whose training has been invested in in this manner will be able to "take behaviors on the road" quickly - or even immediately!
Here we called in seven dogs to sit for a group photo. It's pretty clear the cue was delivered: all six residents of the home are sitting quite nicely, but the visiting poodle is understandably distracted and, despite being an accomplished sitter in many other scenarios, still standing.
Although these are the exact steps certified professional dog trainers use, they frequently will take the below additional steps and considerations into account (they are in gray text):
(1) Plan in advance the exact criteria for the behavior, and especially each approximation if using shaping, and how to teach it. Research "state of the art" techniques - like teaching weave poles in days with agility world champion Susan Garrett's 2x2 Weaves method rather than months (as it used to be done with luring alone). Understand the dog's preferred rewards (food? praise? play?) and most tempting distractions (squirrels?). Be prepared to work with these to enhance your teaching and proof your final result. The photo to the left shows how a trainer has created peace in a multi-dog household by (1) proofing against distance, duration, and distraction, and (2) conditioning with relaxation on beds, a calm down simultaneously on all members.
(2) Get the dog to perform the desired behavior according to your plan, and video sessions for review and critique later by yourself and peers.
(3) Let the dog know when they've done the behavior ("mark" it) and reward. Jackpot especially stylish or high-quality versions of the behavior. Work quickly with smooth and efficient mechanics to get in dozens of repetitions a session.
(3) Practice this until your dog fully understands the behavior
(4) Add a cue
(5) Practice the behavior in increasingly distracting environments, at distance, and for longer duration. Do not set up situations too demanding for the dog so that they learn ignoring the cue is okay.
(5) Transfer the behavior to a random or differential reinforcement reward schedule and vary rewards. (For example: "Less than 50% of behaviors should be rewarded, selecting out for those that most closely approach the desired criteria".) Treats are not the only way to reward sit and other polite doggy manners - Jax reports that getting to go outside is very reinforcing!