Chains and Sequencing
The following article was originally printed in Teaching Dogs magazine.
A sequence is a series of multiple, individually cued behaviors performed consecutively and usually without added reinforcement between them. An agility course is a long sequence made up of not only the obstacles, but also the directional cues given between the obstacles.
A chain is a specialized sequence in which a series of behaviors is performed in the same order each time and elicited by a single cue. For example, on the cue for the A-frame the dog touches a contact zone, goes over the top of the obstacle, and then touches the lower contact zone.
Four key guidelines govern successful sequencing or chaining:
- Train the individual component behaviors separately.
- Add new components only when the performance of included parts is exactly how you want it.
- If the dog makes an error in the middle of a sequence or chain, stop. Do not reinforce sub-par performances by continuing.
- Make the reinforcement offered for the performance commensurate to the degree of effort.
Individual Behaviors First
As trainer, your job is to make learning as easy as possible for your dog. Large tasks should, then, be broken into smaller chunks. First, examine your chain or sequence and list its component behaviors. If the sequence includes a chain, list the components of it as well.
In competition obedience, the retrieve exercise is comprised of a sit in heel position, the retrieve, giving the dumbbell to the handler, and going to heel position from front (called a “finish”). Three of the components are individual behaviors, but the retrieve is a four-part chain, a sequence within the sequence. That chain includes going out to the dumbbell, picking it up, returning to the handler, and sitting in front of the handler holding the dumbbell.
Train the components of the sequence or chain as you would any individual behaviors. Get these behaviors exactly as you want them, on cue, and strong. In some cases, practicality requires individual parts of chains to be taught together. For example, taking, carrying, and giving the dumbbell are so closely related that even though you may try to isolate one part, you cannot help but include the others at some level. This is fine. Isolate what you can to make learning as easy as possible, and make sure the behaviors you can’t separate are in their finished form before you attempt to add additional behaviors.
Get your behaviors perfect before you sequence or chain them. When you get to sequencing, the dog should be thinking about performing, not trying to learn what to do. Once the component behaviors are perfect, fade the clicker from the picture. Rule of thumb: If you need the clicker to let the dog know he’s doing a behavior correctly, it’s too soon to chain.
Let’s look at the process for building sequences and chains separately.
Building a Sequence: Step-By-Step
Sequencing is as simple as putting known behaviors together, but for a dog who has previously worked on continuous reinforcement, multiple behaviors is a new concept. Sequencing is exactly that—a concept. Once your dog understands the concept of performing multiple behaviors consecutively, putting together new sequences is easy.
To teach the concept of sequences, start with two behaviors, Behavior A and Behavior B. Begin by cueing several reps of Behavior B, each followed with a powerful reinforcer. Get the dog excited about performing that behavior! Then cue Behavior A and quickly follow it with the cue for Behavior B, which he was just excited about performing. Jackpot! Because Behavior B has a recent reinforcement history, the cue for that behavior acts as a conditioned reinforcer, predicting the payoff that follows its performance. In a sense, you “charge” the behavior, just as you initially charged the clicker.
Repeat the sequence in the opposite direction (B to A), starting with multiple reps of Behavior A before sequencing A and B together. Mix and match other behaviors, using the same technique. It won’t take many two-behavior sequences for your dog to catch onto the idea that two behaviors in a row are followed by a great reinforcer. Once he has grasped the concept of performing two consecutive behaviors, you can skip the step of “charging” the second behavior.
When your dog will happily and eagerly perform two behaviors, you can add a third. Use a technique similar to the one above, only this time, “charge” the final two behaviors. For example, to train Behaviors A, B, and C, first do several highly-reinforced reps of the sequence B and C. When the dog is enthusiastically performing B and C, cue Behavior A, and then follow it with B and C.
Again mix and match using the above technique until your dog eagerly performs a three-behavior sequence. When your three-behavior sequences are solid, add a fourth behavior. By this point, your dog will likely have grasped the concept of performing consecutive behaviors, so the new behaviors can be added more easily.
Being able to successfully complete longer sequences doesn’t mean you should do long sequences every time. If you did, your rate of reinforcement would likely fall low enough to affect your dog’s eagerness to perform. To prevent this, vary the length of your sequences, and, just as importantly, continue to work on and reinforce individual behaviors. Keep your dog guessing!
The process of building a chain is similar to building a sequence, except that the order of the behaviors doesn’t change. Again, begin with just two behaviors. You can start with the first two behaviors in the chain and chain in order (called front chaining or, simply, chaining), or you can start with the last two behaviors and work in reverse (called back chaining).
Though working in reverse seems odd, it’s actually quite effective, because the dog knows just how much he has to do to get to the reinforcer at the end. Each successive behavior is just a bit more reinforcing because each is closer in proximity to that payoff.
Additionally, the Premack principle plays a factor in back chaining. The Premack principle states that a preferred or more familiar behavior can reinforce a less-preferred or less familiar one. As you build your chain, one behavior at a time, progressing only when the current ones are perfect, the behaviors within the chain are becoming more and more familiar to the dog. A new behavior at the beginning of the chain may be difficult, but then he can relax a bit because what comes next is known and comfortable.
Consistency is your ally when you chain. Initially you will have to cue each part of the chain, but as the dog learns the routine, he will anticipate the coming cue and offer the behavior. Jackpot! When you have all of the behaviors in place, you add a new cue, a cue to elicit the entire chain rather than an individual piece. The process is straightforward—simply give the new cue before the cue for the first behavior in the chain. Repeat until the dog associates that cue with the chain and anticipates the behaviors that follow.
In a sequence or chain, every successive behavior reinforces the behavior that came before it. Thus, if the dog performs less than perfectly, stop the sequence. If you let the sequence continue, you make it more likely that the sub-par performances will be repeated in the future.
“Perfect” performance includes more than technical correctness. It includes factors such as speed, latency, and the aspects of the dog’s physical appearance that we interpret as “attitude.” If the dog’s attitude begins to lag, or if a previously fast performance becomes slow, you may have sequenced too many behaviors too quickly, and rate of reinforcement may have fallen too low. Again, doing frequent short sequences and individual behaviors will keep your dog’s reinforcement bank account full.
Would you expect to be paid the same for mowing, raking, and bagging an entire yard as you would for planting a single rosebush? Of course not. It takes a lot more effort and time to take care of an entire yard, so you expect to be paid more. So does your dog. The piece of kibble he gets for a sit probably won’t be enough to motivate him to perform a flawless heeling routine.
When choosing your reinforcement, consider aspects like the duration of the sequence and the difficulty of the behaviors, but also weigh your dog’s enjoyment of the behaviors. Your dog may be willing to do an extra long, extra difficult agility routine for just a tug on his leash at the end, but a shorter competition obedience routine may require a whole bowl of his favorite treat doled out a piece at a time. If his attitude is lagging, you might need to increase your rate of reinforcement by practicing shorter sequences, but you may also need to bump up the quality and quantity of the reinforcement for the longer ones. Make it worthwhile to play, and your dog won’t quit!