Dog Discoveries

Dog Word of the Day: Accidental Behavior Chains

 

Accidental behavior chains in dog training are a phenomenon that dog owners and dog professionals should be aware of, considering that they can turn problematic. The fact is, there are behavior chains and behavior chains. Behavior chains composed of desirable behaviors may be trainer-mediated (the dog is purposely trained to perform them) or they may occur as an unexpected “side effect.” More problematic though are behavior chains where the undesirable behavior presents within the chain. How to undo an undesirable, accidental behavior chain once established? There are fortunately several strategies.

Picking up a toy, carrying it to a basket and dropping it in it, is an example of a behavior chain of desirable behaviors.

Behavior Chains in Dog Training 

Dogs are masters of associations, they have a tendency to put two and two together. This quality after all has an adaptive function if we think about it. In a past evolutionary history, where a dog’s ancestors were predators and potential prey, it was vitally important to form associations.

For example, if a dog encountered a predator emerging from the water (like an alligator) he would come to associate that area of water with the predator, and therefore, chances are, he would avoid such area with water, or  at least, be very wary around it. On the other hand, if he encountered prey animals under a bush that made for an easy meal, chances are, he would seek bushy areas more and more in hopes of finding more prey.

Dogs with separation anxiety know chained behaviors too well: they dread all their owner’s pre-departure cues such as putting on shoes, wearing a jacket and grabbing the car keys which culminate in the owner leaving the home. On a more lighthearted note, dogs who are eager to go on walks,  rather readily learn to get increasingly excited when their owners put on their shoes, then go on to wear a jacket and go grab the leash to snap on their collars.

On top of being very observant of chains of behavior occurring in their owners, dogs are adept to learning to perform operant behavior chains. For example, a dog can be purposely trained to perform a series of discrete behaviors such as going towards a toy, picking it up, walking towards a basket and placing the toy in the basket or they can be trained to go through a series of obstacles as it happens in the sport of canine agility.

Of course, just as behavior chain may be composed by a series of desirable behaviors, they may also be composed by multiple
undesirable behaviors, or even a combo of undesirable and desirable behaviors.

Accidental Behavior Chains in Dog Training 

Blessed with such an uncanny ability to chain one event with another, dogs are also capable of learning chains exclusively made of undesirable behaviors. For instance, dogs may learn to bark for attention when their owner is sitting on the couch, but then when the barking no longer works in garnering attention (because the owner decides to ignore), the dog may try to bark and then nip the owner’s hand.

Nipping is a behavior that is difficult to ignore. It may come unexpectedly and it may hurt or shock the owner emotionally, so in many cases the owner reacts. The owner upon feeling the nip may therefore make eye contact with the dog perhaps also remarking something along the lines of: “Ouch!! Are you crazy?” To attention-seeking dogs, any form of attention is better than no attention at all, so being looked at and talked to (even if in a derogatory tone) still qualifies as attention. Inadvertently, the owner has therefore reinforced the new chain which has formed courtesy of an extinction burst.A new bark and nip behavior chain establishes.

The above was an example of a behavior chain composed of inappropriate behaviors, but as mentioned, chains may also contain a combo of undesirable and desirable behaviors.  This often occurs also inadvertently: the dog owner just non purposely happens to reinforce the undesirable behavior and the dog thinks it’s part of the exercise. Let’s now take a closer look into some common accidental behavior chains that may occur in dog training scenarios. Here are a few examples:

  • A dog who jumps is taught to sit and is rewarded for sitting. The dog then learns to jump on people and then sit in expectation of a treat.
  • A dog who pulls on the leash is rewarded for heeling next to the owner. The dog then learns to pull on the leash and then heel for a treat, in a yo-yo-like fashion.
  • A dog who steals a lot of items, is rewarded for dropping items. The dog next learns to pick up items in hopes of being asked to drop in expectation of a treat.
  • A dog who chases squirrels, is rewarded for listening to a “leave it cue.” The owner therefore rewards the dog for listening, but unknowingly the chasing has been rewarded as well.

What do these instances have in common? They share the fact that such dogs believe that the reinforcer (reward) is contingent upon the performance of all the behaviors (including the undesirable one that dog owners are ironically trying to extinguish) rather than just the last one. In other words, these dogs believe that the undesirable behavior is part of the chain.

How to Undo Accidental Behavior Chains 

So your dog believes that the undesirable behavior is part of the chain and is therefore worthy of reinforcement, but if you inadvertently provide reinforcement, this will maintain and strengthen the chain, making it more and more likely to happen, quickly turning into a bad habit, so what should you do?

Several remedies for accidental behavior chains have been suggested according to some studies but several of these suggestions may be often impractical, depending on the type of behavior occurring. Let’s take a look at them regardless.

Satiation: this simply means providing access to whatever the dog finds reinforcing (attention, praise, food, toys) so that he no longer feels the need to engage in the undesirable behavior to gain it. According to one study, this method might not be effective considering that there may be other subtle reinforcements at play.

For example, in the case of the dog chasing squirrels, the adrenaline rush associated with chasing may be self-reinforcing alone and the squirrels’ unpredictable movements may never cause a dog to grow tired of chasing, therefore, even if the dog is fed a meal prior and his tummy is full, he may still feel compelled to chase regardless of the tasty treats you may offer.

Extinction: this simply means removing reinforcement so that the undesirable behavior weakens and stops occurring. Now, in behavior chains, something to consider is the fact that cues that predict a behavior that generates some form of reward (attention, food, toys), become reinforcing on their own.

So in the case of the dog chasing squirrels, just saying ‘”leave it” can be perceived as reinforcing since it has a history of generating a reward and therefore reinforces the preceding behavior of chasing. Omitting the verbal cue “leave it” may therefore help weaken the desire to chase, but one must be careful here, considering that, as mentioned, when it comes to chasing there are other reinforcers at play that maintain the behavior (adrenaline rush, instinctive attraction to unpredictable movements).

However, in the case of the dog jumping first in expectation for the owner to cue the dog to sit and earn a reward, this method may possibly work. Many dog trainers indeed suggest totally ignoring the jumping considering that cuing the dog to sit is already a form of attention since we are looking at the dog and talking to him.

“Cues that precede a behavior that is followed by a positive reinforcer become conditioned reinforcers. It has been shown that these cues can be used in a chain to reinforce the preceding behavior. “~McKnight, Debra Gayle.

Unchaining: This method consists of technically “un-chaining” the two behaviors. So in the case of dogs who jumps and then sits, you  proceed to reinforce only the sits with no jumping taking place prior. With time, you should see an increase in your dog offering sits in hopes of a reward, rather than jumping and then sitting.

Reinforcing other desirable behaviors: this strategy involves reinforcing some acceptable behaviors that may take place before the undesirable behavior. Dog owners are used to giving dogs attention when they misbehave, failing to acknowledge dogs for carrying out desirable behaviors. In the case of the dog who jumps and then sits, therefore one would reward with a treat the dog for just approaching and then ask for the sit and reinforce that. With this approach, the jumping doesn’t have a chance to partake in the chain and the dog is more likely to approach and sit.

Adding more behaviors: in this case, after the undesirable behavior occurs, the dog is asked several more behaviors (at least three) in hopes that the dog sort of “disassociates” the undesirable behavior from the reward. So in the case of a dog who jumps and expects the owner to cue him to sit after the jump, the dog can be asked to off, sit and then stay before he is rewarded. However, according to the study, this might not work as hoped considering how cues potentially become conditioned reinforcers on their own in behavior chains as stated in the quote above. On top of this, dogs may get frustrated in performing many behaviors prior to getting reinforced.

Adding a delay: in this case a delay is added in between the undesirable behavior and delivery of reinforcement in hopes of extending the chain and confusing the dog a bit considering that reinforcement, in order to be effective, is suggested to be delivered within 0.5 seconds of the desired behavior (Karen Overall, Manual of Clinical Behavioral Medicine for Dogs and Cats). 

In the case of the dog who pulls and then is reinforced for heeling, it may therefore help to not immediately reward the dog when heeling, but asking the dog for a longer heel (like several steps in a time lapse of 5 seconds). This should be taught gradually, starting by rewarding one step, then two, then three, then four steps, staring in areas with little distractions.

Prevention: This is the most recommended strategy for accidental behavior chains and perhaps the most effective considering that it prevents setting the dog up for failure. This strategy aims to go a step ahead of the first chain, preventing the inappropriate behavior from occurring in the first place. In technical terms, it’s known as antecedent control.

If you therefore pave the path for no jumping, it is easier to help your dog succeed. How do you prevent jumping? You use antecedent control techniques to prevent the undesirable behavior from occurring in the first place. For example, try not to engage in overly enthusiastic greetings, keep your hands low, no talking.

In the case of the dog chasing squirrels, you would have to pay extra attention to your dog’s environment and move away from areas with squirrels or catch your dog when he sees the squirrel, intervening before he gets to lunge in the squirrel’s direction (taking distance, turning the opposite way).

References:

  • McKnight, Debra Gayle. Breaking Accidental Behavior Chains.thesisMay 2010; Denton, Texas.  University of North Texas Libraries, UNT Digital Library

Share Your Comments

comments



Please Follow Us on: