Last time I shared examples of an extremely aversive experience poisoning what had once been a pleasant, rewarding behavior. So when something outside of your control goes wrong, how likely is it to destroy the behavior you’ve trained?
Ringtones and Repairing Cues
First, realize that the songs referenced in earlier posts were fairly neutral stimuli; I didn’t have a strong prior attachment to them, so it was relatively simple to forge a new emotional association. But what about stimuli that already have long-standing associations? What about a positively-reinforced cue that the subject already knows? If your dog has years of experience and knows that Sit means she can get a click and a cookie for plopping her tail on the floor, will one loud, scary noise at the wrong moment undo all that training?
Probably not. The more positive or negative experiences the subject has with a stimulus, the more solid that association becomes. If you have a thousand good associations and only one bad one, odds are good that the stimulus will still be good (though don’t be surprised if there’s a moment of hesitation, and be ready to reinforce the correct response when it happens).
However, sometimes something really good or bad can still make an impact. An extremely rewarding experience can overshadow an unpleasant stimulus, just as an extremely traumatic experience can instantly undermine our association with a favorite or positive stimulus. An example of the latter (once again, back to music):
For most of 2006, Laura had an upbeat song she liked set as the ringtone on her mobile phone. One rainy Monday in September, my little blue Doberman Lucrezia went missing from my yard. I contacted Laura and we set out separately around the neighborhood to search for her. When I found Lucrezia dead, I called Laura and told her. She came to help me recover the body. It was a horrible experience.
Some time later, I noticed that Laura’s ringtone had changed, but at the time I didn’t think anything of it; people often change their ringtones. It was five years later that Laura confessed she’d had to change the ringtone and take the song off her MP3 playlist because after the phone call about Lucrezia, it upset her to hear it. Even though she still liked the song itself, the terrible experience had poisoned her enjoyment of listening to it.
No matter how well-established the stimulus is, if the consequence is jarring enough – either for bad or for good – it can change the emotional association the subject has with it. This is one reason we trainers tend to be control freaks where training environments are concerned; we try to predict what might happen and make sure our subjects (and ourselves) are prepared for contingencies.
Still, we can’t always predict or prevent weird things from happening. In addition to his competitive career, Shakespeare (appropriately named) has been in a couple of stage plays. To prepare, Laura carefully coached the cast and crew on what to do around the dog, and trained with all sorts of props, treats, placement cues, recalls, etc. to make sure the dog always had clear signals and knew what to do. Then one night during rehearsal, the entire set on the back of the stage collapsed. Shakespeare bolted from the stage, understandably frightened by the building falling down around him and the panicked, shrieking humans running by. After that, in spite of copious reinforcement, he was a little suspicious of going on stage for that scene.
Being prepared is still a good idea – but the real question is not if, but when, something goes wrong in training, how do we fix it? We have a few options:
Make something good happen, instantly. So you’re working on sit-stays, and suddenly the roof collapses. Instead of grabbing your dog and telling her to Sit again, you just say, “Oops!” and dump an entire bucket of treats in front of the dog. (If you tend to work in unpredictable environments, you can even pre-condition this cue, so it already has a positive and calming emotional association when you need it.) It won’t completely negate the bad consequence, but it can help bring the dog out of her frightened state and lessen the impact of what scared her.
Give it time. Believe it or not, you don’t have to “get right back up on the horse that threw you.” In most cases, there’s no harm to be done by waiting a little while before dealing with the issue. If it was one freak incident that really frightened or hurt your animal, just leave that stimulus alone for a bit and give the fear or other emotions time to fade. In general it takes about 72 hours for the panic chemistry (adrenalin, et al.) to break down in the bloodstream, so anything you do during that time frame could potentially be even more traumatic.
In the example of Laura’s ringtone, after a few years had passed (with no additional negative association to that stimulus), she could listen to the song again. But had she tried to force herself to “get over it” quickly and listen to the song right away, the associated stress and emotion probably would have made the reaction much worse.
Go somewhere else. Environment plays a big role in associative learning. We often reference this in terms of generalization: If you always work with your pet in the same environment (one room in your house, or the back yard, or the classroom at your training facility) and nowhere else, your pet will only know the behaviors reliably in that area. But those environments also become classically conditioned, based on what the dog experiences there: Think of the adolescent dog who becomes excitable (or downright out-of-control) when you walk into the doggy daycare facility, or the animal who instantly becomes fearful or reactive when you walk into a vet’s exam room.
In the case of a dog that has a terrible experience with an object in Location A, it might not be so bad if you re-introduce that object positively in Location B.
Change the cue. If you called the dog to “come” and then lightning struck nearby, and now your dog is forever terrified of that word, you don’t need to expend a lot of effort repairing that cue. It’s probably easier for both you and the dog if you just use a different word, teaching the dog to run to you when you say “here” or “front” instead. The scary cue ceases to exist, and your dog has a new favorite behavior to practice!
Now, obviously some cues – particularly visual, contextual or environmental cues (Agility equipment, the vet’s exam table, the basement stairs, the harness, etc.) can’t easily be changed. In that case, you’ll need the next option.
Systematic Desensitization & Counter-Conditioning. This is where respondent conditioning can really come in handy, especially when paired with operant protocols. The dog is scared of an object? Pair that object with something pleasant. It’s even more effective if you use a systematic approach, starting well within the dog’s comfort zone and gradually increasing exposure to the stimulus. (There are a number of excellent guides to this system already in print, so I won’t try to replicate them in this quick overview.)
Ultimately, classical conditioning is going on ALL THE TIME, no matter what you or your animal are doing. Whenever possible, take advantage of it and make it work for you. Want your dog to be more energetic while heeling? Think of things you can do to associate that left-side position with exciting, enthusiastic play. Want your dog to be more confident in the vet’s office? Think of ways to associate good things with being in the building. Want your dog to calm down around strangers? Build a calm, relaxed state and pair it with the approach of new people. There are countless ways to incorporate respondent elements into your operant training.
And remember – the clicker itself is one of the most powerful applications of classical conditioning you will ever use. If one respondent pairing can be so useful, think of how much more efficient our training can be if we take advantage of classical conditioning elsewhere!