Successive approximations work best in training when a discrete, conscious behavior is being shaped.
In our previous installment of the Elevation Blog Series, we talked about when learning occurs with respect to elevated hides. (You can read that blog here: https://scentsabilitiesnw.com/blog/elevation-blog-series-3-when-does-learning-occur/) In this installment, we will explore why the practice of rewarding a dog for a glance up (using successive approximations) may actually hurt your training.
Successive approximations are powerful tools when used for training dogs. It’s a technique used when “shaping” a dog to perform a behavior. When we “shape” a dog, we mark the dog for incremental “correct behaviors” in order to evolve towards our chosen behavior.
Here’s a shaping video that I found of a training session that I did with Brava when she was 6 months old. She is quite successful but note that the training is based on her guessing and me telling her if she is guessing correctly.
This technique is often used in Nosework when teaching elevated hides. The dog is rewarded for a glance up. This is especially common when the hide is very difficult and the handler wants the dog to win.
This practice can cause more confusion when applied to High Hides than good. Why?
When you were a child, did you ever play “Hot or Cold”? In Hot or Cold, one child searches for an object guided by the other children’s callings of different temperatures based on the searchers’ proximity to the object: “cold”, “warm”, “warmer”, “hot” etc. The closer the first child gets to the chosen object, the “hotter” they are. This is basically shaping!
The issue when applying this to finding a high hide is that the handler is essentially giving information to the dog. Granted, the dog doesn’t get rewarded for sequential guesses… in that search… but over time this practice takes accountability away from the dog and puts the knowledge in the hands of the human, not the dog. This is fine for shaping behaviors… but unless the dog truly understands the outcome, all you might actually succeed in shaping is a head lift when the dog is confused. Then, when this is applied in a blind situation, the handler starts to call Alerts when the dog is confused and lifting the head. This is one way that a handler can manufacture a hide! It’s also a great way to extinguish actually searching for elevated hides.
In Nosework, we really DON’T want the dog to think we have reliable information about the hide
This is really a general statement. We never want the dog to start to rely on us to find the hide. Not only does that lower the dog’s self confidence in general, but it also sets the team up for failure when they have to search under blind searching conditions. We often see the dogs who have become habituated to getting information from their handler when we see a dog asking questions such as “Is this it?”, “Is this it?” This questioning messes with the handler’s trust and soon you have a downward spiral.
In the case of elevated hides, the handler will stop calling them because of a history (even a brief one) of false alerts. Very quickly this can turn into working a dog who ignores elevated odor altogether.
By rewarding for a glance up, we are teaching the dog that we have information. Soon the dog starts to glance up when confused because that has worked in the past. Just because you reward at a glance up doesn’t mean that the dog has any idea WHY they have just been rewarded.
As long as you reward for effort after leaving the search area, it’s OK not to find a hide
It’s totally ok not to find a hide! In fact in general, I encourage my students to sometimes search beyond finding the last hide and end on searching rather than a find in preparation for unknown number of hides.
Your dog may have put out significant effort. You can STILL reward that! Just leave the search CHEERFULLY and then give a cookie outside of the search area. (Make sure you wait until you leave the search area or you can create a dog who ends the search prematurely.)
Dogs really DON’T have to find all of the hides to be happy!
Giving the dog successive information about the hide assumes that the odor is behaving consistently with the human’s direction
Keep in mind that when you mark for a glance up that the odor might NOT be in that area! Just because we assume that odor is in a certain location doesn’t mean that it is. In fact, you can have a scent void directly under the hide.
In this video, we can see that Judd alerts around the corner of the building. Even though odor is under the eave of the building closer to the glass doors, the odor is around the corner. Had I set this on my own, I never would have guessed this initially. However, it makes sense. The odor is lifting due to the high ceilings and traveling under the eaves. There is a scent void directly under the hide… meaning NO SCENT THERE.
Odor can MISBEHAVE! Let’s watch to see what a difference a day makes
This is a rather long video but worth watching. The orange and the red smoke is from the first day. The blue and the purple smoke is from the second day. Setting a high hide doesn’t mean that the odor will fall, nor does it mean that the odor will fall in a location accessible to your dog. Now what happens if you reward a glance up?
If you want your dog to be confident in finding elevated hides, clarity in marking the AH-HA moment is much more helpful
Remember how we talked about the Ah-Ha moment in Part 3 of the blog series? That is the point where the dog REALIZES where the hide is located. If you want repeatable behavior, your dog needs to know WHY he’s getting rewarded. You will also find that your dog’s confidence will soar knowing that he CAN find the high hide.
Consistent rewards and allowing the dog to learn how to be correct without our input is essential in creating a CONFIDENT and INDEPENDENT worker.
See also: Bulletproof Elevation, a 3 Part Workshop