We recently saw an article on a blog about the use of negative reinforcement (www.EileenAndDogs.com – http://eileenanddogs.com/2015/07/30/natural-vs-contrived-negative-reinforcement/). One of the main points of the article seemed to be that “natural” negative reinforcement was acceptable, and didn’t cause harm or problems for dogs, while “contrived” reinforcement was to be avoided.
The article started by quoting Paul Chance’s book:
“Natural reinforcers are events that follow spontaneously from a behavior.”– Paul Chance, Learning and Behavior, 7th Edition, 2013
“Contrived reinforcers are events that are provided by someone for the purpose of modifying behavior.”–Paul Chance, Learning and Behavior, 7th Edition, 2013”
In the context of his book, Chance’s use of “contrived” is taken as “purposeful”. In fact, “purposeful” is in his definition. That’s what training is about – having a plan to purposefully modify behavior. Nothing wrong with that.
It’s also important to notice that Chance is talking about reinforcers , without reference to whether the reinforcer is negative or positive. Trainers use contrived positive reinforcement all the time! We couldn’t do most of the training we do without contriving reinforcement! So right away the premise that “natural” is good and “contrived” is bad goes right out the window.
There are uncountable behaviors we want dogs to do that we reinforce with food that would never spontaneously cause food to be delivered or fall from the sky. What about giving a treat when the dog sits on cue? That is certainly an example of a contrived reinforcer based on Chance’s definition. But I don’t know of anyone who believes that use of contrived reinforcement is bad for dogs.
The author states that “No longer does the human or animal necessarily respond with a behavior that directly relieves her discomfort”, in reference to contrived negative reinforcement. The same could be said of a “natural” situation involving negative reinforcement. Take a dog that is afraid of thunder. A behavior that “directly relieves her discomfort” would be the dog seeking out an area of the house or yard that minimizes her exposure to the sound of thunder. For our dog Coral, going to the basement is the behavior that should relieve or reduce her discomfort because the sights and sounds of the storm are significantly muted. But what do many dogs do? They jump the fence and run around the neighborhood – certainly not a behavior that reduces their exposure to the frightening elements of the storm.
The other problem with the author’s statement is the implication that it is the behavior itself that is the consequence rather than the termination of something unpleasant. If the behavior doesn’t increase in frequency, and result in termination/avoidance/escape from the aversive, then by definition it can’t have been negatively reinforced.
Negative reinforcement learning begins with the animal performing a behavior to escape the unpleasant stimulus, but ultimately (usually) results in avoidance of it all together. That’s the sign the task has been learned. In both the article and subsequent comments, it is stated that in “contrived” negative reinforcement, the aversive is presented repeatedly, supposedly not the case in “natural” negative reinforcement.
The important point is that in properly executed negative reinforcement learning, the animal learns to avoid the unpleasantness – meaning he doesn’t experience the aversive at all after the first few trials. That’s why negative reinforcement is often referred to as escape and avoidance learning. The animal is learning to escape/avoid the stimulus. The behavior that accomplishes that doesn’t have to be what we think of strictly as “escape” – running away – as the author so aptly points out in her hand washing example. Escaping or avoiding the aversive even in “natural” situations can be accomplished with any number of behaviors – standing still, jumping up, jumping over, backing up.
If the animal is experiencing the aversive over and over again, that’s a sign of bad implementation. Or perhaps it’s not even an avoidance task– just an example of someone who doesn’t understand how negative reinforcement – or learning theory in general – works at all doing something harmful to a dog.
In the example of the hot chaise lounge in the article – the hot surface would occur/be presented repeatedly – so what? What’s relevant is the dog learning not to jump on it because it was unpleasant. No properly executed avoidance learning forces an animal to experience an aversive without a means to turn it off. And usually without teaching the animal the correct response before applying the aversive.
Take no-pull harnesses, or even head collars/halters. No pull harnesses make use of “contrived” negative reinforcement, yet their use is generally accepted among the “positive” and even “force free” training proponents. And the unpleasantness of pressure on the chest from a no-pull harness is something dogs do indeed experience repeatedly because of imperfect use of the harness.
BAT is another example of the use of “contrived” (purposeful) negative reinforcement, and is still a highly regarded technique among most of those same segments of the training community. In BAT, behaviors are negatively reinforced repeatedly. Is that “bad”?
The harness also illustrates another erroneous assumption made in the article – that learning to escape or avoid an aversive stimulus requires actual escape or leaving. To avoid the unpleasant pressure/tension on chest (harness) or head (halter) requires just the opposite – stopping, slowing down, staying with the handler. In BAT, a lip lick might be the behavior chosen for negative reinforcement.
Have negative reinforcement procedures (or attempts at them) been improperly implemented and caused harm to dogs? Absolutely. Does negative reinforcement as a behavior modification technique always and consistently result in harm to dogs? Absolutely not. Are “contrived” reinforcers “bad” and “natural” reinforcers “good”? We shouldn’t even have to pose the question.
Most any technique, improperly implemented, can result in a bad outcome for the animal. If backward conditioning is used when attempting classical counter conditioning (good thing presented before the bad thing), conditioned food aversion can be the outcome, as Kathy Sdao has so aptly pointed out.
Rather than setting up these false dichotomies that do not help us choose the best technique to help a particular dog learn a different behavior in a way that benefits both the dog and the owner, let’s think of other guidelines that are more helpful.
I use the Four Keys to Cooperative Training© that I teach my students:
- Focus on the animal – if you are carefully and consistently observing what affect your training is having on the animal, that alone should tell you if what you are doing is producing stress, fear, or anxiety, and you need to modify what you’re doing.
- Follow the science – there is too much information these days that is being masked as “science based” that really isn’t – it’s either pseudo –science or one – sided science. Be willing to dig deeper and question “science based” claims.
- Think critically – that means asking oneself – how do we know this? What are other, alternative explanations? Where is the evidence?
- Train creatively – that’s the fun part! Coming up with new ways to elicit behaviors, make it hard for the animal to make a mistake, and easy to do what’s wanted.
Want to learn more about cooperative training and other strategic approaches to behavior consulting and problem resolution? Take our “Keys, Guidelines and Protocols for Working with Pet Behavior Problems” course from PetProWebinars.com
Leave A Response