Frustration is an unwanted side effect of positive dog training. Although we strive for errorless learning and errorless performance, we can�t always control the antecedents as well as we might like. Rewarding a dog for the behavior we want has the consequence that if a dog expects a reward and that reward is withheld, some frustration might ensue.
Recently there has been a lot of debate about the place of �no-reward markers� (NRMs) in dog training. Some trainers have argued that an NRM is a useful piece of information for the dog and can prevent longer-term frustration, others claim that it is an aversive and therefore has no place in �purely positive� training. Some clicker trainers claim that there is no need for an NRM because not clicking for a behavior is sufficient for the dog to understand that he must try something else to get a reward.1
For the purposes of this article, I am defining an NRM as a noise or gesture that signifies to the dog that no reward will be presented for a behavior, but that does not cause the dog extra pain or fright. The dog has to learn the connection between the no-reward marker and not being rewarded in the same way as he learned the connection between the clicker and a reward. So, rolling one�s eyes and making a noise like �oops!� count as no-reward markers; dropping one�s keys on the kitchen floor or kicking the dog in the ribs do not.
It�s reasonable to assume that the presence of an NRM would eventually become mildly frustrating in itself - pair it enough times with the disappointment of not getting an expected reward. So when we use an NRM, we�re causing the dog to become mildly frustrated - this is a conditioned emotional response.
Not clicking doesn�t cause the dog to become frustrated by itself - the frustration comes from an expected reward being withheld. This means that the frustration the dog feels isn�t tied to an event like an NRM but it does still happen; the dog realizes that no expected treat is coming at some point after he performs the behavior, and that�s frustrating for him. The difference is, in this case the trainer hasn�t performed an action that frustrates the dog, the dog has become frustrated �on his own�.
So is there a moral difference between causing frustration with an NRM, and allowing frustration to happen by withholding a click? I will argue that, in this particular case, there is not.
In ethics, there is an ongoing debate about whether it is morally worse to cause harm than it is to allow harm to happen. The general consensus is that it is worse to do harm than to allow harm, and there are literally hundreds of examples that illustrate the distinction.2
One justification for the claim that doing and allowing are morally different has to do with probability - by doing an action, we make it certain that the consequence of that action will occur. By allowing the action, we only make it much more likely that the consequence will occur. There might, for example, be a tiny chance of a recovery if we turn off a patient�s life support, but no chance at all if we perform euthanasia.
In dog training, we could make the claim that using an NRM definitely will cause frustration, whereas not clicking only probably causes frustration. Therefore, it could be argued, we should avoid using the NRM.
This argument doesn�t work, however, because adding an NRM doesn�t make frustration more likely. We could imagine that although many dogs get frustrated if they don�t get the reward they expect, some just don�t. However, a dog that doesn�t get frustrated when he doesn�t get the reward he expects also wouldn�t be frustrated by an NRM. If not getting an expected reward just isn�t frustrating, he�d never develop a conditioned emotional response to the eye rolling, or the word �oops!�. It would just be information. The likelihood of frustration depends on the dog, and on the trainer�s skill, not on the mere presence of an NRM.
The factors that make the difference between examples of doing and allowing in ethics don�t exist in dog training - factors like whether the harm would have happened if we were not present, whether we intended to cause the harm and so on. This is because when we are training a dog, we are putting him in a situation we have decided he should be in and we have created. What we do, or don�t do, in that situation is always our responsibility.
The frustration the dog feels at not getting a reward is our responsibility whether we cause it with an NRM or allow it to happen by withholding a click. This being the case, it is better to focus on the way training is set up and create conditions to minimize frustration rather than focus on how that frustration comes to happen.
Whether a dog gets frustrated during training is down to a combination of the dog�s personality, the task at hand, and the trainer�s skill in setting him up for success, not merely whether an NRM is used. If both approaches can lead to frustration, there is no obvious ethical difference between using an NRM and withholding a click. If there is a difference, it must be empirical - that one approach leads to more frustration; that the conditioned emotional response to the NRM is worse than the frustration of not getting a click.
1 Materials like this article from the Karen Pryor Academy make this claim.
2 The Stanford Encyclopedia of Philosophy has an excellent introductory article to the Doing and Allowing distinction with a bibliography.
No comments:
Post a Comment