What is the impact of using words such as “ahh” or “no” when our learners get things wrong? When I was in university studying a Bachelors of Science in Canine Behaviour and Training I decided to do my dissertation on the use of Non-Reward Markers (NRMs) which are effectively words or markers that tell our dogs that they are wrong. I hear so many people using “no” or “ahh” when their dogs don’t perform the correct behaviour and I wanted to determine if it was useful. A lot of people believe that it is important to provide feedback if a learner makes a mistake but along with many others who advocate errorless learning, I suspected that it was unnecessary and potentially detrimental.
So my study consisted of me shaping 30 dogs to touch a cone with their muzzle – 15 using a clicker and treats for correct responses and doing nothing when they made a mistake and a further 15 using clicker and treats for correct responses and “ahh” when they got it wrong.
During the study I found that if I was teaching a really simple behaviour that had two options i.e. right or wrong, then using a non-reward marker such as “no” or “ahh” was effective – it immediately stopped the dog from offering the wrong response. On the other hand when I was trying to teach the dogs a more complex behaviour it significantly impaired the learning process. More worrying though is that those dogs subjected to the “ahh” showed significantly more stress response behaviours than those who were taught using an errorless learning process; they did not enjoy the learning experience nearly as much.
Avoiding the use of markers that highlight errors made the learning environment much better for the dogs and as a consequence they achieved the correct behaviour on average over 2 ½ minutes quicker – that’s over 25%!
So this is one of the reasons why I always try to avoid marking incorrect behaviour, especially when I am shaping a complex behaviour – I also find that the same principles apply to the human end of the lead!
Do you use non-reward markers in training? Do you find them helpful?
The abstract is below:
Shaping new behaviour through an audible marker (clicker) to acknowledge increasingly successful approximations is a well-established protocol developed with efficacy in mind. That said, disagreement exists regarding the benefits or drawbacks of using markers to acknowledge errors. Known as Non-Reward Markers (NRM), they are executed verbally by a timely, yet neutral “ahh”, and are designed to inform learners that no reinforcement will follow; opponents consider NRMs to be conditioned punishers. The purpose of this study was to determine if NRMs would enhance or impair learning. Two groups of 15 dogs (n=30) were trained to muzzle target a cone using a clicker to mark successful approximations, with one group also subjected to NRMs. The time taken (m=645 seconds) to complete the task for dogs exposed to NRMs was significantly longer (F= 0.555, df= 28, p= 0.002) than the control group (m=482 seconds), suggesting that NRMs impair the speed at which dogs learn new behaviour. There was also a significant difference (χ²= 10.465a, df= 1, p= 0.001) in the total number of stress related behaviours evident for dogs within the experimental group (NRM n=58, m=3.87, control n=28, m=1.87). The results of this study suggest NRMs could impair wellbeing and unnecessarily prolong the learning process when shaping dogs to perform novel behaviours. Consequently, the absence of NRMs in training protocols may improve efficacy.