REVIEW TRIAL-BY-ERROR VS ERRORLESS LEARNING, How might this apply when using No Reward Markers? 

I reviewed the following three articles written on “No Reward Markers” found at those links. I occasionally use NRM’s so am interested in what naysayers often say. My concern was the claims made that errorless learning is superior to trial-by-error learning. Using No Reward Markers could be said to be form of trial-by-error learning since the purpose informs the learner they’ve made an error and should try again. Please keep in mind; this review was several years ago (2012), and links often change and/or become lost:

1. No Reward Markers Jul/Aug 2010 http://www.lauraromanik.com/id114.html
2. NRMs No Reward Markers by Melissa Alexander on 07/01/2003 filed in Training Theoryhttp://www.clickertraining.com/node/179
3. Errorless Learning versus the use of No Reward Markers by Dog Tec Emily Larlman 2012http://clicktreat.blogsot.com/2012/06/errorless-learningt-versus-use-of-no.html

I was prompted (pun intended) to review these articles/blog posts because they often cite science, example “errorless learning as opposed to trial and error learning has been scientifically proven with animals and humans to”  (Larlman, 2012) and included these statements as evidence.

“Minimize the number of errors in the training session” (Larlman, 2012)

“Decrease time spent learning a skill”

“Reduce future errors, as they have never been practiced” (Larlman, 2012)

“Create less frustration, stress, and aggression” (Larlman, 2012)

“Not inhibit behavior” (Larlman, 2012)

“Not create a conditioned emotional response associated with punishment to any part of the behavior task” (Larlman, 2012)

“Not create a conditioned emotional response associated with punishment to the trainer or the training environment” (Larlman, 2012)

Errorless learning has not been “proven” to be “scientifically” superior in dogs, not even in humans. This is where the bone of contention is often debated!

DO I CARE IF SOMEONE ACCEPTS OR REJECTS THIS INFORMATION?

Absolutely not. Am I concerned? Yes, because it’s commonplace in materials disseminated in the dog training industry to use words like science and/or scientific evidence without references and/or include critical (alternative) reviews.

Let’s keep alternative views secret, even if they’re actually published in scientific Journals, so we can perpetuate one’s own point of view. There is nothing wrong with this choice until someone finds actual published articles that dispute such claims. It is not intended to pick on any particular person, other than I have limited time to bring this issue forward, so those who might be interested can actually critically review the science themselves. It really helps one learn to think critically, remain skeptical of claims citing science when there are no references to actual published papers. That should be your first clue. One should also be familiar with both concepts of learning, errorless and trial-by-error.

I reviewed each article. I’m going to leave it up to you to critically read these example articles. In addition, you should read “The Effectiveness of Trial-and-Error and Errorless Learning in Promoting the Transfer of Training”, Robert S. P. Jones, Linda Clare, Conor MacPartlin and Olivia Murphy School of Psychology, Bangor University, UK, published in European Journal Of Behavior Analysis, July 2010, 11, 29-36 Number 1 (Summer 2010).

To whet your appetite here are some key statements included in this published journal paper.

1. Despite an extensive literature showing successful application of errorless techniques in applied settings there has been some debate about the optimal circumstances surrounding the use of these techniques.
2. The general paradigm involves a variant of traditional stimulus discrimination training whereby the S+ (the stimulus associated with reinforcement) is easily available, while the S- (the stimulus not associated with reinforcement) is difficult or impossible to access in the early stages of training”
3. Jones and Eayrs (1992) later expanded on this suggestion stating that although errorless learning may confer an advantage when learning a skill for a situation that will never change, such techniques may not be superior when training has to be transferred to a novel situation.
4. There is no longer a single methodology than can be said to represent ‘errorless learning’ per se, and the field has moved beyond the simple fading paradigm adopted by Terrace (1963a, 1963b).
5. “healthy student participants were deliberately chosen to evaluate the initial testing of this hypothesis.
6. Data from both experiments show that during the transfer task, the trial-and-error group performed significantly better than the errorless group when there was a requirement to transfer the skill learned during the acquisition phase to a new related task,
7. This is not to deny the huge advantages that errorless learning can provide in a variety of training situations (e.g. Jerome, Frantino, & Sturmey, 2007). It is important, however, not to rely on this technique exclusively.
8. The present experiments show that making errors when learning may be more advantageous when training has to be transferred to a novel situation.
9. Clinical implications of this finding point to the importance of case-by-case decision making in relations to the choice of learning methodologies employed.
10. Generic statements about the superiority of errorless learning over trial-and-error learning are not sustainable (Mount et al., 2007).

The authors closing statement is included in entirety as it includes an important distinction concerning individuals ability to monitor and detect errors. It is also important to note this article addressed using both methodologies to treat impaired individuals. Since we are concerned with dogs, the same would apply, selection of methodology on case-by-case basis and that in a normal training environment, working with normal dogs, trial-by-error may be a wiser choice with faster learning possibilities. Whereas when working with abnormal expressed behavior in dogs/cats (other animals) errorless learning may be the better choice.

“In applied settings the advantages of learn­ing using errorless techniques may need to be balanced against slower but more generalizable learning using more traditional trial-and-error learning paradigms. This will not always be an easy decision to make. There may be situations where individuals with severe memory problems will be effectively unable to learn with trial-and-error techniques. In such circumstances error­less learning may constitute the only available methodology to promote new learning (Clare & Jones, 2008). Errorless techniques may facilitate a much more rapid acquisition of new material when compared to trial-and-error learning and this may have important clinical advantages for client and carer alike. In other situations, people who retain the ability to monitor and detect errors, and to update knowledge of their performance on the basis of feedback, may fare better with a rehabilitation intervention based on trial-and-error principles (c.f. Kessels, van Loon, & Wester, 2007). As was predicted by Gollin and Savoy (1968), it does indeed ap­pear that any learning situation that involves the need for a later transfer of learning to novel situations is likely to confer advantages to a trial-and-error methodology.”

An additional source for critical review on this same subject, specifically the use of verbal feedback and/or trial-and-error learning, I recommend reading make it stick The Science of Successful Learning by Peter C. Brown, Henry L. Roediger III and Mark A. McDaniel, 2014.

If you wish to forgo reading this book in entirety, skip to page 90-94, and read Failure and the Myth of Errorless Learning.

Enough on No Reward Markers, already! Their use requires first defining their use by an individual! It is apparent from reading the nays and yays that each person has their own individual interpretation on using No Reward Markers.

Taking into consideration trial-and-error learning provides essential feedback to the individual learner, both positive and negative, using No Reward Markers can potentially provide that same kind of critical feedback. Taking this one step further, this critical error feedback, can potentially save individuals from making the same mistakes in new (novel) environments!

AND here’s quote from Karen Pryor! 

Karen Pryor, in her book “Getting Started: Clicker Training for Dogs,” revised edition, copyright 2002, page 53, answered this question.

“But doesn’t that mean I should never punish the dog? What if the dog jumpsup, or nips me, or takes food off the counter, or runs away?”
“Some people think that using positive reinforcement means you never reprimand the dog or control it physically. That’s unrealistic. Leashes are a fact of life for dogs. It’s necessary to keep a dog on a leash when you are going to strange places, or out in traffic, or amongst strange dogs. And of course your dog needs to understand the meaning of “No”. You need to interrupt behavior such as mouthing your hands and clothes, for example, or trying to grab food from the kitchen counter. Remember that timing is just as important in correcting misbehavior as it is in reinforcing good behavior. Your response should occur while the behavior is happening, not afterwards, or just before you think it might happen.

While correction and scolding can stop behavior (while you’re around, anyway) they are not efficient ways to teach the dog to do something new. For that, clicks and treats work best. As for stealing food, tipping over the garbage, and so on, it’s up to you to monitor the environment and put temptation out of reach.”

Other Reads:

Errorless learning how could that be a bad thing? by Dr. Deb Jones