I think in error based training we usually use negative feedbacks or sometimes punishments to train the patient, but we also can focus in positive feedbacks or encouragements to internalize and train the errorless steps. otherwise to mix these two systems can prevent habituation of patients to one repetitive training method and also prevent the fear of always having errors and consequently a bad response. and on the other hand errorless training may put some errors as a usual and normal step of the task. So combined method may be the superior system.

As I understand the document, the trail is ongoing and it is an interesting one.

http://www.trialsjournal.com/content/14/1/369

I would certainly never include punishment as a reaction to any training ever.....

Patients come to us to 'get better' not to 'get hurt'. I would like to wait until there are some results before I react in depth.

Dear Esther, you are certainly right. I did mean the patient may interpret the errors and negative response of trainer as a punishment. I am totally against the punishment you may concern in mind too! but the word punishment has a wide range of meaning see link below: http://en.wikipedia.org/wiki/Punishment

"The unpleasant imposition may include a fine, penalty, or confinement, or be the removal or denial of something pleasant or desirable"

For example what is time out? a method to put the child in a position that is not usual and pleasant for the child for a while, in order to accentuate the bad behaviors of him/her.

Thank you guys for your interest in the topic. From what I understood in the document, both methods use feedback and feedfoward. However, error-based learning aims at teaching the patient a structured method to self-evaluate its performance and anticipate further errors. They are encourage to use this method in trial-and-error learning. Conversely, errorless learning uses a lot more feedback and feedfoward to correct/prevent any error. It is proposed that this technique is efficient with patients with severe cognitive impairment, but the patients may become dependent of the therapist feedbacks.

It may be useful to consider two kinds of feedback: Informational feedback and motivational feedback. Informational feedback helps us understand when errors occur and help direct us toward better responses in the future. Motivational feedback provides reward or punishment. It is possible to avoid punishment for errors and reward participation but at the same time provide informational feedback to shape better future responses. Thus, in this version of "errorless learning," any response is "correct" but there is also information provided to shape future responses. Instead of "that was wrong, do it like this," the message is "that was great, now try it this way." This approach can be very helpful with people who have experienced a lot of failures.

I'm afraid that there is no general answer to this very interesting question, as there are good arguments for both approaches, and studies showing that both can work. I only know very few studies directly comparing error-based to error-free learning (e.g., http://www.tandfonline.com/doi/abs/10.1080/713756014?journalCode=pqja20#.UxXGGH6iYUc linking error-free learning to implicit learning), and even these studies probably cannot answer the question, as a error-based approach normally differs in many ways from an error-free approach, making it difficult to attribute group differences solely to the concepts of error-based vs. error-free.

In any case, I can recommend the review by Wolpert and colleagues on sensorimotor learning, perhaps you can extract some ideas about the general topic of motor learning: http://brain.phgy.queensu.ca/flanagan/papers/WolDieFla_NRN_11.pdf

Thank you Dr Kramer and Mcnevin for your answer and references! Could you give us more details or references for the bandwidth training method?

I agree that the literature is too scant for now (with only a few that directly compared both methods) to know when use one or the other. From the answers we had so far, it is clear that no matter which method we choose, the feedback must be positive to avoid de-motivation.

Dear Dr Mcnevin, thank you again for these details and reference! I understand better now what is bandwidth feedback. I find it much more representative of the kind of feedback that is generally used in neurorehab. For example, during the initial stages post-stroke, the sensorimotor deficits are more severe. Clinicians have therefore to select goals that can be acheived (with a certain degree of error) and give the proper feedback. While the patient is recovering, we have to adjust the goals to maintain the right level of difficulty (i.e. less error).

Another concept you might want to look into is "fading feedback", i.e., reducing the frequency of the feedback with increasing skill level, see for example http://physicaltherapyjournal.com/content/88/6/720.short

Interesting.  May be the answer is within cerebellum functioning.  It helps coordinating with basal ganglia and cerebral cortex for learning and motor memory.