Errorless Learning

At some point in my training in neuropsychological rehabilitation, I was introduced to the concept of errorless learning. This technique was designed to help folks with cognitive impairment accomplish tasks with greater accuracy by removing the opportunity for error.  A non-clinical example:  if I want to teach my daughter how to lock the front door, I may break the task down into 3 steps and then train her in each task in a hierarchical fashion. “Okay honey, Step 1: We check the position of the lock. If it is up-and-down, that means it is unlocked.” Once my little angel masters the check, she is ready for Step 2. “Okay honey, Step 2: Turn the lock that way. We only turn the lock that way when we want it to be locked. Now go ahead, you try, turn the lock that way.”

You can see the advantages of such an approach, right? If not, I invite you to observe me in the morning trying to figure out where I keep my socks. Rather, I invite you to read what I write about the matter. You see, I have a 6 drawer dresser. Only 1 of those drawers contains socks. On a good day I will open the sock drawer on my 4th attempt. On a less-than-good day, I will open the sock drawer on my 7th attempt (yes, there are only 6 drawers, but I may very well open and close the same drawer 2 or 3 times in the course of my haphazard quest.). The process is kind of like playing a game of Guess Who with my children, “Does the drawer have a chip on the corner? No? Ok, that eliminates the top drawer. Is the handle loose? No? Ok, can’t be the bottom one. Hmm, only 4 left. Does it have a brown sock-like item protruding out from the top? Yes? Alright, I win!!”

So if my daughter wanted, she could reciprocate the lock lesson and employ an errorless learning paradigm to make my SockQuest far more efficient. “Daddy, socks are kept in the 2nd drawer from the top. If you want your socks, open up that drawer. No, not that drawer, stop right there, that’s where your undees are kept. You are already wearing your undees, right now we are looking for socks. Ok, there you go, 2nd drawer from the top. Now you got it.”

However, can you see the disadvantages to this approach? Here’s what I see; you can train a person using errorless learning procedures to find her socks. However, unless your trainee works in a sock store and her sole responsibility is to retrieve socks from the correct drawer on the first try, you have not helped her in a way that is all that meaningful.

In the SockQuest example, my daughter can teach me how to find my socks in 1 second flat, but has she helped me with what matters most in my life (i.e. my Work/Love/Play triad). So errorless learning is wonderful for turning out research with impressive effect sizes because it works. The limitation, however, is what it “works” for. See, for example, Erica L. Middleton and Myrna F. Schwartz’s 2012 article in Neuropsychological Rehabilitation where the authors conclude that errorless learning is quite effective for straightforward and highly circumscribed tasks but not very useful for teaching more complex ones.

Could you imagine if I actually recommended an errorless learning paradigm to a frustrated wife whose husband has been unemployed, irritable, and reclusive since his traumatic brain injury?

Dr. Gross: Yes Mrs. Watson, I would like you to teach Mr. Watson how to pick out his socks and lock the door in a hierarchical fashion. Once you do that successfully, please come back to my office so I can record how much he has  improved.

Mrs. Watson: Improved in what?

Dr. Gross: Well, improved in locking the doors and picking out his socks, of course.

Naturally, the limitation here is that Mrs. Watson did not come to see me because of locks & socks. She came to see me because her husband has been impossible to live with ever since his Work/Love/Play triad was busted up in the car wreck! Errorless learning would only “help” Mrs. Watson become even more dissatisfied with her husband by having her become his 4th grade teacher.

The alternative to errorless learning, say Middleton and Schwartz, is potentially errorful learning, which they argue is better suited for robust learning and prolonged performance gains (i.e. life).  So for Mr. Watson, pick a few key tasks for which to apply errorless learning training (i.e. helping him to self-administer his medications and pay his gas bill) and spend more time helping Mr. Watson increase his tolerance for the unavoidable fact that his will be making A LOT MORE ERRORS since his car wreck. Then help him to identify potentially errorful learning strategies to accomplish the complex tasks he cares about. And in the process, help him to redefine and recreate his Work/Love/Play triad.