-Homer Simpson
In Homer Simpson's common sense view, mistakes are universal, inevitable, and easily fixed. No big deal. I suspect he would be totally mystified by the debates about error-free learning. One group of educators argues that errors hamper learning and should be avoided; another group argues that errors are potentially beneficial as long as you know how to interpret and use them. I first met the idea of error-free learning in the 60s when our clinic began offering Lovaas-type therapies for autistic children. It made sense to me at the time as a way to help children view themselves as successful learners, and avoid the catastrophic reactions that for some children could be set off by the slightest failure. Forty years later, it makes much less sense to me to set error-free learning as a goal of teaching or intervention. It seems more important to work with children in a way that helps them develop healthy and productive responses to error. In this article, I discuss the ideas and evidence that led me to change my mind.
Who Invented Error-Free Learning and Why?
Error-free learning was a product of the behaviorist view of humankind. If the mind is viewed as an indiscriminate record of behavior, errors could well impede progress. In tracing the history of the term error-free learning, Pashler, Zarow and Triplett (2003) cite the behaviorist Guthrie who claimed animals, "Learn what they do (and especially what they do last)" (p. 5). With this view of learning it certainly makes sense to avoid doing the wrong thing. Minimization of error was incorporated into the programmed instruction of the 60s, and in particular into the ABA (applied behavioral analysis) programming of Lovaas in his work with autistic children.
Error-free learning wasn't just motivated by theory though. This approach to teaching was the then radical response of a group of special educators to the challenge of parents and the courts who said that persons with severe intellectual disabilities nevertheless had the right to learn (Tawney, Allen, O"Reilly, Cobb, & Aeschleman, 1974). These educators reasoned that the symptoms of successful learning included (a) new behaviors, and (b) the absence of errors. Error-free teaching methods seemed capable of producing both sorts of evidence of learning, and programmed lesson materials soon sprang up everywhere. More importantly, persons with severe intellectual disabilities in many states were treated as learners...and learned! This historical shift in educational opportunity was an important victory for social justice, but ultimately provided only weak support for the proponents of error-free learning. There were two problems with the evidence. The absence of errors might mean that learning had occurred, but was it the error-free nature of the materials that was the key to this success? And secondly, if there had been errors, would that have meant that learning had not occurred? Let's take these questions in turn.
The Evidence in Support of Error-Free Learning
When I went to the research reports, I discovered that scientists have been studying error-free learning for nearly a century. Here are three studies that typify this literature. You may be surprised at the outcomes.
Study 1
In 1919, Carr and Koch began a series of experiments that contrasted trial-and-error learning with "controlled" learning because "The value of errors...has not received any extended discussion in the literature on learning" (p. 292). Like many psychologists of the day, they investigated this issue by putting rats in a cleverly designed maze and counting the minutes it took for them to find the food. In this particular study, the rats had to choose between two routes, and the correct route (i.e., the one leading to the food) alternated from trial to trial. One group of rats was left to discover this pattern in whatever way they could and, naturally, made some mistakes along the way. The other group was physically prevented from going the wrong way by a gate, and thus made no errors on the training trials. Analysis of the learning curves revealed no differences in level of achievement between groups on later tests, and of the 7 rats that learned particularly well, 4 were trial-and-error learners and 3 were error-free learners. The researchers rightly decided that these results were inconclusive.
Study 2
In the 1990s, a group of clinical scientists at the Applied Psychology Unit in Cambridge studied the use of error-free rehabilitation therapies for persons with memory impairments. They were impressed by the successes reported by the special educators who were using error-free methods with children who had learning disabilities and they wished to extend this work into adult rehabilitation. In one of their early experiments (Wilson & Evans, 1996; Evans et al., 2000), the team asked persons with acquired memory deficits to participate in two tasks. In the first one, they had to learn the names of 15 people presented in photographs, and in the second one they had to learn a particular route through a large room again as presented in photographs. For each task, the participants learned half of the items with trial-and-error methods and half with error-free methods. In the "with error" condition, patients were told to name the person in the picture without ever having been given that information. All they were given was the first sound in the name. In the "error-free" condition, patients were immediately given the name. With each successive trial they got fewer and fewer letters of the name in print and had to remember the name with diminishing cues. Analogous differences in teaching method were used in the route task. The results indicated that indeed, error-free learning was more effective then learning with errors, but only in the proper name task. This task difference was subsequently confirmed in a number of experiments.