Verified by 
By John Sweller (University of New South Wales)
Imagine a student learning to solve mathematical problems or learning to write an essay in a history class. With time and practice, they are likely to improve—but what are they actually learning that is driving the improvement? What should we be teaching them to assist improved performance? These questions are fundamental to those areas of educational psychology concerned with cognitive processes and instructional design. In order to answer these questions—and to guide educational practice—a solid evidence base is needed. Cognitive load theory (Sweller, Ayres, & Kalyuga, 2011) was designed to address these questions. It is an instructional theory that generates testable hypotheses using randomized, controlled trials. The data generated by these trials are needed to create an evidence base that can be used to inform educational practice. Curiously, over the years, cognitive load theory’s emphases have turned out to be at odds with much of the work currently carried out in the fields concerned with cognitive processes and instructional design.
Students use cognitive procedures such as general problem solving strategies, planning, monitoring, and adjusting their thought processes. These are general cognitive processes essential to functioning irrespective of the curriculum area. Much research examining cognitive processes and instructional design is concerned with teaching generic cognitive skills that cross curriculum boundaries. Although domain-general skills are essential, we do not need to teach them.
Rather, we need to teach domain-specific skills. Students need to learn to read and write specific words, sentences, and combinations of sentences. They need to learn how to solve specific types of arithmetic and algebra problems. Learning how to solve algebra problems is different from learning how to write. These skills are domain-specific and need to be explicitly taught.
Cognitive load theory is concerned with teaching domain-specific (rather than domain-general) skills. The theory assumes that skilled performance in any area comes from large amounts of domain-specific knowledge stored in long-term memory after extensive practice over many years. The main aim of the theory is determining how to present information to learners in a manner that reduces unnecessary working memory load so that they can acquire the immense, domain-specific knowledge bases required by anyone living in a sophisticated society. It assumes that schools and other educational institutions were devised precisely to enable learners to acquire that knowledge. Lastly, and somewhat controversially, it assumes that domain-general skills do not need to be taught because they will be learned without tuition (Tricot & Sweller, 2014).
Although cognitive load theory indicates that we need to explicitly teach domain-specific skills—as they are not learned automatically without instruction—this notion is often at odds with commonly held beliefs about teaching and learning. How would one learn generic cognitive skills without instruction? Generic cognitive skills (such as learning to plan) are far more important than domain-specific skills; we can survive as humans without knowing mathematics, but we cannot survive without having acquired general problem solving skills. And, because general skills such as “think of a similar problem” are relevant to solving any problem in, for example, mathematics, we have evolved to acquire them easily, automatically, and without conscious effort (Geary, 2012). Domain-specific skills, on the other hand, will rarely be acquired without instruction. Those skills need to be explicitly taught, and educational institutions were invented precisely because those skills tend not to be acquired simply by membership of a functioning society. Cognitive load theory tells us how to organize information and its presentation to facilitate the acquisition of domain-specific knowledge, and it assumes domain-general knowledge will take care of itself. Empirical evidence for the theory can be found in many hundreds of randomized, controlled experiments testing the predictions of the theory applied to a variety of students from early learners to adults (Sweller et al., 2011).
Ultimately, of course, whether we need to teach generic cognitive skills is an empirical issue. If experiments can demonstrate that teaching generic cognitive skills improves performance across a wide range of unrelated areas, then generic cognitive skills need to be taught. Educational psychology, like all science, is an empirical as well as a theoretical discipline. For example, let us assume that we have hypothesised that teaching students general problem solving skills will benefit their problem solving performance. We could run an experiment in which one group of learners was taught some general problem solving strategies (such as “When solving this problem, try to think of a similar problem for which you know the solution”), while another group was simply presented with the problem without instructions on how to solve it. Both groups then would be asked to solve transfer test problems that were entirely unrelated to the learning problems. The test problems must be unrelated to the learning problems to ensure that any difference between the two groups was not due to domain-specific skills. If the group taught the generic skill demonstrated improved test performance on transfer problems compared to the group not taught the generic skill, that result would provide evidence for a learnable/teachable generic skill. To this point, however, there is no body of empirical literature demonstrating such a result despite decades of advocacy. The onus is on those who advocate the teaching of generic skills to provide that literature using the standard experimental procedures of psychology.
In order to answer fundamental questions regarding teaching and learning—and to guide educational practice—theories must be in accord with empirical evidence. I believe the failure to provide empirical evidence from randomized, controlled trials supporting the teaching of generic cognitive skills is telling. Of course, the crowning glory of all scientific investigation is that we never can assume that we have obtained definitive, final answers. Our theories must constantly change and develop along with the available evidence. Educational science, like all science, should rest on empirical foundations. Our procedures should depend on a solid evidence base, not ideology. Educational psychology, with its rigorous procedures, can provide that base. By using randomized, controlled trials to empirically test competing viewpoints regarding teaching domain-general and domain-specific skills, we can inform effective teaching and learning across the lifespan, meeting the criteria outlined in Karen Harris’ “Message from the President.”
This post is part of a special series contributed in response to Karen R. Harris’ Division 15 Presidential theme, “Impacting Education Pre-K to Gray.” President Harris has emphasized the importance of impacting education by maintaining and enriching the ways in which educational psychology research enhances and impacts education at all ages. Such impact depends upon treating competing viewpoints with thoughtfulness and respect, thus allowing collaborative, cross/interdisciplinary work that leverages what we know from different viewpoints. She has also argued that we need to set aside paradigm biases and reject false dichotomies as we review research for publication or funding, develop the next generation of researchers, support early career researchers, and work with each other and the larger field.
Geary, D. (2012). Evolutionary Educational Psychology. In K. Harris, S. Graham & T. Urdan (Eds.), APA Educational Psychology Handbook (Vol. 1, pp. 597-621). Washington, D.C.: American Psychological Association.
Sweller, J., Ayres, P., & Kalyuga, S. (2011). Cognitive load theory. New York: Springer.
Tricot, A., & Sweller, J. (2014). Domain-specific knowledge and why teaching generic skills does not work. Educational Psychology Review, 26, 265-283. doi: 10.1007/s10648-013-9243-1
