As a response to a recent Tweet by Eric van den Branden, and as a service to Pedro de Bruyckere (@thebandb) here a piece that I originally wrote for Van 12-18 in 2013.

OK, we’ve now handled (and debunked) learning styles. But, what if – just what if – there was any value to learning styles for learning and teaching? If we set choose to set all of the difficulties associated with the measurement of learning styles aside, the next question would then be how to tailor instruction to particular learning styles and would this lead to better teaching and better learning?

This is where the learning-styles hypothesis (Pashler, McDaniel, Rohrer, & Bjork, 2008) comes into play. The required evidence for the learning-styles hypothesis is a crossover interaction in which type A learners learn better with instructional method A, whereas type B learners learn better with instructional method B. For example, according to this hypothesis, verbal learners will learn best with verbal instructional methods (e.g., reading a book) whereas visual learners will learn best with visual instructional methods (e.g., watching a video). Here, it is important to note that a statistically significant interaction between learning style and instructional method is in itself not enough to have any practical educational meaning. Van Merriënboer (1990), for example, compared two instructional methods for teaching computer programming: The ‘generation method’ stressed the writing of programming code (matching an impulsive approach), whereas the ‘completion method’ stressed the study and completion of existing programming code (matching a reflective approach). Although there was a tendency showing that reflective learners profited more from the completion strategy than impulsive learners, the completion strategy was yet superior to the generation strategy for both types of learners. Thus, studies may show interactions between learning styles and instructional methods that have no practical educational implications; only cross-over interactions provide acceptable evidence for the learning-styles hypothesis.

In learning styles literature, the theoretical basis for the formulation of cross-over interactions is typically based on a preferential model, also called the meshing hypothesis. The basic idea is that instruction should be provided in the mode that best matches the learner’s style; thus, the learner is assumed to “know” what is best for him or her. As already indicated, the preferred way of learning, however, does not need to be the most productive way of learning. Let us make a comparison with food. Suppose that we ask children what food they prefer. Some children might prefer fruit and milk but the majority will prefer candy and soft drinks. Would this be a justified reason to give these children the food they prefer? We think not, simply because the preferred food will have a negative effect on their health. Similarly, Clark (1982) found in a meta-analysis of studies using learner preference for selecting particular instructional methods that learner preference was typically uncorrelated or negatively correlated to learning and learning outcomes. That is, learners who reported preferring a particular instructional technique typically did not derive any instructional benefit from experiencing it. Frequently, so-called mathemathantic (from the Latin mathema = learning + thanatos = death) effects are found, that is, teaching kills learning when instructional methods match a preferred but unproductive learning style (Clark, 1989). In such situations, a compensation model that makes up for the undesirable effects of a particular learning style would be more applicable than a preferential model.

Another complication that has, to our knowledge, not been discussed in the literature so far is that a learning style that might be desirable in one situation might be undesirable in another situation due to the multifaceted nature of complex skills. Suppose that one is teaching in the medical domain: One instructional method may emphasize learning-by-observation and matches – according to the preferential model – a reflective style while the other instructional method may emphasize learning-by-doing and better matches an impulsive style. When medical diagnosis is taught, it is particularly important that learners base their decisions on all available information and systematically exclude alternative hypotheses before they reach a diagnosis; the preferred style for learning this particular complex skill is a reflective one. For teaching medical diagnosis, it thus makes sense to use learning-by-observation for reflective students but it seems better not to teach impulsive students through learning-by-doing but rather to compensate them for their counterproductive style. In contrast, when emergency care is taught, it is particularly important that learners can make decisions on the basis of incomplete information and give priority to speed over fastidiousness; the preferred style is an impulsive one. Consequently, for teaching emergency care, it makes sense to use learning-by-doing for impulsive students but it seems better not to teach reflective students through learning-by-observation but now to compensate them for their counterproductive style. The point is that simple two-way interactions between learning styles and instructional methods, either based on a preferential model or a compensation model, do not take additional relevant factors such as the nature of the knowledge and skills that are taught or the context in which they are taught into account.

Given these complexities, it is particularly interesting to search the literature for studies that report crossover interactions between learning styles and instructional methods, irrespective of the question whether these interactions are based on a preferential model or a compensation model. Unfortunately, review studies indicate that the evidence for such interactions is almost completely absent (see Pashler, McDaniel, Rohrer, & Bjork, 2008). One exception is the study by Sternberg et al. (1999) reported earlier, but due to its methodological limitations (i.e., reported results were based on only a fraction of the participants) this study provided less than compelling evidence. On the other side, several well-designed recent studies contradict the learning-styles hypothesis. Constantinidou and Baker (2002), for example, found no relationship between a visual learning style and the actual learning of verbal items that were presented either visually or auditory. In a series of carefully designed experiments, Massa and Mayer (2006) also found no support for the idea that different instructional methods, emphasizing either pictorial or verbal information, should be used for visualizers and verbalizers. Similar negative findings are reported for other learning styles. In the medical domain, Cook, Thompson, Thomas, and Thomas (2009), for example, found no support for the idea that different instructional methods, working either from problems to theoretical information (inductive) or from theoretical information to problems (deductive), should be used for sensing/concrete learners and intuitive/abstract learners.

To summarize, the idea that learners with different learning styles should be taught with different instructional methods is a belief for which little, if any, scientific evidence exists. There are fundamental problems with regard to the measurement of learning styles and the theoretical basis for the assumed interactions between learning styles and instructional methods, and, last but not least, substantial empirical evidence for the learning-styles hypothesis is missing. We would like to stress that this does not mean that instruction should not take individual differences into account. For example, there is scientific evidence that objectively measured abilities and, especially, prior knowledge should be taken into account when instructional methods are applied. The expertise-reversal effect (Kalyuaga, Ayers, Chandler, & Sweller, 2003), for example, indicates that learners with low prior knowledge learn more from studying examples than from solving the equivalent problems and that this pattern reverses for learners with higher prior knowledge. The point we want to make is that there is no scientific evidence whatsoever that learning styles provide a useful construct to base the selection of instructional methods on, not that instructional methods should not be sensitive for differences between learners. Despite decades of research, the field of learning styles has failed to make significant progress and so far it does not yield any valid educational implications.

0 0 votes
Article Rating
Abonneren op

Deze site gebruikt Akismet om spam te verminderen. Bekijk hoe je reactie-gegevens worden verwerkt.

1 Reactie
oudste meest gestemd
Inline Feedbacks
View all comments

About Paul Kirschner

Nederlands: Prof. dr. Paul A. Kirschner, dr.h.c. is Universiteishoogleraar en hoogleraar Onderwijspsychologie aan de Open Universiteit. Hij is ook Visiting Professor Onderwijs met een leerstoel in Leren en Interactie in de Lerarenopleiding aan Oulu University (Finland) waar hij ook een Eredoctoraat heeft (doctor honoris causa). Hij is een internationaal erkende expert op zijn gebied en heeft zitting gehad in de Onderwijsraad in de periode 2000-2004 en is lid van de Wetenschappelijk Technische Raad van SURF. Hij is Fellow of the American Educational Research Association (AERA; NB de eerste Europeaan aan wie deze eer werd toegekend), de International Society of the Learning Sciences (ISLS) en van de Netherlands Institute for Advanced Study in the Humanities and Social Science of the Royal Dutch Academy of Sciences (NIAS-KNAW). Hij was President van de International Society for the Learning Sciences (ISLS) in de periode 2010-2011. Hij is Hoofdredacteur van de Journal of Computer Assisted Learning en Commissioning Editor van Computers in Human Behavior, en hij is auteur van Ten steps to complex learning (Routledge/Erlbaum). Hij schrift ook regelmatig voor Didactief (de kolom KirschnerKiest over wat docenten kunnen met wetenschappelijke resultaten). Hij is ook medeauteur van het boek Jongens zijn slimmer dan meisjes XL (EN: Urban Myths about Learning and Education). Hij wordt gezien als expert op veel gebieden en vooral computerondersteund samenwerkend leren (CSCL), het ontwerpen van innovatieve, elektronische leeromgevingen, mediagebruik in het onderwijs en het verwerven van complex cognitieve vaardigheden. English: Paul A. Kirschner (1951) is Distinguished University Professor and professor of Educational Psychology at the Open University of the Netherlands as well as Visiting Professor of Education with a special emphasis on Learning and Interaction in Teacher Education at the University of Oulu, Finland where he was also honoured with an Honorary Doctorate (doctor honoris causa). He was previously professor of Educational Psychology and Programme Director of the Fostering Effective, Efficient and Enjoyable Learning environments (FEEEL) programme at the Welten Institute, Research Centre for Learning, Teaching and Technology at the Open University of the Netherlands. He is an internationally recognised expert in the fields of educational psychology and instructional design. He is Research Fellow of the American Educational Research Association and the Netherlands Institute for Advanced Study in the Humanities and Social Science. He was President of the International Society for the Learning Sciences (ISLS) in 2010-2011, member of both the ISLS CSCL Board and the Executive Committee of the Society and he is an AERA Research Fellow (the first European to receive this honour). He is currently a member of the Scientific Technical Council of the Foundation for University Computing Facilities (SURF WTR) in the Netherlands and was a member of the Dutch Educational Council and, as such, was advisor to the Minister of Education (2000-2004). He is chief editor of the Journal of Computer Assisted Learning, commissioning editor of Computers in Human Behavior, and has published two very successful books: Ten Steps to Complex Learning (now in its third revised edition and translated/published in Korea and China) and Urban Legends about Learning and Education (also in Dutch, Swedish, and Chinese). He also co-edited two other books (Visualizing Argumentation and What we know about CSCL). His areas of expertise include interaction in learning, collaboration for learning (computer supported collaborative learning), and regulation of learning.