Learning styles as a neuromyth
I found this article to be intriguing. It’s a long post but worth the read. Touloumakos, Vlachou & Papdatou-Pastou (2023) published “’Visual Type? Not My Type’: A systematic study on the learning styles neuromyth employing frequentist and Bayesian statistics” in Mind, Brain & Education. Here’s the highly edited article:
The term learning styles (LS) describes the notion that individuals have a preferred modality of learning (i.e., vision, audition, or kinesthesis) and that matching instruction to this modality results in optimal learning. During the last decades, LS has received extensive criticism, yet they remain a virtual truism within education. One of the major strands of criticism is the fact that only a handful of studies have systematically put the LS assumptions to the test. In this study, we aimed to explore whether learners who are visual types will be better at learning sign-words (i.e., ecologically valid stimuli) compared to auditory and kinesthetic types. Ninety-nine volunteers (67 females, mean age = 28.66 years) naive to Greek Sign Language (GSL) were instructed to learn 20 GSL sign-words. The volunteers further completed two LS questionnaires (i.e., the Barsch Learning Styles Inventory and the Learning Channels Inventory) and they also reported what their LS they believed was. No evidence of a difference in learning sign-words among individuals with different LS (as identified by either of the LS questionnaires or by direct self-report) was found, neither using a frequentist nor using a Bayesian approach to data analysis. Moreover, inconsistencies between the way participants were classified based on the different measures and direct self-report were detected. These findings add further support to the criticism of the LS theory and its use in educational settings. We suggest that research and practice resources should be allocated to evidence-based approaches.
The term learning styles (LS) describes the notion that individuals have a preferred modality of learning and that presenting to-be-learned material in this modality results in optimal learning for them. To date, studies around the globe suggest that LS is very popular among preservice and in-service teachers, including school principals, student teachers, and university instructors, even among the more esteemed and internationally recognized of them. Across Europe, in the United Kingdom and the Netherlands, Spain, and Portugal, teachers' and student teachers' belief in this neuromyth was found to be the case for 90% of those surveyed or higher. In Greece in particular, where the present study takes place, these numbers are even higher; 97% of in-service teachers believe that matching LS type with mode of instruction enhances educational outcomes and 94% of student teachers agree with this view.
In line with a recent document, “there is very limited evidence for any consistent set of learning ‘styles’ that can be used reliably to identify genuine differences in the learning needs of young people” (Education Endowment Foundation [EEF], 2019, para 2). In fact, LS was pronounced as a neuromyth already since the early 2000s (see the Centre for Educational Research and Innovation [CERI], at the OECD in 2002).
Neuromyths are defined as misconceptions based on facts that are misinterpreted (OECD, 2002). Grospietsch and Mayer (2019) call these misrepresented facts “kernels of truth” and posit that neuromyths start from a valid point and through subsequent logical yet unsubstantiated steps this point is then repackaged to a new reality, which is neither accurate nor scientifically tested itself. The kernel of truth for the LS neuromyth is that people have preferred modalities—typically the visual, the auditory, or the kinesthetic (VAK) modality. Expecting, however, that fitting the task to their preference will have a positive effect on performance in where the problem lies because preference is different from ability.
Evidence on learners' differences in educational needs, which teachers should accommodate, and the importance in developing learner's metacognitive and self-regulation skills to enhance learning further add to the appeal—and need—of individualized instruction. However, anything between this truth and teachers' effort to cater instruction to match learners' preferred modalities to enhance learning is the LS myth. Of note, this practice of matching instruction to preferred modality is commonly encountered in the literature as the meshing or matching, but also the LS hypothesis or the LS notion In this study, the term “LS hypothesis” will be adopted.
In addition to having been labeled as a neuromyth by educational bodies (CERI and EEF), as described earlier, LS has received considerable criticism in the academic literature. Some of the more prominent strands of criticism refer to the following:
The number of LS categories and LS taxonomies is so big that the term has become conceptually equivocal. As an illustration, Coffield et al. (2004) identified 71 different LS models. Moreover, different educators mean different things when they refer to LS.
The lack of psychometrically sound LS measures.
The lack of experimental evidence (or more accurately evidence coming from systematic, empirical studies, as participants cannot be randomly allocated to learning styles) that each person has one single LS and that they learn best when instructed in this single modality.
The experimental evidence on the LS hypothesis is somewhat rare before 2010, with only one study published. Krätzig and Arbuthnott (2006) tested whether LS preference—as measured by direct self-report and the Barsch Learning Styles Inventory (BLSI)—correlated with memory performance for material presented in each modality (as measured by auditory, pictorial, and tactile tests) and found no evidence of such a correlation. Nearly 10 years later, Rogowsky et al. (2015) recruited a group of 121 participants and investigated the effect of LS preference, as assessed by the Building Excellence (BE) Online Learning Styles Assessment Inventory, on (a) listening and reading comprehension and (b) on learning, when the identified LS was matched with the modality of instruction.
Other experimental work conducted on LS recently involved the comparison between visual subtypes (i.e., visual and verbal learning), presenting no evidence of an effect in this case either. Knoll, Otani, Skeel, and van Horn (2016) tested the hypothesis that preferred LS correlates significantly with subjective—but not objective—aspects of learning. In particular, the authors, first, assessed participants' preferred LS as verbal or visual. Then they assessed (a) the cued recall of pairs of stimuli presented in either verbal or visual modality (objective learning aspect) and (b) the participants' immediate or delayed judgments of the likelihood (JOL) of recalling the stimuli (subjective learning aspect). As hypothesized, higher visualizer scores were associated with higher immediate JOLs for pictures, and higher verbalizer scores were associated with higher immediate JOL for words (the subjective aspect of learning). However, there was no effect of preferred LS on actual recall scores (the objective aspect of learning) or accuracy of judgments.
Similarly, Koć-Januchta, Höffler, Thoma, Prechtl, and Leutner (2017) examined whether being a visual or verbal type leads to different learning behaviors, through eye-tracking techniques, and found no differences in retention of information, regardless of the learning topic or the mental effort invested in the task. Overall, to the best of our knowledge, only the seven studies described earlier have empirically tested the basic premise of the LS hypothesis to date, namely, that optimal learning takes place if instruction is matched to the LS of the learner. None of the studies provided any evidence for this premise.
Of note, the LS hypothesis was tested with various samples of adults and school-aged participants; with different outcome variables including comprehension (listening and reading), recall, and memory performance; with a wide variety of approaches and/or measures employed to classify participants across the different LS types: the BLSI, the BE Learning Styles Inventory, the LSCY, the LSS, the VAK Learning Styles Inventory, modified versions of the O′ Brien Learning Styles Questionnaire, and direct self-report; and finally with different materials (e.g., material on lightning or from a nonfiction historical book). As evident, only in one previous study, direct self-report was used in addition to an LS questionnaire measure to control for the possibility that the specific type of assessment was not appropriate. Moreover, all to-be-learned material came from either educational resources or everyday material; therefore, participants might have encountered it before or it might have been already known to them.
In the present study, we aimed to systematically test the LS hypothesis, focusing on the visual modality. We took an innovative approach with regards to the to-be-learned material by using sign-words for the first time in the literature. Sign-words are unique, rich, and unfamiliar in nature as visual stimuli. They are moreover ecologically valid, as individuals with a hearing impairment need to (and do) learn sign language in order to be able to communicate. Sign-words further neutralize any potential effect of familiarity that visual stimuli typically used in LS studies (e.g., pictures of objects or animals or other educational or everyday material) might have on performance. Lastly, to ensure increased reliability of the LS assessment, we chose to use three separate LS measures (two LS questionnaires and direct self-report).
The average quiz scores of the visual information learning test and the standard deviations based on the BLSI were close among the three LS types, with visual types being outscored by both auditory and kinesthetic types. The same pattern can be identified when looking at the quiz scores of the visual learning test of the three LS types based on the LCI. However, based on direct self-report, the visual types have slightly outscored the auditory types and considerably outscored those who did not identify as any visual/auditory/kinesthetic in the visual learning test.
As Willingham et al. (2015) suggest, to provide evidence for the LS hypothesis, one should be able to show that (a) there is a positive effect of the modality of teaching on achievement (main effect) and that (b) matching the LS of the participants “optimizes achievement for each group” (p. 267). Focusing specifically on “a,” we hypothesized that a difference between the three LS for learning material presented visually (i.e., sign-words) should be found if the LS hypothesis is true. In fact, “visual” learners would be expected to learn visual material better than “auditory” or “kinesthetic” types (or even participants who did not self-identify with any LS) if the LS hypothesis is true. Using a frequentist statistical approach, we found an absence of evidence that there are differences between the three LS in terms of learning visual material. When supplementing the frequentist analysis with a Bayesian statistical approach, we could further establish evidence of an absence of a difference between the three LS for learning visually presented material.
With regards to the secondary aim of this study, namely, to explore the extent to which the three VAK measures assign participants to visual, auditory, and kinesthetic LS consistently, results were mixed. On the one hand, the classification of participants based on the BLSI and the LCI, although not identical, was found to be consistent. On the other hand, the classification based on direct self-report was not consistent with the evidence from the LS measures employed (this was not tested via a statistical criterion, however), as none of the participants selected the “kinesthetic” type when using direct self-report. This finding contradicts previous findings with primary school students, where over 60% of the students had self-identified as “kinesthetic.”This finding further contradicts the classifications made by means of the BLSI and LCI used in this study (whereby a number of participants were classified as “kinesthetic”).
Based on this evidence, it can be argued that (a) the self-reported LS and the LS preference as measured by questionnaires do not coincide, or that (b) the self-reported percentages of “kinesthetic” types differ between students and adults. In addition, and based on the reported low reliability for the two measures (BLSI and CLI) in this study, arguably there is much room for research and improvement in the area of LS measurement. This study follows many others that cast doubt in the measurement of LS and/or the existence of distinct LS. With regards to the issues of measurement, the discrepancies in the choices made by participants with regards to their preferred modality (and the poor reliabilities found) point to either the lack of rigorous work in the inception and development of the measures and/or the lack of consensus among theorists about the operational definitions of the construct. In every case, the results are (as in our case) contradicting and doubtful. Focusing on the issue of the existence of LS, arguably, a construct that is not psychologically real cannot be measured accurately or reliably.
The findings from this work add further support to the criticism of the LS hypothesis and its use in educational settings. This link is problematic at the neurological level too, for a modality to be dominant, one region of the brain (or a network) should be better (more efficient/productive or quicker) in processing information than others; but there is no such evidence. More importantly, information processing activates brain regions engaged in a task, and when the task is comprised of visual, auditory, and kinetic components, then networks of corresponding regions are activated together, share information, and even compensate for the loss of information in other components. Accordingly, and based on other available evidence, it seems that multimodality may be a more appropriate approach to effective teaching and learning.
In line with the evidence, it is argued that adopting the LS theory does not merely waste time, money, and energy resources, but is potentially harmful to learners, as it promotes unrealistic expectations. In essence, it promises that everyone can learn, as long as instruction modality meets their LS. Moreover, this belief can result in students shying away from learning material that is not presented in their alleged LS, instead of them taking on new challenges and working toward their weaknesses. Furthermore, the adoption of LS by educators reveals a lack of critical and research skills. Teachers, researchers, and policymakers could use their typically limited resources in more meaningful and evidence-informed methods. We, therefore, join voices with previous researchers who have called for the debunking of the myth of LS—among other myths—in teacher education settings, as well as for educating teachers in the use of evidence-based practices, such as the self-regulation of learning.
Overall, this study sought to test the LS hypothesis through a new approach (three different LS measures, novel and ecologically valid visual stimuli); it failed to provide support for it. Regardless of the LS measures used, visual types were not found to be better at learning material presented through the visual modality, namely sign-words. Catering to students' needs lies at the heart of good educational practice. However, doing so using theories not supported by empirical evidence is suboptimal both in terms of resource allocation and meeting students' needs. As Willingham et al. (2015) put it, “it is impossible to prove that something does not exist” (p. 267), but teaching LS should be evidence-based, and this was another systematic research work that has failed to provide such empirical evidence.
I think this is an important study with a clever methodology. It also highlights the problem that many educators assume that learning styles exist, should be catered to, and impact learning, when their major function may be to engender learned helplessness about non-preferred styles.