Do you ascribe to the idea that people have different learning styles, such that learning is supported by the optimal match of instruction with style but hindered by a mismatch? In other words, to use a common learning-style discriminator, optimal learning requires instructing identified verbal learners with a verbal sensory experience while implementing a visually oriented instructional design for the visual learners. If you do hold this view, then you must understand that you are swimming upstream against a torrential current of contrary evidence.
Lack of evidence or meaningful theory
Criticism of learning styles as indicators for teaching have existed as long as the many popular instruments that supposedly measure these differences. A recent decade of loud objection to learning styles began with an important 2008 publication by a team of cognitive psychologists led by Harold Pashler. Their article was commissioned by Psychology in the Public Interest to assess the scientific evidence for applying learning-styles assessments in school contexts. The authors addressed this task by reviewing the literature to test the “meshing hypothesis”. Consider students A and B who are assessed as having opposite styles (e.g., verbal versus visual learners). According to the meshing hypothesis, student A will outperform student B when instruction matches A’s style but the reverse will be true if the instruction of the same material matches B’s style. Despite the importance of the meshing hypothesis to decades-long advocacy for teaching to individual learning styles, Pashler and his colleagues found very few studies with research methodologies appropriate for testing the hypothesis. Furthermore, the relevant existing studies failed to support the hypothesis. The Pashler et al. article received great attention in professional publications and the popular media – I remember reading about it in a syndicated education column in my local newspaper. The results were championed by skeptics of learning styles in not only educational psychology contexts but also in disciplines ranging from culinary training to medical school. The assertion of learning styles as myth is now highly visible.
Gazing beyond the problematic meshing hypothesis, it is also essential to scrutinize that most learning-styles advocates pay little attention to the validity and reliability of the instruments that they use to measure their learners’ styles. In an exhaustive 2004 review of learning-styles instruments, Frank Coffield and colleagues analyzed 13 instruments that are widely applied in higher education. The instruments subjected to psychometric analysis scored weakly on reliability, construct validity, or both. These include the family of surveys that focus on verbal and visual (and commonly kinesthetic) learning styles, the very popular Kolb Learning Styles Inventory and the venerable Myers-Briggs Type Indicator, which although focused on personality type has been extended into the learning realm. Even in the business management world where Myers-Briggs consultants have earned their keep, the measure has fallen out of favor.
There is also the limitation of learning style self-assessment, which is not an objective measurement of learning. For instance, one study showed no relationship between stated preferences for visual or verbal learning and the measurement of learning that emphasized one modality over the other. Therefore, is there any justification for adjusting instruction to match a learner’s style, or to be inclusive of an array of learners’ styles, if there is so little confidence in the assessment of these styles? Ironically, one could also argue that the conclusions of the Pashler group are weakened by testing a hypothesis with data collected with instruments of questionable validity and reliability. Nonetheless, at best, learning-styles advocates can only say that the meshing hypothesis remains untested while also acknowledging that they have little more than anecdote to continue using practices for which there is no evidence.
Despite the compelling evidence-based argument against learning styles, their popularity and acceptance remains strong. Nearly all of 109 articles published between 2013 and 2015 regarding learning styles and higher-education instruction expressed positive intentions regarding learning styles and 89% reached positive-outcomes conclusions regarding use of learning styles; even though only one of the studies included a test of the fundamental meshing hypothesis. Recent surveys show that a large majority of teachers at all levels – school to university – continue to accept the validity of accommodating learning styles into their instruction.
The cognitive and educational psychologists who labored to demonstrate many flaws of popular perceptions of learning styles are frustrated by the persistent advocacy and use of purported assessment of learning styles in education. Thirty European and North American psychologists and neuroscientists signed a letter to the influential British daily, The Guardian, during the 2017 Brain Awareness Week. They not only argued the absence of evidence for learning styles but also that there was potential educational and fiscal harm to continue using the idea in education. A commentary by prominent educational psychologist Paul Kirschner was triggered when the journal’s editors invited him to enlighten their readership after Kirschner posted a tweet berating the journal for publishing another paper about “learning styles bull!”
Notably, and curiously, missing from recent critical scrutiny of learning-styles instruments is mention of the Felder-Silverman Index of Learning Styles (ILS). Developed by engineering education expert Richard Felder, the ILS is the second most commonly used learning-styles assessment tool in higher education. The ILS differs from surveys that have been strongly attacked. The ILS has been the subject of psychometric analyses that add confidence in the reliability and validity of what is being measured (although we should still debate what “it” really is). Unlike most surveys that pigeonhole learners into particular categories, the ILS reports varying levels of preference for modalities of information intake and approaches to knowledge processing. Even more importantly, Felder never advocated for matching learners’ learning style with particular instruction. Instead, he emphasized placing learning preferences in context with the learning task, and stretching learners to use less desirable learning approaches in order to develop broader learning skills. He encouraged faculty to use the ILS scores to employ a broad battery of approaches to be inclusive of learning styles. In response to the Pashler study, Felder vigorously defended the ILS and pointed to several studies where the meshing hypothesis was supported by ILS data.
Nonetheless, it is difficult to understand what conceptual or theoretical framework encapsulates the various learning-styles approaches, including the ILS. The typical learning-styles survey classifies a respondent on several dichotomous scales or categorical end points. While some, such as the popular visual-verbal distinction, appear in many survey outcomes, most instruments have distinct ways of categorizing a learner. Examples include: intuitive versus analytic, arousal avoidance versus arousal seeking, simultaneous or successive planning of learning, concrete versus abstract, sequential versus global, accommodating or assimilating, convergent or divergent … and the list goes on. These many ways of labeling a learner must raise the question of what “learning style” really means, whether these labels can really be confidently measured, and if they need to be scrutinized in different contexts. Felder defends this state of affairs by contending that the number of learning styles are unlimited and cannot practically be encompassed in a single theory. However, the theory that we seek is a rationalization for why learning styles (a) should exist, and (b) should impact learning. That framework remains largely lacking and, as a result, some learning styles scales may miss the point.
For instance, the visual-verbal scale is featured in many learning-styles instruments, including Felder’s ILS. Regardless of subjective statements of preference, the human brain depends on a dual coding system that utilizes both phonological and visual inputs. Therefore, purposeful emphasis on one sensory modality over the other based on a subjective measure of a preference runs the risk of impeding learning processes. The alternative view is that modality of information input during learning depends on the learning task. Therefore, the modality does matter, but it matters the same way for everyone who pursues the same learning activity.
The concrete-abstract distinction is also problematic. This distinction originates with Kolb’s LSI and loads heavily on Felder’s sensing-intuitive scale. However, rather than being a learning style, per se, operational learning from concrete or abstract approaches has long been regarded – since the work of Jean Piaget in the early 1900s – as a developmental process. Learners progress from concrete to abstract thinking over time. For example, in reviewing my ILS data collected from about 400 college students and 300 faculty members, there is a strong skew among students to be sensors (concrete learning) whereas the more mature learners composing faculty are strongly skewed toward intuition (abstract learning). We should expect learners in different development stages and promote learning from both concrete example and abstract generalization. However, these are not different learning styles.
But – humans don’t all learn the same way, either
I could end, here, and this essay would simply be another effort to wake up educators to a fallacy of adjusting instruction to account for alleged learning-styles differences among their students. However, I feel that several issues require further scrutiny.
In addition to the Pashler et al. paper, I always encourage those interested in learning styles to read the 2014 paper by Maria Kozhevnikov and colleagues, which places learning styles into the broader and older concept of cognitive styles. Although difficult to measure, cognitive style refers to individual approaches to cognitive functioning, including acquiring information, processing information, and decision making. Kozhevnikov and co-authors argue that the falsification of the meshing hypothesis does not preclude the existence of cognitive style. Even Pashler, Kirschner, and others acknowledge that people have learning preferences, which are less strongly perceived than alleged styles, but these are dismissed as being of minimal importance for educational practice. However, these authors cite no evidence to justify “minimal importance”. Needing further attention is the affective consequence of encountering instruction that is contrary to preferences. Just because the cognitive function of the human brain accommodates wide ranging instructional approaches does not mean that students will engage equally and self-efficaciously with different approaches if there is a mismatch with a cognitive-style preference.
I have had many conversations with scholars who approach learning differences and styles from a critical-theory perspective. Their work is typically rooted in anthropological and sociological frameworks, rather than psychology, and they rarely or never cite the psychology research mentioned previously. They usually avoid discussion of quantitative learning-styles surveys but commonly encourage application of Howard Gardner’s multiple intelligences concept, despite the comparable demolition of that idea by cognitive psychologists. They view cognitive psychologists as seeking to generalize human behavior and brain function, which means brushing over the individual or group differences that are so meaningful to anthropologists and sociologists and scholars seeking social justice for those not flourishing in higher education constructs founded in western European traditions. Favoring these opinions, the Kozhevnikov review summarizes neuroimaging studies that demonstrate differences in brain activity for Asians versus Americans during learning experiences, as well as other research that justifies thinking that there is a neurological basis for cultural differences in cognitive style.
A related idea arising from the anthropological perspective is epistemological differences of learning focused in individual actions versus the collected action of a group. These differences are referred to as high- versus low-context learning, individuated versus integrated epistemologies, or independent versus interdependent approaches. Collectively, the research argues that these contrasting views of where knowledge comes from along with how, and with whom, it is constructed varies by cultural ethnicity and race, country of origin, and socioeconomic status. Northwestern University’s Nicole Stephens argues that most colleges and universities favor independent learners to the exclusion of interdependent approaches more common to first-generation college students from working-class families. As a result, higher education enlarges, rather than shrinks, socioeconomic inequality in the United States. Here, again, data from neurological scans reveal differences in cognitive processing by learners of differing social class that correlate to independent versus interdependent self-construal. These sociologically grounded concepts, supported by psychology and neuroscience research and connected to the idea of cognitive style, are arguably more important than debating the meshing hypothesis when we seek to understand how best to teach diverse learners.
Other authors investigated the impact of different instructional approaches on the achievement gaps associated with students of varying race/ethnicity, socioeconomic status, or both. The surging interest to replace the centuries-old tradition of the university lecture with active-learning pedagogies has been shown to decrease these achievement gaps. My critical-theory colleagues claim that this is evidence of different learning styles that are rooted in culture, family, and community structures during childhood learning. In this view, the traditional university curriculum favors the individuated, independent learning approaches of those from groups who have been privileged for success in higher education whereas active learning favors the integrated, collectivist culture of learning in working-class contexts and among African American, African, Hispanic, Indigenous, and Southeast Asian communities. However, I am also impressed that these data demonstrating quantitative improvements in achievement gaps show that there was no decrease in achievement of the traditionally high-performing demographic groups. Instead, all boats floated higher on the tide of active learning and the gap closed because of preferentially greater learning improvement for members of those populations traditionally less successful in higher education. To me, this suggests an alternative hypothesis: Active learning is more natural to neurological function that evolved for hundreds of millennia before the first college classroom and is more closely associated with the psychological needs of autonomy, competence, and relatedness that should motivate learning according to self-determination theory. Therefore, rather than active learning being favored in high-context/integrated/interdependent epistemological contexts, it may be that the low-context/individuated/independent students arose from more privileged experiences to prepare for and succeed in college and have better adapted to system of learning that is, frankly, unnatural, but who do flourish when engaging instead with active learning.
Time for a reboot
So, are learning styles an illusion? In our desire to honor perceived differences among learners, have we gone too far? All educators should accept that the meshing hypothesis stands as unsupported and can only be possibly, if unlikely, resurrected by focused study, rather than inference, using valid and reliable instruments that have a sound theoretical basis, and which do not currently exist. In the meantime, I feel that the term “learning style” is too burdened by a history of using a confusing array of measurements of dubious quality and then linked to the unsubstantiated meshing hypothesis. A possible way forward is to abandon the learning-style concept, and stop using the phrase, in favor of exploring new approaches to understanding individual and group differences that engage cognitive psychologists and neuroscientists alongside social psychologists, anthropologists, and sociologists. Nonetheless, the ongoing learning styles debate is a battlefield engaged by opposing protagonists who have strong biases toward their point of view and are challenged to understand the other side. As with political partisanship, it is unclear that confirmation biases will permit those on opposite sides of the debate to seek common ground for the necessary reboot to understand the educational significance of learner differences.
By Gary A. Smith, Assistant Dean of Faculty Development in Education and Director, Office for Medical Educator Development, School of Medicine, University of New Mexico. This work is licensed under a Creative Commons Attribution-NonCommercial-Share Alike 4.0 International License.