How do Learning Styles Affect Learning Predictions?

By Megan Sumeracki

In spite of many cognitive psychologists’ efforts to dispel the myth, the learning styles theory is still popular in many educational circles. Learning styles is the idea that individual students have learning preferences, and importantly, in order to maximize learning, students need instruction in their preferred style. The implications of this theory, if there was evidence to support it, are that teachers would need to assess each student’s learning style and then tailor their instruction to match that student’s style. Of course, if you read our blog, you know that learning styles do not have evidence to support its effectiveness. In fact, catering to learning styles isn’t just ineffective! Learning styles can also be harmful. Note, learning styles is different dual coding, and you can read about learning styles vs. dual coding here.

In today’s blog, I walk through yet another study (1) demonstrating that catering to learning styles do not improve learning, and can be harmful. In this blog, the focus is how learning styles affect learning predictions, or metacognition.

The Experimental Design and Procedure

In Knoll and colleagues’ 2017 (1) experiment, University students learned simple materials that were visual and verbal. Simple materials mean this experiment is in the “basic research” level of the lab-to-classroom model, allowing for greater control and determining cause-and-effect relationships.

Assessment Phase:

First, the students were assessed to determine whether they had a verbal or visual learning style. They also completed tasks that tested their vocabulary and visual reasoning ability.

Learning and Predictions Phase:

The students studied two lists: pairs of line drawing pictures and pairs of words. Some of the participants learned the picture list first, while others learned the word list first to account for any effects of order (called counterbalancing, and you can learn about this in the research methods post if you are interested).

Sample of images from the Snodgrass & Vanderwart (1980)*(2)* picture norms. Knoll et al. (2017) created the picture pairs from the materials in this paper.

Students learned one list at a time. While the students were learning the lists, they were asked to make a judgment of learning (JOL). A JOL is just a prediction about how well they think they have learned the material. In this case, they were asked to rate the likelihood that they would remember the pair in the testing phase later. Sometimes they made the JOL right after seeing a pair, and other times it came after a delay of a couple of minutes (at least intervening 10 items).

Testing Phase:

After learning the first list, the students performed an unrelated “distractor task” for 2 minutes, and then took a cued-recall test to measure their learning of the items. The students saw a cue word or picture one at a time, and were asked to recall the other word or picture from the pair they originally studied.

Repeat for the Second List:

Finally, the learning and predictions phase and the testing phase were repeated for the second list. (I.e., the students who learned the pairs of words first now learned the pairs of pictures; the students who learned the pairs of pictures first now learned the pairs of words.)

The Results

The researchers found that learning style was not related to actual learning. In other words, the verbal students did not remember the words any better than the pictures, and the visual students did not remember the pictures any better than the words.

However, their predictions about how well they would remember the pictures and words did depend on their learning style.

The verbal students had higher JOLs for the words than pictures, indicating they predicted they would remember the words better than the pictures (even though they didn’t).
The visual students had higher JOLs for the pictures than words, indicating they predicted they would remember the pictures better than the words (even though they didn’t).

The predictions were only related to learning style when the JOLs were made right after studying the pair. This means that after waiting a little bit, the students were better able to assess how well they would remember the item. In general, delaying one’s prediction of how much is learned (i.e., metacognitive monitoring) leads to a more accurate prediction.

Why is it a problem that learning styles are related to learning predictions but not actual learning?

If a student is studying, and they believe in learning styles, they may study the information in their own particular style. While they are studying, they are likely assessing how well they think they know the material, likely in preparation for a test. If they think they learn best with verbal information, and they are studying verbal information, this pairing alone should drive up their prediction of how well they are learning without actually meaning they are learning more. (And, students tend to overestimate how much they will remember in general as it is.) Their prediction that they know the material better may lead them to terminate studying early, potentially leading to a disappointing surprise when it comes time for exam day. If the students were told that learning styles don’t matter for learning, and encouraged to use dual coding (and delayed predictions of their own learning for that matter), their assessments would generally be more accurate.

Bottom Line

Catering to learning styles isn’t just ineffective at producing learning. It can also be harmful. While this experiment utilized very simple experimental materials for learning, that does not necessarily mean it doesn’t have any validity for the classroom. This type of experiment can allow for greater control over the materials and experimental procedure, making it easier to determine cause-and-effect relationships (see this post).

Instead, we should engage students in multimodal learning, or dual coding, and match learning activities to the material to be learned. Imagine trying to learn the anatomy of this flower and how these parts work together without a diagram. It would be very difficult for many people to learn just based on a verbal description. On the other hand, if we remove all of the words from this diagram, the chances of learning would likely be greatly reduced.

This does not mean that every single learner is identical and that there is zero diversity in learning. Clearly, students do have learning preferences! We can capture their preferences by providing students with variety rather than putting them in a learning styles box and only letting them learn in their one specific style. Teaching them to only learn in their style limits the learning strategies they can use, and increases how much they think they are learning, without actually increasing their learning.

Want to read more on this topic? Click the link to see our posts tagged “learning styles”.

References:

(1) Knoll, A. R., Otani, H., Skeel, R. L., & Van Horn, K. R. (2017). Learning style, judgments of learning, and learning of verbal and visual information. British Journal of Psychology, 108(3), 544-563. https://doi.org/10.1111/bjop.12214

(2) Snodgrass, J. G., & Vanderwart, M. (1980). A standardized set of 260 pictures: Norms for name agreement, image agreement, familiarity, and visual complexity. Journal of Experimental Psychology: Human Learning and Memory, 6(2), 174-215. https://doi.org/10.1037/0278-7393.6.2.174

May 12 How do Learning Styles Affect Learning Predictions?