Learning styles: types, evidence and how to learn best with AI

Learning styles are the formats a person prefers to take in new information: images, spoken explanation, written text or practice. The best-known learning style model, VARK, sorts these preferences into four types: visual, auditory, reading/writing and kinesthetic. Most people do have preferences, but decades of controlled research have found no reliable benefit from matching teaching to a learner’s preferred style. Below: the main types, what the research says, what works instead and how AI tutoring helps teachers and trainers.

What you’ll take from this

• The best-known model describes four learning styles: visual, auditory, reading/writing and kinesthetic. Most learners use a mix.
• The “meshing hypothesis”, the idea that teaching to a learner’s style improves results, is not supported by the research reviewed by Pashler and colleagues (2008).
• Practice testing, spaced practice and pairing words with visuals improve learning for almost every learner, whatever their stated style.
• AI tutors can personalize pace, examples and feedback. Early trials show strong gains when the tutor is designed to make learners think, and losses when it simply hands out answers.
• You’ll find copy/paste prompts that help teachers and trainers use AI for evidence-based lesson design.

This guide is written for teachers, trainers and anyone who designs learning. If you want to apply AI to your own teaching, the Founderz IA para docentes program is built for educators. The program is currently taught in Spanish; if you need another option, contact Founderz.

What is a learning style?

A learning style is a preference for the way you take in new information, process information and retain the information later. One person likes a chart more than a paragraph. Another remembers a spoken explanation better than a written one. That preference is your style of learning. Psychologists and education experts usually call it a learning preference, because it describes what you like rather than how your ability to learn works.

The idea is popular because it matches the way we learn day to day. Learners tend to enjoy a lesson more when the format suits them, and enjoyment matters for motivation. Being aware of your own strengths, weaknesses and preferences is a healthy part of personal learning. The problem starts when a preference turns into a rule, such as “I’m a visual learner, so I can’t learn from lectures.” The research covered below shows why that rule does not hold.

Learning styles, cognitive styles and learning strategies

These three terms are often mixed up:

• Learning style is a perceptual or format preference, such as favoring visuals over audio.
• Cognitive styles describe how you organize information at a deeper level, for example detail first or big picture first.
• Learning strategies and learning techniques are the actions you take, such as self-testing or spacing out study sessions.

Your approach to learning drives results far more than your style. A learner who quizzes themselves regularly will usually outperform one who rereads notes in their “preferred” format.

The 4 types of learning styles in the VARK model

Neil Fleming introduced VARK in 1987 to help teachers and learners talk about how learners prefer to learn. It builds on the older VAK model by splitting reading/writing from purely visual information. It remains the most widely cited of the different learning styles frameworks because it is simple to apply.

Treat the categories as a vocabulary, not a diagnosis. Most people lean on two or three channels depending on the subject, and the same person can prefer text for law and practice for cooking. Each type of learning preference below comes with what the evidence supports for everyone.

Visual learning

A visual learner likes images, spatial layout and color more than blocks of text. Visual learners often remember where information sat on a page and follow a flowchart faster than a written description. Visual aids such as diagrams and charts, timelines and mind maps feel natural to them.

What the evidence supports for everyone: creating visual summaries that pair words with a relevant image. Allan Paivio’s dual coding theory holds that information stored both verbally and visually is easier to recall. That benefit applies to all learners, not only those who call themselves visual.

Auditory learning

An auditory learner likes spoken explanation, discussion and sound. These learners often understand concepts by using conversation, and podcasts, lectures and group discussion suit this preference.

What helps everyone: explaining an idea out loud in your own words. Explaining forces you to retrieve and organize what you know, which exposes gaps faster than rereading.

Reading/writing learning

A reading/writing learner takes in information best through text. These learners like written definitions, structured outlines and annotated notes. Traditional assessment, such as essays, reports and written exams, rewards this channel.

What helps everyone: active writing, not passive reading. Summarizing from memory and answering your own questions move material into long-term memory far better than highlighting or rereading.

Kinesthetic learning

A kinesthetic learner likes movement, touch and doing. If you learn best by doing, you probably retain more from a simulation than from reading about the same task. Role-play, labs and case exercises suit this style.

What helps everyone: learning tasks that require action and decisions. A trainee who works through realistic scenarios has to apply the rule, and applying it is a form of retrieval practice. Learners retain procedures better when they perform them.

Other learning style models: 7 types of learners, Kolb, Honey and Mumford

VARK is one of many models and theories. Researchers have proposed dozens of them, and they disagree on what a “style” even measures or which aspects of learning it covers. These are the ones you are most likely to meet.

The 7 types of learners

A popular extension adds three categories to reach seven types of learners, each a different type of learner on paper: logical (reasoning and patterns), social (learning in groups) and solitary (learning alone). The extra categories describe motivation and setting more than senses. They carry the same evidence problem as VARK: no strong research shows that sorting learners this way improves outcomes.

Kolb’s experiential learning cycle

David Kolb’s experiential learning theory, published in 1984, describes learning as a cycle with four stages: concrete experience, reflective observation, abstract conceptualization and active experimentation. Kolb also proposed four styles (diverging, assimilating, converging and accommodating) based on where a person feels most comfortable in the cycle.

The cycle itself is useful for lesson design because it moves every learner through doing, reflective learning, theorizing and testing. The style labels attached to it have the same weak support as VARK.

Honey and Mumford’s learning styles

Peter Honey and Alan Mumford adapted Kolb’s model for workplace training. Their questionnaire sorts people into activists, reflectors, theorists and pragmatists. Many corporate training teams still use it as a conversation starter about how a team approaches new tasks and learning activities.

Gregorc, Felder-Silverman and other models

Anthony Gregorc’s model describes concrete or abstract perception combined with sequential or random ordering. The Felder-Silverman Index of Learning Styles places people on scales, such as active to reflective, and is common in engineering education. The National Association of Secondary School Principals also published a learning style profile for schools. Each model offers a different lens on an individual’s learning, and none has strong evidence that matching instruction to one particular learning style consistently improves results.

Are learning styles real? What the research says

Learning preferences are real. The claim that teaching to a person’s preferred learning format improves learning outcomes is not supported. This distinction is the heart of the debate about using learning styles in schools and training.

The key test is called the meshing hypothesis: if styles matter, visual learners taught visually should outperform visual learners taught verbally, and the reverse should hold for verbal learners. In 2008, Harold Pashler, Mark McDaniel, Doug Rohrer and Robert Bjork reviewed the literature for Psychological Science in the Public Interest. They found very few studies with a design able to test the hypothesis, and the ones that did largely contradicted it. Their conclusion was that there is no adequate evidence base to justify building learning styles assessments into general educational practice.

The UK’s Education Endowment Foundation reaches a similar view in its Teaching and Learning Toolkit. It rates the security of evidence on learning styles as extremely low, warns that learners are very unlikely to have a single learning style, and notes that labelling learners can undermine their belief that effort leads to success.

The belief is still widespread. In a study of 242 teachers in the UK and the Netherlands, published in Frontiers in Psychology in 2012, Sanne Dekker and colleagues found that more than 80% agreed that individuals learn better when they receive information in their preferred learning style. Matching learners’ learning styles to teaching styles remains a common practice, even though the research does not back it.

Three practical conclusions follow:

• A learning styles questionnaire can start a useful conversation about preferences and motivation. It should not decide how someone is taught.
• Match the format to the content, not to the person. Geography needs maps, music needs sound and surgery needs physical practice, for every learner.
• Variety helps. Presenting each key learning point in more than one format gives everyone more ways to connect with it, and a variety of learning formats keeps attention up.

What works instead: evidence-based learning strategies

The strongest evidence for effective learning points to a handful of methods of learning that work for almost everyone. In 2013, John Dunlosky and colleagues reviewed ten common study techniques, also in Psychological Science in the Public Interest. Two received their highest utility rating: practice testing and distributed practice. Rereading, highlighting and summarization were rated low utility.

• Practice testing (retrieval practice). Pulling information out of memory, through quizzes, flashcards or explaining from memory, strengthens it more than reviewing it.
• Distributed practice (spacing). Spreading study across several sessions beats cramming the same hours into one.
• Dual coding. Pairing words with a relevant image helps learners build two routes back to the same idea.
• Worked examples and feedback. Showing a solved problem step by step, then giving fast feedback on the learner’s attempt, reduces the load on beginners.

These strategies work across different types of learners and should shape your teaching methods and learning materials, whatever the learning environment. They also translate well to online courses: our guide to e-learning content development shows how to build them into digital training.

How AI changes personalized learning

The appeal of learning styles was always personalization: teach each person in the way that suits them. AI makes real personalization cheaper, but the variables that improve educational results are prior knowledge, pace, examples and feedback. Sensory style is not one of them, and a specific learning label adds little.

Two recent studies show both sides. At Harvard, Greg Kestin and colleagues ran a randomized crossover trial in an introductory physics course, published in Scientific Reports in 2025. Learners using an AI tutor at home achieved median learning gains more than double those of the same learners in an in-class active learning session, in less time. The authors stress that their tutor was carefully designed: it walked learners through problems step by step, used expert-written solutions and followed established teaching practice. They do not claim the result holds in every context.

The second study is a warning. Hamsa Bastani and colleagues, in a 2025 paper in the Proceedings of the National Academy of Sciences, gave nearly a thousand high school math learners access to GPT-4. With an unrestricted chatbot, practice grades rose 48%, but scores on a later exam without AI fell 17% compared with learners who never had access. A version designed as a tutor, which gave hints instead of answers, largely removed that harm.

The lesson for teachers and trainers: AI personalization helps when it makes learners think, retrieve and practice. It hurts when it does the thinking for them. That lines up with the learning strategies above far more than with any learning style label.

Limits, privacy and human judgment

AI tools still make factual mistakes, so check any generated quiz or explanation before learners see it. Avoid pasting personal data about learners into public AI tools, and follow your school’s or employer’s data policy. AI can reformat material and generate practice, but deciding what learners need next stays with you.

Copy/paste AI prompts for teachers and trainers

These prompts work in ChatGPT, Microsoft Copilot, Gemini or Claude. Each one includes a filled-in example; replace the text in [brackets] with your own context.

1. Turn a lesson into retrieval practice

You are an experienced [secondary school biology] teacher. Using the lesson notes below, write [8] retrieval practice questions for [15-year-old] learners.
Mix [4] short-answer questions, [2] multiple-choice questions with plausible wrong answers, and [2] "explain in your own words" questions.
Put the answer key at the end, with one sentence explaining each answer.
Do not invent facts that are not in the notes. If the notes are missing something important, list it separately.
Lesson notes: [paste notes on photosynthesis here]

2. Build a spaced practice plan

I teach a [6-week] [Excel for finance teams] course with [2] sessions per week. The core topics are: [lookups, pivot tables, data validation, charts].
Create a spaced review schedule that brings back each topic at increasing intervals.
For each session, list: the new topic, which earlier topics to review, and one 5-minute review activity.
Format the result as a table.

3. Dual-code a hard concept

Explain [how compound interest works] for [adult learners with no finance background].
Give me: (1) a plain-language explanation in under 120 words, (2) a description of a simple diagram I could draw on a whiteboard that shows the same idea, and (3) one worked example with real numbers.
Keep the explanation and the diagram consistent with each other.

4. Create a tutor that gives hints, not answers

Act as a tutor for a [first-year university statistics] learner working on [this problem: paste problem].
Do not give the final answer. Ask one guiding question at a time and wait for my reply.
If my answer is wrong, point to the step where the mistake is and give a hint.
After I solve it, ask me one similar question to check that I can do it on my own.

5. Check a lesson plan against the evidence

Review the lesson plan below for a [Year 9 history] class.
Tell me: (1) where learners actively retrieve information, (2) where earlier content is revisited, (3) where words are paired with visuals, and (4) any activity that relies on sorting learners by learning style.
Suggest specific changes for anything missing. Keep it to one page.
Lesson plan: [paste plan here]

A note on the last prompt: if your plan groups learners as “visual” or “auditory” learners, swap that for format variety for the whole group. Everyone benefits from the diagram, the discussion and the practical task.

How to find your preferred learning style and use it well

Knowing your preferences is still useful, as long as you treat them as preferences and not limits. A simple approach:

1. Reflect. Think of subjects that came easily. How did you study them, and in what format did you enjoy learning?
2. Test formats. Take the same material as a chart, an audio summary and a practical task, then quiz yourself a few days later. Which one did you actually remember?
3. Add the strategies that work. Whatever your preferred style, build in self-testing and spaced review.

AI makes step two quick. Paste a source into ChatGPT or Copilot and ask for a summary table, a short spoken script and five practice scenarios. You will often find that the format you enjoy most is not the one you remember best.

Frequently asked questions

What are the 4 types of learning styles?

The four types in the VARK model are visual, auditory, reading/writing and kinesthetic. Visual learners like images, auditory learners like spoken explanation, reading/writing learners like text and kinesthetic learners like learning by doing. Most people use a combination of learning styles rather than one.

What are the 7 types of learners?

The seven types add logical, social and solitary learners to the four styles above. Like VARK, this framework describes preferences. There is no strong evidence that teaching to these categories improves results.

Are learning styles real or a myth?

Preferences are real, but the idea that matching teaching to your learning style improves learning is a myth. Reviews such as Pashler and colleagues (2008) found no adequate evidence for it. Varied formats and strategies like practice testing help far more.

What is the VARK learning style model?

VARK is a model created by Neil Fleming in 1987. It groups preferences into four channels by the first letter of each: visual, aural, read/write and kinesthetic. It works well as a vocabulary for preferences, not as a guide to how someone should be taught.

Can you have more than one learning style?

Yes. Most people prefer different formats for different subjects and situations. That flexibility is one reason researchers question fixed labels.

How many learning styles are there?

There is no agreed number. A 2004 review led by Frank Coffield for the UK’s Learning and Skills Research Centre identified 71 learning style models and examined 13 of the most influential in detail. The most cited models use four styles (VARK) or seven types of learners.

What are the 8 main learning styles?

There is no standard list of eight. Lists of eight usually take the seven types of learners (visual, auditory, reading/writing, kinesthetic, logical, social and solitary) and add one more category, which varies from source to source.

What are Gregorc’s learning styles?

Anthony Gregorc described four styles built from two dimensions: concrete sequential, abstract sequential, abstract random and concrete random. Like other style models, it describes preferences rather than proven differences in how people learn.

Why is it important to understand my learning style?

Knowing your preferences helps you notice which formats you enjoy and which ones you avoid. Use that awareness to stay motivated, then add practice testing and spaced review, which improve results whatever your preference.

Should teachers use only one learning style?

No. Match the format to the content and give everyone more than one way in: a visual, an explanation and a task to do. Sorting a class into style groups is not supported by the evidence.

How can I identify my child’s learning style?

Watch what they choose when they have a free choice: pictures, talking, reading or building. Then check what they actually remember a few days later. Keep offering every format, because children benefit from all of them.

How can AI help teachers personalize learning?

AI can adjust pace, examples, difficulty and feedback for each learner, and generate practice questions in minutes. Studies suggest the benefit depends on design: tutors that give hints and ask questions help, while tools that hand over answers can reduce learning.

Teach for how people actually learn

Learning styles give teachers and learners a shared language for how a person absorbs and retains information. Results come from retrieval, spacing, clear visuals and feedback, and AI is making all four easier to deliver at scale.

Using AI well in a classroom or training room is a skill in itself: knowing what to ask, what to check and what to keep away from public tools. The Founderz IA para docentes program is a practical, online starting point for educators, from a school with more than 700,000 learners in 170+ countries. It is currently taught in Spanish; for other options, contact Founderz.