Taxonomy of Learning Theories

Academia is teeming with learning theories.

Some of them are old, some of them are new. Some are flash-in-the-pan, others stand the test of time and remain applicable to this very day. Some of them are controversial, while others have assumed the aura of conventional wisdom. Some of them are simple, while others are incomprehensible to mere mortals.

It can be quite a challenge for the modern learning professional to identify an appropriate learning theory, draw practical ideas from it, and apply it to their daily work.

Where do you start? Which theory do you choose? What is its central premise? How does it relate to other theories?
Frustrated man with post-it notes stuck to his face.


To clear some of the obfuscation that surrounds learning theory, I have developed the following Taxonomy of Learning Theories.

Tracey's Taxonomy of Learning Theories

This taxonomy identifies key theories that apply to workplace learning, categorises them according to common properties, and illustrates the relationships among them.

I hope that this taxonomy, along with the corresponding notes below, will assist you in using learning theory to inform your instructional design decisions.

Close up of The Thinker by Rodin

Overarching themes

Almost all learning theory is derived from one or more of the following psychological schools of thought:

  • Behaviourism
  • Cognitivism
  • Constructivism
  • Connectivism

These four psychologies form the overarching themes of my taxonomy.

Cartoon of one dog dog saying to another 'Watch what I can make Pavlov do. As soon as I drool, he'll smile and write in his little book' while Pavlov looks on.


Classical conditioning maintains that a neutral stimulus can be associated with another stimulus that elicits a particular response. This concept was demonstrated in the early 1900s by Ivan Pavlov, who reported that after a period of conditioning, a dog will associate the sound of a beating metronome (neutral stimulus) with food, and respond to it in the same manner (salivate).

Operant conditioning maintains that behaviour is controlled by its consequences: behaviours that are rewarded are likely to be repeated, while behaviours that are punished are unlikely to be repeated. This concept was demonstrated by Edward Thorndike, who placed a cat in a “puzzle box”. The cat discovered that by pulling a ring, a side door fell open which allowed it to escape. So when Thorndike put the cat back in the box, it pulled the ring again.

Social Learning Theory is another theory that has its roots in behaviourism. I somewhat amateurishly consider it operant conditioning by proxy, whereby the learner (especially a child) observes the rewarded actions of someone else, and thus behaves similarly. It’s important to note that Social Learning Theory now extends beyond the behaviourist domain to encompass cognition, particularly through the work of Julian Rotter and Albert Bandura.

Stylised x-ray of a brain in a skull.


Since behaviourism focuses on external behaviour, it considers the mind a black box. In contrast, cognitivism peers inside the box to explain the inner structures and processes of learning.

Models of memory

Numerous cognitivist learning theories derive from the Modal Model of Memory developed by Richard Atkinson and Richard Shiffrin since 1968. They proposed that human memory comprises three components: (1) Sensory memory, which perceives the information that is collected by our senses, such as visual information (eg a drawing) and auditory information (eg a bell toll); (2) Short-term memory, which processes the information that has been supplied by the sensory memory; and (3) Long-term memory, our more-or-less permanent knowledge storage area.

The original Atkinson-Shiffrin memory model, lacking the sensory memory stage which was devised later, showing incoming information going to short-term memory storage then to long-term memory storage.

Atkinson & Shiffrin’s concept of short-term memory was superseded in 1974 by Alan Baddeley and Graham Hitch’s concept of working memory, which comprises the central executive and three slave systems: (1) The phonological loop, which processes verbal information; (2) The visuospatial sketchpad, which processes visual imagery and spatial information; and (3) an integrative component called the episodic buffer.

Baddeley's model of working memory in which the phonological loop, the visuospatial sketchpad and the episodic buffer connect to the central executive.

In 1956, George Miller reported that the “span of immediate memory” is limited to the magical number 7±2 items. From this, he deduced that the amount of information that could be processed at any one time could be increased by “chunking” it.

In the 1970s, John Anderson started to develop ACT-R, which maintains that long-term memory comprises declarative memory which is explicitly stored and retrieved (eg crashing your bike into a tree on your birthday when you were a child) and procedural memory which is unconsciously stored and retrieved (eg the motor skills required for riding a bike generally).

Stylised neurons

Schema theories

Models of memory provide the foundation for subsequent cognitivist theories that (arguably) have more direct implications for instructional design.

In 1977, Richard Anderson extended the work of earlier theorists such as Frederic Bartlett and Jean Piaget. His Schema Theory of Learning maintains that within long-term memory (or more specifically, declarative memory), knowledge is arranged in a hierarchical network of constructs called “schemas”.

Similarly, David Ausubel’s Subsumption Theory proposes that learning involves the linking of new information to relevant points in the learner’s existing cognitive structure. During the learning process, new information is subsumed under more general information in the hierarchical arrangement of the learner’s prior knowledge.

Charles Reigeluth’s Elaboration Theory complements Ausubel’s principle of ideational scaffolding. Reigeluth maintains that instruction should be organised in increasing order of complexity. In particular, the simplest (or epitomised) version of the domain should be provided initially, and elaborated upon subsequently. This approach develops a broad, meaningful context into which the learner can assimilate the narrow, detailed information.

A woman studying.

Cognitive load

In 1988, John Sweller synthesised key principles of memory and schema under a new proposal called Cognitive Load Theory.

Cognitive Load Theory maintains that the mental effort required for learning imposes a cognitive load on working memory. The total cognitive load consists of three components: (1) Intrinsic cognitive load, which is imposed by the intrinsic characteristics of the content that is to be learned; (2) Germane cognitive load, which refers to the mental effort required to organise the elements of the content into a schema, integrate it into long-term memory, and automate its processing; and (3) Extraneous cognitive load, which does not contribute to the learning process (eg the mental effort required to block out loud music).

If the total cognitive load of the learning task exceeds the processing capacity of working memory, learning fails. This suggests that instruction should be designed with a view to reduce cognitive load and thereby avoid overload.

Two workers on a construction site.


Constructivism has a rich history. Numerous theorists have contributed to its development over the last century (eg Jean Piaget, Lev Vygotsky, Jerome Bruner, Ernst von Glaserfeld), and several brands are recognised in the domain (eg cognitive constructivism, social constructivism, radical constructivism).

Regardless of the theorist or the brand, however, constructivism essentially maintains that people learn by constructing their own knowledge on the basis of their experiences. Constructivist learning theories recognise that everyone’s framework of prior knowledge is unique, thus they have their own needs, goals and contexts.

A baby playing with books.


In his study of child development, Jean Piaget posited that every learner has a mental representation of the world which he or she constructs through their experiences.

When a person experiences cognitive conflict (a discrepancy between their mental representation and what they are currently experiencing), they undergo a process of adaptation. If the new experience aligns with their mental representation, the learner assimilates it in the form of new knowledge into their existing schema. If, however, the new experience does not align with their mental representation, the learner must rearrange their existing schema to accommodate the new knowledge.

Clearly, adaptation is complementary to Schema Theory; however, the constructivist perspective emphasises the learner centredness of the activity.

A blacksmith hammering a piece of metal while his colleague supervises.

Situated Learning Theory

A valuable means by which a learner can close the gaps in their existing schema, and broaden and deepen their knowledge, is to engage with other people, ask questions, debate ideas, and share experiences.

Situated Learning Theory focuses on this social, practice-based approach to learning. The theory views learning in terms of participation in a community of practice, and considers knowledge to be highly dependent on its context.

A standing businessman facilitating a training session with a group of colleagues seated in a semi circle.


Andragogy focuses on adult learning, and it adopts a strong constructivist perspective.

It boils down to 5 assumptions about adult learners as articulated by Malcolm Knowles: (1) Adult learners are self directed; (2) Adults bring experience with them to the learning environment; (3) Adults are ready to learn to perform their role in society; (4) Adults are problem oriented, and they seek immediate application of their new knowledge; and (5) Adults are motivated to learn by internal factors.

I have stated previously that I believe Knowles’ 5 assumptions generally hold true – but not for all adults, and certainly not all of the time. An andragogical approach is appropriate for adults who are intellectually mature, self directed and intrinsically motivated, with time to learn and their heads in the right space.

A humanoid figure with networked nodes extended from its head.


While cognitivism focuses on knowledge inside the mind, connectivism focuses on knowledge outside the mind.

George Siemens describes connectivism as “a learning theory for the digital age”. He maintains that in today’s world, there’s simply too much knowledge to take in – and it changes too quickly anyway.

So forget about trying to know everything; instead, exploit technology to extend your knowledge beyond your own brain. Build a network of knowledge sources which you can access as the need arises.

Recognising meaningful patterns among distributed sets of information, rather than storing it all in your head, re-defines what it means to “learn”.

39 thoughts on “Taxonomy of Learning Theories

  1. There’s some things off here though I think. Social learning theories I wouldn’t put under behaviorism. I saw the wikipedia articles on these topics, and they have some issues, too, like labeling Vygotsky a behaviorist and so forth.

    Connectivism is still just a proposal by Siemens and others, it hasn’t been researched and tested or made any predictions or published much in journals. It’s essentially social learning + connectionism. If we’re including every blogger’s learning theory, there are others too, like rhizomatic instruction and so forth.

    Here are some better sources for info on learning theories:

    And see the diagram on that first page:

    Really though this is all decades out of date. That’s the problem with textbooks (and wikipedia articles on this topic, which apparently derive from textbooks). There is newer research for example seriously criticizing cognitive load theory:
    And newer theories and research on learning such as embodied cognition and enactivism:

    Click to access aera10_proceeding_session_99190.pdf

    Along with older theories still thriving today that are ignored by learning textbooks like phenomenology.

  2. There’s some things off with the previous comment I think.

    First – Social learning theories are clearly linked to Behaviorism and Cognitivism in the diagram not just Behaviorism. Second – Connectivism isn’t just some blogger’s theory, its an influential shift in the way we think about workplace training. Third – Research criticizing a theory does not ipso facto invalidate that theory. Fourth – Newer theories and research on learning exists, but this taxonomy is a good starting point.

    Ryan leave the academics in their ivory towers. Keep up the good work for the rest of us in the real world!

  3. Hey’ this is me shaista Ambe all the information regarding learning theories contain vaild information that are easily understandable, especially diagrams are very informative.

  4. You’re very welcome! Just bear in mind that everyone has their own view on theory, and this is simply mine.

  5. Thanks Caitlyn. To be fair to Gerry Stahl (the originator of that diagram on Wikipedia), it provides a graphical representation of how individual philosophers contributed to the development of different learning theories – hence the personal names.

    But I take your point: That diagram is far too abstract to be meaningful to anyone in the corporate sector.

    My objective here was to assist practitioners in applying theory to their daily work; to provide them with something practical. I take it from your comment that I have done so – cheers!

  6. Like this compilation of theories. Concise and avoids the argumentation by who has the most extensive collection of web links hidden in their cod piece.

    In defense of text books, they represent a distillation of knowledge that someone actually takes some responsibility for.


  7. New technology is an excellent resource to learn and look for information beside the traditional book.The point is to be able to move thoughout this spiderweb , to select the best information in the lyberinth.When are we ready? Do we need a previous training?

  8. I agree, Nautilus.

    The constructivist/connectivist style of learning that you describe is not necessarily what the learner is used to. It demands serious meta-learning skills which they may not have.

    Are you familiar with Stephen Downes’ views of PLEs and modern learning? See:

  9. Nautilus, you ask if we need previous training to successfully work our way through all the content so freely available to the self-motivated learner? Yes we do but where will it come from? Institutional education has abandoned critical thinking skills in favour of countable lower order thinking “proofs”. The latest soon-to-save-us commercial design crowd hardly has any interest in creating anything more complex than rapid-response, edu-nuggets served from public vending machines.

    Our back-up is a culture that respects independent thought over slogans and marketing nostrums. In the West this training of the next generation has been left up to soccer moms and pimply peer groups. My guess is the whole self education thing will be broken into people forced into self training (paid out of their own pockets) as education and even professional development falls into the user-pay category of what we considered a social value at one time. And the fun learning stuff will become the marginal pastime of grumpy malcontents (like myself) and hopelessly curious people who think learning still has value.


  10. Dear Ryan,

    Great work! What is the copyright policy on this article? Can it be distributed to students for educational purposes provided the work is fully cited and gives credit to you?


  11. Reblogged this on My Mind Bursts and commented:
    Once learning theories are second nature I won’t need to blog about them so much. This post does it well, covering the ground in a succinct way.

  12. I would like to cite this as you recommended out of your book, but I won’t get the book in time for my project due date. Can you let me know what the APA citation would be so that I can still cite this correctly since I don’t have a hard copy of the book to see what the details are of the publisher, year, location, etc. I can still order the book for future reference, but I would like to at least be able to use one quote for now from this. Thanks.

  13. Tracey, R. (2011). E-Learning Provocateur (Vol. 1). Sydney, Australia: Ryan Tracey.

  14. It IS very confusing AND much of it is based on each person’s opinion and experience. There is so much overlap between theories, I do not believe there is a right or a wrong. Just my opinion.

  15. Connectivism -“George Siemens describes connectivism as “a learning theory for the digital age”. He maintains that in today’s world, there’s simply too much knowledge to take in – and it changes too quickly anyway.

    Question- is this a tongue-in-cheek statement, or this person’s position that the access to information that we have in this era is a back thing?

  16. Hi mrsbrittons. I don’t know what you mean by a “back thing”…?

  17. Well I guess you’d have to ask George that, but my interpretation would be no, the access to information that we have in this era is not a bad thing. Notwithstanding that, however, there is an incredible (unmanageable?) amount of it.

    I think the advent of the internet and, subsequently, social media, has proliferated the amount of information we have access to. Erik Qualman’s Socialnomics 2014 provides a nice overview , while the stats on Wikipedia’s information explosion page about web servers and blogs are sobering.

    All this is a wonderful wonderful thing, but of course it’s becoming harder and harder to handle. Whether connectivism is a theory or otherwise, I consider it a lens through which we see knowledge and learning in the digital age. It acknowledges that how we learn today is necessarily different from how it was 20 or 30 years ago.

  18. It could be that we can now see more knowledge in a short review of the raw web than we could through resources in the past where we were presented with material curated down to a manageable amount? A magazine writer in the past may have waded through mountains of nearly identical content and “named” it as one thing, where in a quick view of the web we wouldn’t notice near duplicates but see things each by each–in their pre-connected and individual “wild” state?

    Could it be our brains that need to resolve things to single solutions arising from the application of single theories and not the universe that requires this? We definitely seem fond of patterns. And projecting meaning onto causes that are suggested by those patterns is a good way to look smart like Eric does. Thinking out loud here it seems because we can measure variables to an ever smaller degree we can be right to an ever higher degree? What if these things cancel out and we aren’t moving at all?

  19. Scottx5, I think your point about curation is quite pertinent. In the past the number of information inflows was finite and (in the case of a newspaper or magazine) curated. Of course we still have newspapers and magazines today, but we also have a multitude of other information inflows that aren’t curated. Again, this may be a good thing — for example, to counteract bias — but it obliges the information consumer (learner) to become their own curator.

  20. Maybe more important than availability itself is we no longer have curators selecting and highlighting based on what they think is significant. I’m not sure if it is a benefit to be free of the “influence” of thought leaders as curators if we are merely trading influence for unprocessed volume. I habitually argue with everything but still value being pointed in directions I never would have discovered by myself. Not knowing what I don’t know keeps me open to input, though I can’t explain the paradox.

  21. I couldn’t agree more, Scott. It is in instances such as the one you describe where I think expertise tends to be under valued by some of the more progressive learning theories. If someone who is knowledgeable and highly experienced recommends a particular resource or direction for me, I’ll accept it graciously thank you very much!

  22. Accepting advice graciously is an interesting behaviour Ryan. Part of the “purpose” of knowing people people who “know” is to help us recognize value I think. Not just a convenience for me that they will sort through the over-abundance but point out things worth pursuing. I’ve struggled with authors and ideas I’d never touch in a million years on the assurance of someone I admired that the effort would pay off. Aside from the social qualities of showing interest in the interests of others, there must be something in there of taking someone’s word that it is possible to understand if we try?

  23. Hiya!

    nice work, proposing an order of those approaches. I’m wondering what might be the connections between constructivist and connectivistic approach.

    Also, if you haven’t heard of it, I’d refer you to the Extended Mind concept: This is rather a groundwork and applications of this theory are – I think – rather limited but it’s interesting anyway and quite compelling in terms of instructing others. Thinking of a mind like a closed vat processing information but basically separated from other information storage and processing devices that our culture developed and put all around us is a faulty approach – according to it. Thinking of “mind” is only fragmentary – think of more ecological approach, a system of information exchange in which mind is only one link – to put it in simple words.

  24. Thanks for pointing this out to me, Nikos. It seems to me that the extended mind is very similar to distributed cognition, though I suppose the former would also include elements such as AI and even the handy calculator.

  25. @Ryan

    I think they can be connected, surely. As I browse through materials on distributed cognition, it seems like it sprouted from social sciences and extended mind is of more philosophical character, both being a part of computational paradigm and approach towards explaining behaviour.

    Distributed cognition seems to focus on organisms and their communities, societies they create and extended mind is also about material culture – notepads and recently certainly internet-of-things devices.

  26. This is a great post on the different theories most commonly used. I’m researching data and how it’s used in instructional system design. I think that each theory has a time and place in learning. Data can be what helps a designer determine which theory would be best. What are your throughts on this? Do you feel that data should be used to determine which theory to use?

  27. A very interesting question, Sabrina.

    I’m an advocate for using data to inform all our activities, and that would include the application of learning theory to instructional design.

    An example off the top of my head would be to refer to the usage stats from the enterprise social network to determine your target audience’s aptitude for a connectivist approach to learning.

    Do you have another example in mind?

  28. I love a good blog post and always find the comments underneath revealing. A great form of connectivism and social construction, fun yet inherently false – and all that is wrong with both approaches. I think the glaring issue with Siemens (and Downes) approach (connectivism) and any form of socially constructed knowledge is that it can be driven by the narrative of the times and will always lack objectivity. This is from users own bias towards certain sources and machine algortihms providing content and links which expand on the (subjective) narrative.

  29. Your not the first critic of connectivism, Steve, and you won’t be the last.
    Thanks for adding your point of view.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.