Cognitive Learning Theory: 11 Examples and How to Use It

Scale training programs and cut development time using AI

Start Free

Picture a new account manager at a commercial insurance broker, three weeks into the job. During her induction she read the firm's 12-point renewal checklist, took the quiz at the end of the module and got every question right. Now a panel beating workshop is due for renewal in 30 days. She renews it on last year's terms, and nobody notices that the $85,000 spray booth the client bought in March is missing from the asset schedule. The gap will show up only if the client ever claims for the booth. She had seen the checklist. She did not use it to catch the change when the work needed it. Recall may be part of the problem, but so may access to the checklist, supervision or the renewal workflow.

Cognitive learning theory explains that gap, and most of its practical advice is about closing it. I'm an instructional designer at Coursebox, where we make instructional design software that builds courses from a brief, so I'll mention our product where I tested it. In this guide I explain what the theory says, where it came from and the memory model behind it. Then I walk through 11 examples of cognitive learning theory in one workplace course, each with the evidence behind it and a classroom version. Along the way I cover two real cases, a product test and two free templates.

What is cognitive learning theory?

Cognitive learning theory, or cognitivism, explains learning as mental processing: people attend to information, hold a small amount of it in working memory, connect it to what they already know in long-term memory, and later retrieve it to use it.

View the full-size diagram

Peggy Ertmer and Timothy Newby's comparison of learning theories for instructional designers, first published in 1993 and reprinted in 2013, gives the working version. Cognitive theories "address the issues of how information is received, organized, stored, and retrieved by the mind", and "Learning results when information is stored in memory in an organized, meaningful manner" (2013 reprint, pp. 51 to 52). Forgetting, in the same account, is "the inability to retrieve information from memory because of interference, memory loss, or missing or inadequate cues needed to access information" (p. 52).

That puts cognitivism in contrast with behaviourism, which explains learning through observable responses and their consequences. A cognitivist still cares about what learners do, but treats it as evidence of what is happening in their heads: what they noticed, how they organised it, and whether they can find it again when they need it. Ertmer and Newby describe the learner as "a very active participant in the learning process" (p. 51), and they note that cognitive theories are "usually considered more appropriate for explaining complex forms of learning (reasoning, problem-solving, information-processing)" (p. 52).

Three points set the scope:

• It is a family of ideas, not one theory with one author. Information processing, working memory, meaningful learning, schema and metacognition all sit under the label, and each has its own researchers and dates.
• It is a theory of learning, not a teaching method. Its design rules, such as chunking, advance organisers and retrieval practice, were drawn from it later by researchers and instructional designers.
• Its current form is evidence-led. Much of what designers now take from cognitive psychology comes from controlled studies of memory and practice, such as the research on testing and spacing.

Here are the main ideas at a glance, with the design move each one suggests.

Who developed cognitive learning theory?

No single person founded it. The theory grew out of the cognitive revolution in psychology, which George Miller, one of its leaders, called "a counterrevolution" in a 2003 history of the period. The revolution was against behaviourism, which had defined psychology as the science of behaviour. Noam Chomsky, Miller recalls, said that was "like defining physics as the science of meter reading" (p. 142). Miller dated the moment of conception of cognitive science to 11 September 1956, the second day of a symposium at the Massachusetts Institute of Technology.

The main names and dates:

• 1956 and 1960, Jerome Bruner. Bruner, Jacqueline Goodnow and George Austin published A Study of Thinking in 1956, a book which, in Miller's words, "took seriously the notion of cognitive strategies". In 1960 Bruner's group became the Harvard Center for Cognitive Studies.
• 1967, Ulric Neisser. His book Cognitive Psychology gave the new field its name.
• 1968 and 1974, the memory models. Richard Atkinson and Richard Shiffrin set out the stores of memory, and Alan Baddeley and Graham Hitch replaced the short-term store with a working memory system.
• 1960s, David Ausubel. He argued that meaningful learning depends on connecting new material to what the learner already knows, and introduced advance organisers.
• 1979, John Flavell. He framed metacognition as knowing about, and monitoring, your own thinking.

Jean Piaget is often named as the founder, sometimes with a date in the 1930s. His work on children's thinking came from a separate tradition in Geneva, which Miller lists among the schools abroad that behaviourism had never displaced (p. 142). In instructional design Piaget is usually placed as the root of cognitive constructivism, and his central idea was assimilation: "the fundamental relation is one of assimilation", he wrote in 1964, and "Learning is possible only when there is active assimilation." He is part of the story, as the cognitive-developmental strand, but he did not start the cognitive revolution.

Instructional design took the theory up quickly. Robert Gagné's nine events of instruction are, as one review puts it, "based on the information processing model of the mental events that occur when adults are presented with various stimuli". Our guide to Gagné's instructional design model covers the nine events. Learning objectives are a separate question: our guide to Bloom's taxonomy covers how to classify them.

What to read first

• Ertmer and Newby, "Behaviorism, Cognitivism, Constructivism" (1993; reprinted with commentary in Performance Improvement Quarterly in 2013). Read this first if you design training. It sets out what each theory says about learning, memory and transfer, and lists the design strategies each one supports.
• Dunlosky and colleagues, "Improving Students' Learning With Effective Learning Techniques" (2013). It rates ten techniques students use to study on their own by the strength of the evidence, which makes it a good starting point for choosing which to build into a course.
• Miller, "The cognitive revolution: a historical perspective" (2003). Four pages on how the field began, from someone who was there.

How cognitive learning theory explains memory

Most cognitive accounts of learning start from a model of memory. The best known is Atkinson and Shiffrin's from 1968, which Paul Kirschner, John Sweller and Richard Clark later described as the base most treatments use. In the original paper, Atkinson and Shiffrin divide memory "into three structural components: the sensory register, the short-term store, and the long-term store" (p. 90). "The short-term store is the subject's working memory", and information in it "decays completely and is lost within a period of about 30 seconds" unless a control process such as rehearsal keeps it there. The long-term store is "a fairly permanent repository for information" (p. 91).

In 1974 Baddeley and Hitch replaced the single short-term store with a working memory system of three parts: a central executive that controls attention, a phonological loop for sounds and words, and a visuospatial sketchpad for images and space. In a 2000 update, Baddeley added a fourth part, the episodic buffer, and defined working memory as "a limited capacity system allowing the temporary storage and manipulation of information necessary for such complex tasks as comprehension, learning and reasoning" (p. 418).

Long-term memory is where schemas live: organised structures of knowledge that new information is fitted into. Our guide to schema theory covers how they form and how to activate them. Over all of this sits metacognition, which Flavell framed as "cognitive monitoring" in his 1979 article. He separated the two kinds of strategy neatly: "Cognitive strategies are invoked to make cognitive progress, metacognitive strategies to monitor it" (p. 909).

These parts are not a pipeline that every piece of information passes through once in order. Working memory draws on long-term memory as much as it feeds it, prior knowledge changes what a learner notices in the first place, and monitoring runs across the whole process. For a designer, three practical consequences follow:

• Limit what is new at any moment, because working memory is small and unfamiliar material can be lost quickly without appropriate attention and support.
• Connect new material to what learners already know, because that is how it gets organised in long-term memory.
• Make learners retrieve it, more than once and over time, because, in the cognitive account, forgetting is often a failure to find what was stored.

The "over time" part is the one training most often skips, and one of the clearest workplace tests of spacing comes from a British sorting office.

Case study: Postmen learning to type postcodes for the British Post Office

This was a randomised trial of four training schedules, run with the British Post Office and published in 1978 by Alan Baddeley, of the working memory model, and D. J. A. Longman. The Post Office "estimated that with the advent of mechanical letter sorting it might prove necessary to train up to 10000 operators" on a typewriter-style keyboard, and it had to choose between training a few people intensively or many gradually alongside their normal duties.

• What they did: 72 volunteer postmen and Postmen Higher Grade from the Croydon sorting office were randomly assigned to four groups of 18: one 1-hour session a day, two 1-hour sessions a day, one 2-hour session a day, or two 2-hour sessions a day. Every group had the same instructor-led method and a planned 60 hours of training, alongside normal duties. At one hour a day that took 12 weeks; at two 2-hour sessions a day it took 3. Training was later extended to 80 hours for all but the one-hour-a-day group.
• Evidence and result: In Baddeley and Longman's paper in the journal Ergonomics, among the 70 who completed training, after 60 hours the one-hour-a-day group keyed 79.31 correct keystrokes a minute with 1.09% uncorrected errors, against 64.78 and 2.06% for the two-sessions-of-two-hours group. It learned the whole keyboard in 34.9 hours against 49.7, and after 60 hours it was faster and more accurate than the most intensive group was after 80. Retested one, three or nine months later (55 analysed), the most intensive group again did worst. The trainees' own ratings "would have led to a recommendation of a schedule which is associated with the slowest learning, the least accurate performance and the poorest retention."
• Limitation: Groups of 16 to 18 people at one sorting office, a keying skill rather than knowledge, and 1970s equipment. The trial compares schedules with each other, not with no training, and the paper recommends a schedule without reporting that the Post Office adopted it. The spread-out schedule also took four times as long on the calendar.
• Lesson: Spread the same training hours into shorter sessions over more days when the calendar allows, judge a schedule by performance rather than by how trainees rate it, and plan for the longer elapsed time.

11 cognitive learning theory examples, worked through one course

Here is the hypothetical example I'll use from here on. A mid-size commercial insurance broker in an Australian city looks after about 3,000 small-business clients and renews roughly 250 policies a month. Six new account managers have just joined. Some come from claims handling or personal-lines sales at other firms, and some from the broker's own support team. They know insurance terms, but none has run a small-business renewal alone, and each arrives knowing a different amount. A senior account manager demonstrates the first renewal.

The course builds towards one task: running a renewal with the broker's own 12-point checklist, from 30 days before expiry to cover confirmed, and writing the file note. The checklist belongs to this broker, not to the industry, and the course leaves insurance law and licensing to the firm's own procedures. The checklist comes in three groups:

• Before the review: 1 pull last year's schedule and claims history; 2 check what has changed in the business, such as activities, turnover, wages and staff; 3 list new or sold assets and equipment; 4 check premises changes, such as a new site or lease.
• At the review: 5 test each sum insured against current replacement values; 6 check contract and lease requirements, such as liability limits and parties to be noted; 7 record claims or incidents since the last renewal; 8 ask about subcontractors and work away from the usual site.
• Placing and confirming: 9 approach insurers and compare terms, not only premium; 10 explain changes in cover, excesses and exclusions in writing; 11 get the client's written instructions before expiry; 12 write the file note and confirm cover is in place before the expiry date.

Three clients carry the course, and all the figures are illustrative:

• The worked example, a panel beating workshop due for renewal in 30 days. Last year, plant and equipment were insured for $240,000, public liability was $10 million, turnover was $1.2 million and there was one claim, for stolen tools ($6,400). This year it bought an $85,000 spray booth in March that is not on the asset schedule, turnover is about $1.5 million, and a new lease requires $20 million public liability.

For the simplified valuation in the walkthrough, assume the existing plant’s verified current replacement value is still $240,000, the booth’s installed replacement value is $85,000 and no other equipment was sold or replaced. These supplied facts justify the $325,000 teaching total; real renewals require current values and the applicable cover terms.

• The practice client, a dental practice. It added a second treatment chair ($60,000 of equipment) and a part-time hygienist, with no claims and no change to the lease.
• The transfer client, an electrical contractor. It now works on commercial sites whose contracts require $20 million liability and the head contractor noted on the policy, and it uses two subcontractors.

The intended outcome is that a new account manager can run a small-business renewal with the checklist, practise recalling the groups and their purpose, consult the current checklist during real work, explain why each is there and write a file note a senior colleague would accept. It gives nobody advice authority or a compliance sign-off.

This example suits cognitive learning theory because it puts pressure on memory in every way the theory describes. Twelve unfamiliar items can be demanding; familiarity, organisation and access to the checklist change that demand. Learners need to recognise relevant changes and use the current checklist weeks and months later, because renewals come round every month and each client returns a year on. Mistakes are costly and easy to miss, since an asset left off a schedule surfaces only at claim time. And the six learners bring different prior knowledge. Each example applies one principle to this course. The order follows the course; it is a design choice, not a sequence the theory requires.

1. Ask a question before the content

Cognitive theories start from what the learner already knows. Ertmer and Newby put it plainly: "Instruction must be based on a student's existing mental structures, or schema, to be effective" (pp. 53 to 54). A question before any teaching does two jobs. It shows the designer and the learner what is already there, and it gives the learner a question that the next pages answer.

• In the renewal course: Before any content, learners answer in open text: "A client's renewal is due in 30 days. What would you check first, and why?" Nobody marks it right or wrong. A former claims handler and a former support officer will answer it differently, and both answers are useful.
• Why it works: In Richland, Kornell and Kao's 2009 experiments, learners did better on the final test after a pretest "even though only items that were not successfully retrieved on the pretest were analyzed": getting the pretest wrong still helped. The effect is narrow. A 2024 meta-analysis by St. Hilaire, Chan and Ahn found "a moderate specific effect (g = 0.54, k = 97) but a virtually nonexistent general effect (g = 0.04, k = 91)", so the questioned points improve and the rest of the lesson does not. And as Pan and Carpenter's 2023 review notes, prequestions help when learners have "an opportunity to study the correct answers afterwards".
• In a classroom: "What do you already know about volcanoes?" before the lesson, or three quick questions the lesson goes on to answer.

2. Show the whole map before the detail

An advance organiser is a short overview, given before the content, that the details can hang on. Ausubel introduced the idea, and Ertmer and Newby list "advance organizers" among the core cognitive strategies (p. 53). As Joseph Novak and Alberto Cañas summarise in their account of the theory behind concept maps, "The fundamental idea in Ausubel's cognitive psychology is that learning takes place by the assimilation of new concepts and propositions into existing concept and propositional frameworks held by the learner" (p. 3).

• In the renewal course: One screen shows the renewal timeline (30 days out, the review, placing, expiry) with the three groups named, before any item is taught.
• Why it works: The timeline gives each of the 12 items a place before learners meet it, so an item such as "test each sum insured" arrives as part of "the review" rather than as a free-floating rule. Our guide to elaboration theory compares advance organisers with other ways of sequencing content from general to detailed.
• In a classroom: A one-page concept map or outline of the unit on the first day, returned to at the start of each lesson.

3. Chunk the content into a few groups

George Miller's 1956 paper is remembered for "seven, plus or minus two", but his argument was that memory limits are counted in chunks: "the number of chunks of information is constant for immediate memory" (pp. 92 to 93). He also called the recurring sevens "a pernicious, Pythagorean coincidence" (p. 96). Later work put the limit lower. In his 2010 summary, Nelson Cowan places it at about four chunks, or three to five.

• In the renewal course: The 12 items are taught as three groups of four, each with its own name and its own section: before the review, at the review, and placing and confirming.
• Why it works: Three named groups help learners organise the twelve items; naming a group does not automatically make its four unfamiliar items one mastered chunk, and each group maps to a stage of the work. Our guide to cognitive load theory goes further into working memory limits and how to cut unnecessary load.
• In a classroom: A 12-step science procedure taught as three stages (prepare, test, record), with the stage names on the board.

4. Walk through a worked example, then fade it

A worked example shows an expert solving a problem step by step before learners try one. The worked example effect "was first reported by Sweller and Cooper (1985)", according to Sweller, van Merriënboer and Paas's 2019 review, and the same review warns that it reverses as learners gain skill: "Past a certain point, studying worked examples may be counterproductive and they should be faded out and replaced by problems" (p. 277).

• In the renewal course: The senior account manager walks the panel beater through the checklist out loud. She finds the spray booth at item 3, checks current replacement values at item 5; in this simplified example, the verified $240,000 of existing plant plus $85,000 installed replacement value for the added booth totals $325,000, flags the lease's $20 million liability requirement at item 6, compares two insurers at item 9, puts the changes in writing at item 10, gets written instructions at item 11 and writes the file note at item 12. Learners then attempt the dental practice with support matched to their performance, followed by the electrical contractor without unsolicited hints but with the current checklist available.
• Why it works: Novices who must solve a problem before they have seen one solved spend their working memory searching for a method. Seeing it done first frees that capacity for learning the steps, and fading the help moves them towards doing it alone.
• In a classroom: A fully solved maths problem, then a half-solved one, then one to solve alone.

5. Pair the diagram with words that add to it

Richard Mayer's cognitive theory of multimedia learning assumes that everyone processes words and pictures through two channels, each with limited capacity, and learns by actively organising both. That is a design rule for all learners. It does not mean that some people are "visual learners" and others are not. Our guide to Mayer's cognitive theory of multimedia learning covers his principles in detail.

• In the renewal course: The renewal timeline appears as a simple diagram with short labels, and a few lines of explanation sit beside it. The text does not repeat what the labels already say.
• Why it works: Words and a diagram together give learners two linked ways into the same idea. Avoid needless duplication, but retain labels, accessible text and explanations learners need. Printed words beside a picture are processed through visual input; the channels are not a rule requiring every label to be different from nearby text.
• In a classroom: A labelled diagram of the water cycle explained aloud, rather than a slide of text read out word for word.

6. Ask learners to recall, not reread

Retrieval practice means making learners produce information from memory instead of reviewing it again. Henry Roediger and Jeffrey Karpicke open their 2006 study "Test-enhanced learning" with the line "Testing is a powerful means of improving learning, not just assessing it" (p. 249). In their first experiment, with 120 students learning prose passages, rereading won at five minutes (81% against 75%). After a week, "The tested group recalled 56% of the material, whereas the restudy group recalled only 42%" (p. 251). Recall pays off later, not straight away.

• In the renewal course: After each group, learners type the four items from memory, then see the list and compare. A multiple-choice question that shows all 12 items would test recognition rather than unaided production. Well-designed multiple-choice questions can also involve retrieval and judgement. Choose the format for the intended evidence.
• Why it works: Retrieval with accurate feedback can improve later access to information; difficulty alone is not evidence that learning improved. The effect holds in real classrooms: a 2021 meta-analysis by Yang and colleagues, covering 222 studies and 48,478 students, found that quizzing raised achievement "to a medium extent (g = 0.499)".
• In a classroom: A low-stakes "brain dump" at the start of the next lesson: students write everything they remember before opening their notes.

7. Bring earlier material back later

Spacing means putting a gap between practice sessions instead of massing them together, and spaced retrieval combines it with recall. Repetition in one sitting is not spacing. The evidence for the spacing effect is broad.

In that review, Nicholas Cepeda and colleagues concluded that "separating learning episodes by a period of at least 1 day, rather than concentrating all learning into one session, is extremely useful for maximizing long-term retention" (p. 370). The studies were of verbal recall tasks, so read it as strong evidence for remembering facts and words rather than a guarantee for every skill. Their 2008 follow-up with 1,354 adults found that the best gap depends on how long the knowledge must last: it "declined from about 20 to 40% of a 1-week test delay to about 5 to 10% of a 1-year test delay" (p. 1095). Their advice was blunt: "you need to decide how long you wish to remember something" (p. 1101). One large workplace dataset points the same way. In an observational study of 10,514 employees at retailers, pharmaceutical sales teams and a university, "the optimal amount of spacing between repeated retrieval events increased as the retention interval increased."

• In the renewal course: The check after the second group also asks about the first; the check after the third asks about both earlier groups. A mixed review follows one week after the course and again three weeks after. Renewals come round every month and the same client returns a year later, so the one- and three-week reviews are a proposed starting schedule to test, not intervals established for this job. The within-course checks are cumulative retrieval; a delay is needed to make them spaced practice.
• Why it works: Each return trip to an item, after some forgetting has started, makes the next recall easier and the memory more durable. In a 2019 randomised trial at a Canadian hospital, 58% of staff who practised CPR monthly achieved excellent CPR on a feedback manikin after a year, against 15% of those who practised yearly (167 analysed). The monthly group also practised more in total, so that trial cannot separate spacing from extra practice. Our guide to spaced repetition covers how to schedule it.
• In a classroom: Each week starts with three questions from last week and one from last month.

8. Mix the practice

Interleaving means mixing different kinds of problem in one practice set instead of practising one kind in a block. In John Dunlosky and colleagues' 2013 review of ten study techniques, interleaved practice was rated moderate utility (p. 5), one step below practice testing and distributed practice. Its value lies in making learners decide which approach a problem needs, not only how to carry it out.

• In the renewal course: Three short client changes appear in random order: a new site, a new subcontractor and a claim last month. For each, learners name the checklist item that catches it and say why.
• Why it works: In a blocked set on "premises changes", the answer is always item 4. In a mixed set, learners have to recognise the situation first, which is what a real renewal asks of them.
• In a classroom: A maths set that mixes area, perimeter and volume problems, so students must choose the formula before using it.

9. Ask "why"

Elaboration means connecting new information to what learners already know by explaining it. Dunlosky and colleagues rated two forms, elaborative interrogation (asking "why is this true?") and self-explanation, as moderate utility (p. 5). A "why" question also shows the designer whether a learner has understood the reason for an item or only memorised its wording.

• In the renewal course: Learners answer in open text: "Why does the checklist ask about new equipment before the sums insured?" A model answer follows: if the spray booth is not listed first, the sum insured is tested against an incomplete schedule.
• Why it works: Explaining a reason links the item to other knowledge, which gives the memory more ways back in. It also turns the checklist into a set of reasons, which helps when a client's situation does not fit the list neatly.
• In a classroom: "Explain why the rubber band snapped back" after a physics demonstration, written in a sentence or two.

10. Ask learners how sure they are

Metacognition is knowing about, and monitoring, your own thinking. Flavell's 1979 article is often credited with the phrase "thinking about thinking", but it does not use it. His words are "knowledge and cognition about cognitive phenomena" and "cognitive monitoring" (p. 906), and his verdict was that "there is, overall, far too little rather than enough or too much cognitive monitoring in this world. This is true for adults as well as for children" (p. 910).

• In the renewal course: Every check asks "How sure are you? (sure / unsure / guessing)" before showing the answer. At the end, learners mark which items they recalled without the checklist and name one to practise.
• Why it works: Rating confidence before seeing the answer gives learners a direct comparison between what they thought they knew and what they did know. That comparison is the information they need to decide what to practise.
• In a classroom: Students hold up a green, amber or red card before a quiz answer is revealed, then note where their colour was wrong.

11. Give feedback that explains, then a new case

Cognitivists, in Ertmer and Newby's account, "make use of feedback (knowledge of results) to guide and support accurate mental connections" (p. 53). Feedback that only says "incorrect" tells learners they are wrong; feedback that names the item and the reason tells them how to be right. The final test of understanding is a case that looks different from the ones practised.

• In the renewal course: Feedback on every check names the checklist item and the reason it applies. Then comes the electrical contractor, with no hints: learners use the checklist to write what they would check and a short file note, and compare it with a model file note that ties the new contracts to item 6, the subcontractors to item 8, and the placing steps to items 9 to 12.
• Why it works: The electrical contractor's changes look different from the other two clients' (new contracts, a head contractor to be noted and subcontractors, rather than new equipment), so learners have to apply the checklist's logic rather than match a remembered example. That is the difference between recalling the list and being able to use it.
• In a classroom: The same percentage rule taught with shop discounts, then applied to a sports statistic.

How to use cognitive learning theory to design a course

Cognitive learning theory gives principles, not a procedure, so this workflow is my own. It follows the evidence for the 11 moves, and each step uses the renewal course.

1. Decide how long they must remember it. The answer sets your review gaps. The job needs ongoing accurate checklist use. Start with proposed reviews at one and three weeks, check delayed performance and adjust the intervals; these two reviews do not establish lifelong retention.
2. Find out what they already know. Ask an open question before the content, and read the answers before you finalise the pace.
3. Show the map, then chunk the content. Give the overview first, then teach in a few named groups, an amount matched to prior knowledge and the task, with the current checklist available.
4. Show one worked example, then fade the help. Walk an expert through one full case, then reduce help when performance supports it, retaining aids permitted in the real task.
5. Write checks that ask learners to produce the answer. Typing the four items from memory asks more of learners than picking them from a list.
6. Plan where each item comes back. For every check, name the earlier items it reviews, and schedule a mixed review after the course.
7. Add a confidence step and a "why" question. Ask how sure learners are before each answer, and ask at least one question about why a step exists.
8. End with explained feedback and a new case, then check later. Name the item and the reason in feedback, finish on a case learners have not seen, and look at the delayed review results before you call the course finished.

The cognitive design plan for the renewal course

Here is the plan for the six new account managers, with each move placed in the course.

Adapt the plan for procedures people need over time, such as a site safety inspection, a month-end close or a customer complaint process: chunk the steps, show one done well, and plan when each step comes back.

What Coursebox course requests show about quizzes

To see whether course briefs ask for the moves that make quizzes work, I looked at course requests typed into Coursebox in the 90 days to 7 October 2026: one per person, from English-language browsers, with no uploaded file. Of 266 prompts, 104 were course requests, and about a quarter of those were for training at work. Two independent AI readers each classified all 104 against written definitions, and I re-read every positive request in full. Counts combine requests flagged by either reader. They are classification estimates, not statistical upper bounds.

Twenty of the 104 requests asked for quizzes, knowledge checks, tests or questions in some form. What they asked those questions to do was narrow.

Every one of the 20 treated its questions as a check on the module just finished or as a final test, such as "a quiz in each module" or "a test at the end of the course". The combined classifications flagged 9 as placing checks through the course rather than only at the end, and only 2 framed questions as practice rather than grading. None of the 104 asked learners to rate their own understanding, explain something in their own words, or remember anything as a stated goal. Practice testing and spaced practice were the two techniques Dunlosky's review rated highest. Spaced practice needs questions to come back after a gap, and none of the 20 briefs asked for that.

The sample is small, it is mostly not workplace training, and it shows what people typed rather than what their finished courses contain. Teams may add spacing by editing, through their learning platform or with reminders, and a request is not a course. If you want questions to come back, name the earlier items each check should draw on in the brief.

Cognitive learning theory vs behaviourism, constructivism and social cognitive theory

Ertmer and Newby say they have "consciously chosen not to advocate one theory over the others" (p. 61), and they argue the choice depends on the learners' "present level of competence and the type of learning task". This table is my synthesis along those lines, not a ranking.

The boundary with constructivism is the one most often blurred. Ertmer and Newby call constructivism "a branch of cognitivism (both conceive of learning as a mental activity)" (p. 55); the difference is that constructivists see meaning as built by each learner in context, where cognitivists focus on how information is organised and retrieved. Piaget sits on the constructivist side of that line. Our comparison of behaviourism, cognitivism and constructivism sets the three side by side, and our guide to constructivism covers its strands.

Social cognitive theory is the other common mix-up because of the shared word. Bandura's social cognitive theory explains learning through observation, self-efficacy and the interplay of person, behaviour and environment.

For the new account managers, I would use cognitivism for the checklist itself, social cognitive theory for the senior account manager's demonstration, and situated learning for the weeks they spend running real renewals beside her.

Limitations and criticisms of cognitive learning theory

Cognitive learning theory has strong evidence behind its core techniques and well-known blind spots around them.

• Most of the evidence comes from students in controlled settings. The professional evidence comes mainly from health care: in Larsen, Butler and Roediger's 2009 study, 40 residents remembered 39% of material after repeated testing against 26% after repeated study, about six months later. The non-healthcare examples examined for this guide include the 1978 Post Office trial and workplace-platform observations; they are not an exhaustive account of professional evidence. Treat effect sizes from lab studies as a direction, not a forecast for your course.
• It treats knowing as separable from doing. John Seely Brown, Allan Collins and Paul Duguid argued in their 1989 paper on situated cognition that much cognitive science "assumes that knowing is a process that can be separated from the activities and situations in which it is used". Knowing the checklist is not the same as running a renewal with a client who is in a hurry.
• It is "cold". An early description, by Ann Brown and colleagues in 1983, called academic cognition "relatively effortful, isolated, and cold". Quoting it, Paul Pintrich wrote in 2003 that once cognitive researchers worked in classrooms "they quickly became aware of the importance of motivational factors, as well as social factors" (p. 679). A memory-focused account alone does not explain willingness to engage. Cognitive research also examines motivation and emotion; use the criticism to identify what this design still needs.
• Lab tasks can differ from everyday life. Ulric Neisser, who named the field, later warned that "the artificial situation created for an experiment may differ from the everyday world" (1976, p. 33, as quoted by Holleman and colleagues in 2020). Much of the classic evidence uses word lists, trivia facts and short passages.
• The mind-as-computer picture is a metaphor. At Carnegie-Mellon, Miller recalls, the new field was called "information-processing psychology", and critics, Bruner among them in his later work, argued that processing information is not the same as making meaning.
• Gains can fade when the practice stops. In one hospital trial, an estimated post-training change did not persist; its uncertainty needs to stay visible.

Case study: Clinicians answering spaced quiz questions at Vanderbilt in Nashville

This was a randomised crossover trial at one US academic medical centre, measured on real orders in the health record rather than a quiz. Vanderbilt University Medical Center wanted clinicians to follow the evidence on intravenous fluids and on opioids at discharge, and set out to test whether "a smartphone-based app using spaced education with retrieval practice is an effective method to increase evidence-based practice."

• What they did: Vanderbilt's own app pushed short questions with explanations to clinicians over two four-week modules between January and April 2020, with the order of modules randomised so that each group was the other's control. Of 354 clinicians randomised, 349 were sent 16,868 questions; 70.0% were opened and 92.4% of those were answered.
• Evidence and result: In McEvoy and colleagues' 2022 trial report in JAMA Network Open, analysed on the 206 clinicians who ordered fluids, evidence-concordant ordering was an estimated 3.9 percentage points higher than the comparison group after training (95% CI, -8.2% to 16.0%), with an estimated difference of -9.5 percentage points at the later comparison, and the interaction between intervention and time was significant (P = .002). In the authors' words, "the improvement was not sustained after 8 to 12 weeks." Opioid prescribing did not change at any point.
• Limitation: One centre, volunteers and no blinding. The post-training difference on its own has a confidence interval that crosses zero, and Vanderbilt built and evaluated its own app.
• Lesson: This trial found a significant difference in the pattern over time, but the immediate fluid-ordering estimate was uncertain and opioids did not improve. It does not establish that continued questions or point-of-decision prompts would sustain improvement; those are options to evaluate.

Myths that travel with the theory

Several popular claims are attached to cognitive learning theory without support from its research:

• Learning styles. Preferences exist, but matching teaching to "visual" or "auditory" learners is not supported: Harold Pashler and colleagues concluded in 2008 that "there is no adequate evidence base to justify incorporating learning-styles assessments into general educational practice." The belief is still common: in a 2012 survey by Dekker and colleagues of 242 self-selected teachers, 93% in the UK and 96% in the Netherlands believed it.
• Left-brained and right-brained learners. Some functions are lateralised, but in a study of 1,011 people aged 7 to 29, Nielsen and colleagues found that an individual brain is not left-brained or right-brained "as a global property" (p. 8).
• "We remember 10% of what we read." The percentages often attached to Edgar Dale's cone of experience "were not provided by Dale", as Stice's 2009 refutation shows, and they have no traceable study behind them.
• "People forget 70% within a day." The figure misreads what Hermann Ebbinghaus measured. He tested himself on lists of nonsense syllables and timed how long it took to relearn them. After 24 hours, relearning still took about a third less time than learning the list the first time; after six days it saved about a quarter and after a month about a fifth. That measures one person's relearning effort on meaningless syllables, not the share of work content people forget. A 2015 replication by Murre and Dros again used a single participant. Neither tells you how fast staff forget work procedures.

Using cognitive learning theory with an AI course tool

On 12 October 2026 I briefed Coursebox to build the renewal course. Coursebox is our product, so weigh my account with that in mind. The brief gave the 12-point checklist in its three groups, the three clients' facts and the moves from this guide: a question before any content, chunked teaching sections, the panel beater as a worked example, checks from memory that bring back earlier groups, "How sure are you?" before each answer, a "why" question, the electrical contractor with no hints, and a mixed review to come back to a week later. It asked for about 40 minutes in 11 sections, with the wrap-up last.

After one revision that listed the 11 section titles in order and restated the spacing rules, the outline had 11 sections in the briefed order, with the wrap-up last. On the rendered pages, the opening section asked "What would you check first, and why?" in open text before any content, unmarked, and then showed the renewal timeline. The three teaching sections covered items 1 to 4, 5 to 8 and 9 to 12 in the briefed words. The one-week review came back as an eight-turn scenario, "Mixed Renewal Review", that moves across all three clients and all three groups, with a note to repeat it three weeks after the course. The builder added no learning styles, brain myths, retention percentages or claims that finishing the course makes someone competent.

Then I opened the spaced checks, and they had no questions in them. The check in section 4 told learners to "Respond to each prompt" and to "include at least one question related to items 1 to 4", which passed my design note straight to the learner with no prompts at all. The check in section 5 told learners to "Review items 9 to 12 and relevant earlier group items thoroughly" and then to "respond to the prompts", again with none. That instruction to review first would also have defeated recall from memory. Spacing was in the labels, not in the course.

So I wrote the questions in, using the page editor. The section 4 check now has five questions: four on items 5 to 8 and one that reaches back to the first group, "From the first group: which item found the spray booth, and why does it come before the sums insured?" The section 5 check has six questions mixed across all three groups and out of checklist order, such as "A client has opened a second site. Which item and which group?" Each question asks "How sure are you?" first and shows a model answer afterwards.

Those were not the only page repairs. I also added the missing panel-beater walkthrough, restored the supplied electrical-contractor facts and model file note, and shortened a 250-word “why” task to a few sentences. These edits were checked after reloading, rather than accepted from the outline alone.

What I checked was the outline and the pages I inspected, not a run with account managers. A check that asks about earlier items puts cumulative retrieval into the course; only actual revisits after a gap establish the intended spacing. It does not show that anyone remembers more. The one-week review is a section learners have to come back to a week after they finish.

After the fix, I reloaded the learner view and opened both checks. The questions on earlier groups were there, with the confidence step and model answers, and the old instructions were gone.

Two limitations remained in the inspected course: the mixed-practice answer table appeared before the confidence scenario, and the electrical-contractor task was a reading page without an answer box. The generated outline also estimated about 1 hour 43 minutes against my 40-minute brief; that estimate is not a measured completion time. Before delivery, reorder the answer page and provide a response method. I did not test delayed-release settings.

If you only need questions for content you already have, try our AI quiz generator, and say in the brief which earlier items each check should bring back.

Putting cognitive learning theory to work

Cognitive learning theory says people learn by attending to information, organising it into what they already know and retrieving it when they need it. Its most useful message for training is that the retrieving is part of the learning, not just the test of it. A course that explains the checklist well and checks each module once has done the first half of the work.

Use the theory when learners have to understand and remember a body of knowledge or a procedure, and pair it with other approaches for motivation, demonstration and learning on the job. Treat its evidence as strong on direction and weaker on size, since most of it comes from controlled studies rather than workplaces.

Your next step: take one course you already run and choose its most important procedure. Add one check that asks learners to recall an earlier section from memory, schedule one mixed review a week after the course, and write both into the brief or the plan so they do not drop out in the build.