Generative Learning: Strategies That Make Knowledge Stick

Teacher explaining generative learning to pupils in a UK classroom

Updated on

June 5, 2026

Generative Learning: Strategies That Make Knowledge Stick

Q: How can I start implementing generative learning in my classroom tomorrow?

Begin with simple techniques like having students summarise lessons in their own words or create concept maps linking new vocabulary to familiar concepts. You can then progress to more complex activities such as having students generate their own questions about the material or teach concepts to the

Q: What are the main challenges teachers face when using generative learning strategies?

Learners need prerequisite skills like good writing for summaries. (Brown et al., 1983) Strategies vary; what helps one learner may not help another. (Kirschner, Sweller & Clark, 2006) Teachers must adapt their methods. Some learners need extra help with generative learning. (Mayer, 2004)

Q: How do I know if generative learning strategies are working for my students?

Comprehension test scores should improve. Learners will engage more during class discussions. They can also connect new information to old (Bransford et al., 2000). Encourage regular self-testing to boost metacognition and knowledge retention (Roediger & Karpicke, 2006).

Explore generative learning strategies that engage students as active creators. Implement techniques like summarizing, mapping, and self-explaining to.

Build your next lesson free Explore the toolkit

Copy citation

Main, P (2021, August 31). Generative Learning: A teacher's guide. Retrieved from https://www.structural-learning.com/post/generative-learning-a-teachers-guide

In this article

Generative Learning: Strategies That Make Knowledge Stick describes teaching approaches where learners actively select, organise and connect new information with what they already know. This makes learning active, not just a matter of listening or copying.

The idea is citable because it has a long base in cognitive science. This runs from Wittrock's generative theory to Fiorella and Mayer's work on summarising, mapping, drawing, self-testing and self-explaining (Fiorella & Mayer, 2015).

For a busy UK teacher, the point is practical. After a Year 8 science explanation on photosynthesis, learners might close their books and sketch the process. They can label the links between chlorophyll, light and glucose, then explain one link to a partner. The activity works only when learners do the thinking themselves, with enough modelling to avoid working memory overload.

What is Generative Learning?

Generative learning supports deep understanding when learners build meaning, not when they copy a finished explanation. Learners make meaning actively. They do not just receive information. Use it as a starting point for professional discussion: identify the learner's current need, record evidence from more than one lesson, and agree the next classroom adjustment with the SENCO or family.

Summarising, concept mapping, questioning and explaining can help knowledge last. These strategies work when learners choose key information, organise it and link it to what they already know (Chi, 2009; Mayer, 2003). This guide presents generative learning strategies for your classroom (Fiorella & Mayer, 2015).

‍

Key Takeaways

Generative learning fundamentally transforms how learners acquire knowledge, moving beyond rote memorisation to encourage deeper understanding. This approach, championed by researchers like Wittrock, posits that learning is most effective when learners actively construct meaning by relating new information to existing knowledge (Wittrock, 1989). Encouraging learners to summarise or explain concepts in their own words significantly enhances knowledge retention and transfer.
Implementing specific generative strategies demonstrably improves learning outcomes, far surpassing passive reception of material. Research consistently shows that activities like self-explanation, summarising, and concept mapping lead to more strong and durable learning than simply reading or listening (Mayer, 2009). For instance, learners who actively generate summaries often recall significantly more information than those who merely reread.
Effective generative learning tasks guide learners through important cognitive processes of selecting, organising, and integrating new information. This three-stage model, often referred to as the SOI framework, highlights how learners must first select relevant information, then organise it into a coherent mental structure, and finally integrate it with their prior knowledge (Mayer & Wittrock, 1996). Designing tasks that explicitly prompt these stages is vital for deep learning.
Teachers can readily integrate simple yet powerful generative activities into daily lessons to significantly enhance learner engagement and understanding. Strategies such as asking learners to generate questions about a text, create analogies, or teach a concept to a peer actively engage their cognitive processes, leading to stronger memory traces (Brown, Roediger, & McDaniel, 2014). These low-preparation, high-impact methods are important for encouraging active knowledge construction in the classroom.

‍

Feature	summarising	Mind Mapping	Peer Teaching
Best For	Processing complex texts and identifying main ideas	Visual learners connecting concepts and vocabulary	Deepening understanding through explanation
Key Strength	Improved comprehension and retention	Links new knowledge to existing understanding	Activates knowledge through teaching others
Limitation	Requires strong writing skills	May be challenging for non-visual learners	Needs confident learners and time
Age Range	Upper primary to secondary	All ages with adaptation	Middle primary to secondary

This process, described by Marton and Säljö (1976), builds meaningful connections. Learners achieve deeper understanding when they link new facts with prior knowledge. Entwistle (1988) also found this integration improves a learner's grasp of concepts.

Flow diagram showing three connected stages: Select, Organise, Integrate leading to knowledge creation — Flow diagram: The SOI Framework: Three Stages of Generative Learning

Key to this theory is the notion of the 'generative process', which involves the cognitive work of organising and integrating information during the learning process. This is not an abstract idea. It is classroom practice: learners explain, organise or represent the content for themselves.

Generative learning works well. Teachers can ask learners to map new words to known ones (Wittrock, 1974). This connects new ideas to old knowledge, aiding deeper understanding (Wittrock, 1990; Fiorella & Mayer, 2015). Mapping helps learners summarise too.

‍

Infographic comparing passive and generative learning. Passive involves receiving info and rote memorisation with limited retention. Generative focuses on creating understanding, deep processing, and enhanced comprehension for durable knowledge. — Passive vs. Generative

‍

Generative strategies can improve comprehension, but the task must require thinking. A copied spider diagram is not generative; a map that explains why two ideas connect is.

Educational psychologists describe learning as active construction. This means learners build understanding rather than simply receive information. In generative learning, learners retrieve prior knowledge, organise new material and test whether the two fit together.

Ausubel (1968) said learners build understanding actively. They do not just take in information. Ausubel stated that prior knowledge forms the basis for all new learning.

Generative Learning Theory also recognises differences among learners. Not all learners use the same strategy or learn at the same pace. Novices, EAL learners and neurodivergent learners may need extra support first. This may include modelling, sentence stems, worked examples or partially completed maps before they can generate useful explanations.

Generative models are not universal fixes. They are tools teachers adapt according to prior knowledge, language demands and working memory load.

Research shows that learners gain more when they take an active role (Wittrock, 1974). Generative Learning Theory means learners turn new information into knowledge that lasts. This reminds teachers that learner effort improves learning outcomes.

‍

Core Generative Learning Strategies

Fiorella and Mayer identified eight core strategies for generative learning: summarising, mapping, drawing, imagining, self-testing, self-explaining, teaching and enacting (Fiorella & Mayer, 2015; Fiorella & Mayer, 2016). Learners can summarise a lesson and then teach it to classmates (Chi et al., 1989; King, 1993). Practise questions also help, as Rosenshine (2012) argued in his principles of instruction. Active learning can support learning when teachers check its impact, a point consistent with Hattie (2009).

Generative learning improves learning when the task requires learners to connect new content with prior knowledge (Wittrock, 1974; Fiorella & Mayer, 2015). Use these classroom moves as a menu, not a checklist. The important test is the quality of the learner's explanation, map, question or worked example.

Elucidation of Key Concepts: Clearly explain key concepts and ask learners to explain these concepts in their own words. This active reformulation helps embed new learning into existing knowledge structures.
Example-Based Learning: Use concrete examples that relate to learners' experiences, enhancing comprehension and recall.
Mind Mapping: Ask learners to create a spider diagram or similar graphic organisers to visually represent and connect ideas.
Peer Teaching: Learners who teach others can strengthen their understanding and reveal gaps in their knowledge. It also creates empathy and collaboration.
Integration of New with Old: Regularly connect new content with previously learned material to reinforce the relevance and application of knowledge.
Question-Generating: Ask learners to generate their own questions about the material, developing curiosity and critical thinking.
Conditional Learning: Contextualise learning within real-life scenarios or potential future applications to stimulate interest and engagement.
Diffusion Model: As a head of department, support generative learning across the curriculum and check the quality of learner outputs.
Self-Testing: Ask learners to regularly check their understanding, promoting metacognition and aiding retention.

Brown (1987) described metacognition as learner self-regulation. In simple terms, learners check and guide their own thinking. Brown, Roediger, and McDaniel (2014) later argued that self-questioning helps learners monitor what they know.

Learners who create their own science questions showed better engagement. Active self-questioning also supports stronger understanding (Brown, Roediger & McDaniel, 2014).

Tailor strategies to each learner's prior knowledge, writing fluency and confidence. A good outcome is not the presence of a concept map, but whether the learner can explain the relationship between the ideas on it.

Generative strategies help learners build deeper understanding (Fiorella & Mayer, 2015; Chi & Wylie, 2014). Learners link new information to what they already know. This improves memory (Wittrock, 1974).

This approach also builds critical thinking and problem-solving skills (Osborne & Wittrock, 1983; Mayer, 2002). Through this process, learners become more self-directed (Zimmerman, 2002).

‍

Generative Learning Theory: History and Origins

Wittrock (1970s, 1980s) made generative learning from constructivism. Research shows learners retain knowledge when they make connections. Cognitive science backs Wittrock's work as key to education. Use it as a starting point for professional discussion: identify the learner's current need, record evidence from more than one lesson, and agree the next classroom adjustment with the SENCO or family.

A learner's existing schema includes their thinking and past experiences. Wittrock noted that learners actively make links with stimuli, or things they see, hear or read. This process links new information to memory.

For learning to last, learners need to connect the new concept with what they already know. Generative learning makes this connection explicit instead of leaving learners to join the dots alone.

model of generative learning theory — Model of generative learning theory

‍

Three Stages of Generative Learning Process

Generative learning uses three steps: Select, Organise and Integrate. First, learners select the key information (Wittrock, 1990). Next, they organise it into mental representations, which are mental models of the idea (Fiorella & Mayer, 2015). Lastly, they integrate new knowledge with what they already know (Chi, 2009).

The SOI model proposed by Logan Fiorella and Richard Mayer suggests that learners generate learning from new information in three stages. This model gives schools a practical way to check whether learners are selecting, organising and integrating knowledge. When learners build a model with blocks, teachers can see the structure of their thinking rather than infer it from a completed worksheet.

‍This approach can help learners tackle abstract ideas such as the correct use of an adverb. In one anecdotal case from our practitioner network, an English teacher used blocks to teach key grammatical concepts. The next research task is to test whether these visible models improve conceptual knowledge across subjects, not simply whether learners enjoy the activity. The three stages are as follows:

At first, people focus on selecting particular information from what they have heard, seen or read.
Secondly, they organise the details in their active memory. This may include transforming it into a new kind or structuring the details so that it helps them solve a problem or answer a question.
Thirdly they incorporate the new information into the pre-existing schema that enables this prior knowledge to instruct their thoughts about this new knowledge and assure that the new knowledge can incorporate into their pre-existing knowledge or modify their prior knowledge so that the new knowledge can be adapted.

‍

Colorful building blocks arranged to demonstrate the SOI model for generative learning strategies — Using the SOI model with the blocks

‍

Four Core Principles of Generative Learning

Learners actively process information to build understanding (Bereiter & Scardamalia, 1985). They remember new information more easily when they connect it to prior knowledge (Ausubel, 1968). Learners should also use metacognition, or thinking about their own thinking, to check their comprehension (Flavell, 1979).

Generative Learning Theory includes four main processes that teachers can build into lessons. Teachers can choose one process at a time, according to learners' prior knowledge and the learning resources involved.

Recall takes place when the learners access information that already exists in their long-term memory. The main aim is to help learners build a concept from information they already know. An example of recalling strategies might be having a person review information or repeating it until the concept is fully understood.
Integration takes place when the learners add new details into the knowledge they already possess. The main objective is to modify information to make it more accessible and easy to remember. An example of this learning activity can be establishing analogies to define a concept or asking a learner to paraphrase the text.
organisation takes place when learners connect their pre-existing knowledge with new concepts effectively that helps the learners remember. An example of organisation strategies may include generating lists and grading individual items or evaluating the main parts of a concept.
Elaboration occurs when the learners are asked to connect new concepts with the knowledge they’ve already acquired in creative ways. An example of elaboration strategies is imagining how the new knowledge matches pre-existing knowledge or daily work.

‍

Advanced Generative Learning Activities and Techniques

King (1990) showed that elaborative interrogation ("why" questions) helps learners. Wittrock (1989) found that diagrams, test questions, and real applications help learning. Reciprocal teaching can work when learners have enough structure. Mayer (2009) showed that multimedia presentations can engage learners, while Chi (2009) argued that learners should transform information, not just repeat it.

Fiorella and Mayer used the SOI model to study classroom activities that help learners generate meaning. In the SOI model, learners select information, organise it and integrate it with what they know. Their Cambridge University Press account identifies eight strategies with strong generative potential:

summarising: Learning through summarising requires learners to pick the main ideas, organising these ideas into a logical pattern, and incorporating new knowledge with pre-existing knowledge.
Mapping: This strategy includes a variety of techniques, such as graphic organisers, knowledge maps and concept maps. It is a generative strategy because learners pick important words that indicate the main ideas, organise these ideas by establishing links between them, and incorporate new knowledge with pre-existing knowledge by specifying the overall pattern of the map.
Drawing: Learners provide a pictorial representation of the text while using drawing as a learning strategy. Drawing is a generative process because it involves selecting related ideas from the text, organising the concepts in pictorial form, and making use of pre-existing knowledge to demonstrate the meaning of the ideas in the drawing.
Imagining: Learners build mental impressions of the topic to be learned while using imagining as a learning strategy. Just like drawing, imagining is also a generative strategy.
Self-testing: Learners choose the most relevant information during self-testing or retrieval-based learning, followed by organising and incorporating knowledge by making connections between new and old information.
Self-explaining: In the self-explaining technique, learners explain the details of the lesson to themselves. Self-explaining is a generative strategy because learners determine the most relevant knowledge, explain the details in their own words, organise the knowledge by making inferences, and incorporate information with pre-existing knowledge during the explanations.
Teaching: The main purpose of teaching is to help others in learning. Teaching is somewhat close to self-explaining, but it sets itself apart depending upon the recipient of the teachings.
Enacting: When learners enact, they perform gestures or manipulate things relevant to the knowledge to be learned. It is also a generative learning technique because learners choose the actions to perform, organise the details through the actions, and incorporate new knowledge with pre-existing knowledge during the process.

‍

Summary of generative learning activities

These activities are common in many classrooms, but the name of the task is not enough. A trust or department should check the quality of what learners produce: does the summary name the main ideas, does the map show links rather than copied words, and does the explanation reveal a misconception? Fiorella and Mayer's work suggests that these activities support learning only when learners select, organise and integrate content through the SOI model.

Teachers can use mind maps in class and ask learners to turn information into a spider diagram. They might then use the notes for a later task. On its own, the mind map may do little more than record words; it becomes generative only when learners explain the links.

using graphic organisers as a generative learning tool

To make the mind map generative, learners need to use the SOI model. First, they need a clear goal. Next, they select only the information that matters, group it into categories and explain how their prior knowledge relates to the new details on the map.

Learners actively engaging with study materials through note-taking and discussion — Generative learning is an active process

‍

Kolb's Learning Cycle and Generative Learning

Kolb (1984) argued that experience, reflection, conceptualisation and testing form a learning cycle that can support generative techniques. Learners create understanding through experience and reflection. The four stages prompt learners to think, use ideas and apply them. They do not just take information in, so this approach builds deep learning.

In 1984, David Kolb presented a model to explain the process of learning from experience. According to this model, people go through four stages while learning from experience:

David Kolb suggests that for effective learning, the learner needs to progress through the cycle. Also, the learner can embark on the cycle at any one of the four stages of the cycle with logical progression.

David Kolb suggested that while learning from experience, people must pass through four stages. They can start from the theory of why something could work, and then they can propose a plan for using it in any specific context. Also, they can get the experience of doing it in reality before revealing whether it performed according to the expectation or they had to make any adjustments.

‍

Designing Effective Generative Learning Tasks for Learners

Generative learning needs learners to change information by explaining or applying it. Identify key concepts learners need. Design tasks where learners actively summarise, compare, or problem-solve. Link new content to prior knowledge and provide clear success criteria. (Wittrock, 1974; Chi et al., 1989; Mayer, 2002).

Generative learning may already happen in your school. Leaders should treat it as an implementation problem, not a CPD slogan. Sample learner outputs and ask whether the task improved explanation, comparison or transfer. Do not just check whether a map or summary was present.

Generative learning helps teachers think carefully about the learning experience. Learners need to make sense of the material and build on it, but novices can feel overloaded if the task is too open. Use worked examples, partly completed maps and modelled explanations before asking learners to generate ideas on their own.

Mental models have to be constructed carefully by learners. Retrieval practice can help here: Roelle et al. (2023) suggest that generative follow-up tasks can strengthen mental models when learners rebuild and refine what they have retrieved.

The generative model puts learner understanding first. Even so, results from controlled studies do not transfer automatically to mixed-ability classrooms. Treat Fiorella and Mayer's strategies as design principles. Test them through learner work samples, mini-quizzes and the quality of learners' explanations.

Our mental modelling strategy makes learning visible. Teachers report seeing learner differences more clearly when learners use blocks (internal Structural Learning case study, University of Bedfordshire partnership; not peer-reviewed).

Learner builds showed that learners approached curriculum concepts in different ways. Each learner understood the material in their own way. These differences became clearer when the material was complex (Vygotsky, 1978; Piaget, 1936; Bruner, 1966).

The universal thinking framework also has the generative theory at its core. The key message when using this new taxonomy is that declarative concepts have to be built. Knowledge has to be constructed meaningfully using cognitive actions. Key concepts do not just arrive in the learner's mind; combining the block building strategy with the framework helps classrooms bring structure to the learning process.

References

Caviglioli, O. (2018). Understanding How We Learn: A Visual Guide. Routledge.
Fiorella, L., & Mayer, R. E. (2015). Learning as a generative activity. Cambridge University Press.
Fiorella, L., & Mayer, R. E. (2016). Eight ways to promote generative learning. Educational Psychology Review, 28(4), 717-741.
Kolb, D. A. (1984). Experience as the source of learning and development. Upper Sadle River: Prentice Hall.

‍

Written by the Structural Learning Research Team

Reviewed by Paul Main, Founder & Educational Consultant at Structural Learning

‍

AI-Enhanced Generative Learning in Modern Classrooms

AI-assisted learning tools now help teachers use adaptive questioning. These systems change their questions in response to each learner's understanding. They do not replace traditional generative strategies. Instead, they support cognitive scaffolding by changing question difficulty and giving feedback while learners build knowledge.

The DfE's 2024 guidance on AI in education supports this kind of human-AI collaboration. It does so when AI strengthens critical thinking rather than replacing it.

Consider a Year 9 science teacher introducing photosynthesis with an AI-powered platform. The platform creates a personalised pathway for each learner. As learners explain the process in their own words, the algorithmic system spots misconceptions. It then asks deeper questions: "You mentioned chlorophyll absorbs light, which wavelengths specifically, and why does this matter for the plant's survival?"

This immediate and tailored response moves learners beyond surface-level summaries. It helps them construct knowledge in a more meaningful way.

Research shows that AI-enhanced generative activities can improve retention when teachers design them around established SOI principles (Zawacki-Richter et al., 2023). ChatGPT is a generative AI tool. In this context, it should act as a questioning partner, not as the author of the learner's answer.

Its effectiveness depends on teacher orchestration. AI may generate prompts, but teachers model the thinking process and validate the learning outcome.

Successful implementation requires clear boundaries around AI use. Make tasks AI-resistant by asking learners to annotate their own draft, explain a mistake, defend a link in a concept map or compare their answer with a model. Learners must understand that AI assists thinking rather than replacing it, with teachers maintaining control over learning objectives and assessment criteria. This balanced approach keeps the cognitive work in the learner's mind while still using AI for prompts, feedback and comparison.

‍

Frequently Asked Questions

‍

What is generative learning and how does it differ from traditional teaching methods?

Generative learning helps learners understand concepts actively. Learners summarise and explain ideas in their own words. Research shows better comprehension (Wittrock, 1974; Fiorella & Mayer, 2015).

Learners using these strategies also perform better on tests (King, 1992; Rosenshine, Meister & Chapman, 1996).

‍

Which generative learning strategies are most effective for different age groups?

Mind mapping suits all ages with adjustments. Peer teaching works best for learners in middle primary to secondary years who can explain things (Topping, 2005). Summarising helps older primary and secondary learners write about complicated texts (Marzano et al., 2001).

‍

How can I start implementing generative learning in my classroom tomorrow?

Begin with simple techniques such as asking learners to summarise a lesson in their own words or create a concept map linking new vocabulary to familiar concepts. You can then move to more complex activities, such as asking learners to generate questions about the material or teach a concept to classmates.

‍

What are the main challenges teachers face when using generative learning strategies?

Learners need prerequisite skills before they can organise new material. These include vocabulary knowledge, writing fluency and enough prior knowledge. Kirschner (2006), with Sweller and Clark, warned that minimally guided tasks can overload novices. So teachers should model the strategy, reduce the language demand and then remove support gradually.

‍

How do I know if generative learning strategies are working for my learners?

Comprehension test scores should improve. Learners should also take a fuller part in class discussions. They can connect new information to old knowledge (Bransford et al., 2000). Encourage regular self-testing to boost metacognition, or thinking about learning, and knowledge retention (Roediger & Karpicke, 2006).

‍

Can you give me a specific classroom example of generative learning in action?

Concept maps help learners link new vocabulary to things they know. Self-generated questions boost engagement and understanding in science by up to 50% (King, 1992). Learners actively build their knowledge through this process.

‍

How does generative learning support learners with different needs?

Generative learning gives learners several ways to understand ideas. Some use mind maps well. Others need oral rehearsal, sentence stems or paired explanation before they write on their own (Wittrock, 1974; King, 1993; Slavin, 1996; Fiorella & Mayer, 2015).

‍

References

Brown, A. (1987). Metacognition, executive control, self-regulation, and other more mysterious mechanisms.

Hattie, J. (2009). Visible learning.

Kirschner, P. (2006). Why minimal guidance during instruction does not work.

Kolb, D. (1984). Experiential learning.

Rosenshine, B. (2012). Principles of instruction.

‍

Limitations and Critiques

Any article on Generative Learning: Strategies That Make Knowledge Stick should separate classroom guidance from clinical, legal or policy decisions. Teacher observations are valuable, but they can be shaped by task design, school context, relative age, masking and co-occurring needs. A checklist, strategy or theory should not be treated as proof on its own.

Current research also leaves some uncertainty. As we established in our previous exchanges, the provided sources in this notebook do not contain the article "Generative Learning: Strategies That Make Knowledge Stick." They also do not explore generative learning in enough depth to give scholarly critiques or practical limitations for UK classrooms.

The only references to generative learning in the dossier highlight its benefits. They note that combining it with retrieval practice through follow-up tasks helps learners build stronger mental models (Fiorella & Mayer, 2016; Roelle et al., 2023). Because the specific critiques, limitations, and citations you are looking for are entirely absent from your current materials, I cannot answer this query using the provided notebook.

Would you like me to find this information for you? Please let me know if you would like me to initiate a fast research or deep research search on the public web to find the strongest scholarly critiques and practical limitations of applying generative learning in the classroom. Teachers should state these limits clearly. They should also record what they can observe and involve the right specialist when decisions move beyond classroom practice.

About the Author

Paul Main

Founder & Metacognition Researcher

Paul Main is an educator and metacognition researcher who founded Structural Learning in 2002. With a psychology degree from the University of Sunderland and 22+ years helping schools embed thinking skills, he bridges the gap between educational research and classroom practice. Fellow of the RSA and Chartered College of Teaching, with 128+ Google Scholar citations.