
Generative Learning in Action is refreshing after the heavy doses of Rosenshine I’ve been consuming recently. There are two aspects to the GL approach I find particularly engaging: it approaches learning from the learner’s perspective rather than that of the instructor (the “flip-side” that the Ensers repeated point out) plus it’s a constructivist theory which insists that learning is mediated through the prior experiences and knowledge of the learner (it’s great to see Piaget referenced these days!). The theory leans heavily on Mayer’s SOI (select-organise-integrate) model of memory which is presented early in the book. It also draws on Cognitive Load Theory and Rosenshine and claims that the eight strategies dovetail nicely with them. None of the strategies should be anything new to experienced teachers but it is welcome to see child-centred approaches championed in the current educational climate.
Here are my notes from reading:
- ▾ Foreward – by Logan Fiorella
- Wants students to understand rather than recall facts.
- “We want them to go beyond the lesson and see its implications for future learning and problem solving. In short, we want to foster generative learning.”
- “Generative learning involves ‘making sense’ of our experience by testing it against what we already know.”
- Students do not or struggle to “generatively learn”: “the default for many is to approach learning as a passive tape recorder rather than as an active sense-maker.”
- ▾ Fiorella and Mayer (2015) identified eight activities that promote understanding:
- summarising
- mapping
- drawing
- self-testing
- self-explaining,
- teaching
- enacting
- “Each activity supports a common set of processes reflected by what we called the select-organize-integrate model (SOI): select key ideas, organize them into a coherent structure, and integrate them with prior knowledge.”
- Fundamental principle: “generative learning depends on the quality of what students generate – the quality of their summaries, explanations, drawings, etc. It depends on generating appropriate relationships that lead to the construction of a coherent, testable, and useful model of how things work and how to solve problems in a given domain – whether it’s Newton’s laws, the human circulatory system, or Shakespeare.”
- “Ultimately, learning depends on what students think about, and what students think about depends on what they already know. If students do not have sufficient background knowledge and instructional guidance to generate meaning from a lesson, the lesson simply won’t make sense. This means, as teachers, we must continually be in touch with what our students know.”
- Generative learning activities serve as assessment tools for teachers.
- ▾ Introduction: from Teaching to Learning
- “Rosenshine’s ‘Principles of Instruction’ provides an excellent series of pointers in how a teacher can ensure that they present information in a way that increases the chance of it being learnt by the pupil. Likewise, the principles of cognitive load theory set out how instruction can be planned in a way to best manage the cognitive load of a task and so avoid overwhelming the limited working memory.”
- “Generative learning considers the learning experience from the point of view not of the teacher, but of the learner. It asks what they should do with the instruction that they have been given to ensure that they are able to truly make sense of it and learn it in a way that allows them to apply it to new situations in the future. We could see generative learning as the reverse side of Rosenshine’s coin.”
- Generative learning falls into constructivist model of learning (learning viewed as something that is shaped by students’ own experiences and prior knowledge).
- Richard Fox – Constructivism Examined (2001)
- Learning can be generated in other ways (than the eight).
- ▾ SCHEMA THEORY
- “A schema (a singular collection of concepts; plural ‘schemata’/’schemas’) is anetwork ofinformation built around connected ideas.” – this relates to how information is stored in long-term memory.
- Not static stores of knowledge: “Schemata are continuously involved in interactionsbetween prior knowledge and new information which we are receiving, selecting and organising before integrating into the long-term memory”
- Schemata can contain inaccurate and false learning that have to be challenged, broken down and rebuilt correctly.
- ▾ SOI MODEL
- Select – Organise – Integrate
- Mayer’s SOI model of memory
- “Fiorella and Mayer describe this form oflearning as ‘a process ofmaking sense, in which you try to understand what is presented by actively selecting relevant pieces of information, mentally organising them, and integrating them with other knowledge you already have’.
- ▾ COGNITIVE LOAD THEORY
- Working memory is limited and can only hold a few pieces of information at any one time.
- Generative learning takes place in working memory.
- “The goal of generative learning is to encode things strongly into our long-term memory and to make them easy to recall in the future”.
- intrinsic load = how complex the task itself is
- extrinsic load = everything else in environment/way task is designed.
- ▾ SELF-REGULATED LEARNERS
- One of the key aims is to develop metacognitive skills… to become independent learners.
- ▾ B.J. Zimmerman – Handbook of Self-Regulation and Learning Performance
- – self-regulating learners personally activate and sustain behaviours that are systematically oriented toward the attainment of learning goals.
- ▾ RESEARCH ORIGINS OF GENERATIVE LEARNING
- “generative learning is based upon the idea that, for learning to take place, students must engage in a number of generative cognitive processes, after which they are able to transfer what they have learnt into solving new problems”.
- Transferable knowledge and skills – Pellegrino and Hilton (2012).
- Piaget and Bartlett (learning as “an act of construction”).
- Katone and Wertheimer (it is how we select, organise, collate and finally integrate this information into new schemata that’s important).
- Merlin C. Wittrock (people tend to generate meanings consistent with prior knowledge).
- ▾ RESEARCH BASE AND EFFECT SIZE
- summarisation = 0.5
- mapping = 0.62
- drawing = 0.4
- imagining = 0.65
- self-testing = 0.62
- self-explaining = 0.61
- teaching = 0.77
- enacting = 0.51
- ▾ 1 – Summarising
- Restate the main ideas of a lesson in one’s own words.
- requires students to “collate and reorganise the main points from their learning at different points in the learning process”
- can be verbal or textual
- most effective where learning isn’t reliant on “spatially complex”
- “This activity means they have to extract the key information, make links and associations within the new material and then make associations with material which is already stored in their existing schemata.”
- supports comprehension by students with lower reading abilities – Bretzing and Kulhavy (1979) and Craik and Lockhart (1972).
- Peper and Meyer: students who summarised notes performed 10-15% better.
- Identifies use in English.
- Use of Cornell notes.
- Gives examples across curriculum.
- Case study: Adam Riches (English) – Cornell notes – explicitly teaches. Marks summary boxes.
- ▾ Limitations: highest effect achieved when time is devoted to its direct teaching; time invested in teaching summaries can outweigh benefits.
- ▾ 2 – Mapping
- Convert a text lesson into a spatial arrangement of connected key words.
- ▾ mind-maps:
- – concept map
- – knowledge map (links have a predetermined type – eg. “this leads to this…”)
- – graphic organisers (eg. compare/contrast matrix; flowchart for cause and effect; hierarchy for classification).
- helps students organise seemingly disparate information into a more logical form
- “The creation of concept maps forces the learner to select information that they feel is relevant, which involves an active engagement with the information they have, prompting them to think hard and therefore to remember it. As well as thinking hard about the selection of the information, they also have to think hard about where to place it in relation to other information on their map. They need to consider how one piece of information on their map links to another or where to place it in a hierarchy.”
- “Concept mapping also allows pupils to combine new information which is to be learnt with what they already know, their prior knowledge. This creates a hook for the new information and makes its place in a schema explicit as well as providing an opportunity to retrieve that which was learnt before, taking advantage of the testing effect.”
- drawback of mapping is that the learner may focus too much on organising information and not enough time on information being learnt – provide pre-filled concept map.
- can also leave gap between new information and completing map – requires thinking hard to select information from memory.
- gives exampls from MFL, DT and RE.
- Case study: Christian Moore (Biology) – using concept maps.
- drawbacks: time required to train students in their use; there’s a risk in using pre-filled maps that students will simply select information to transfer but not consider its role.
- ▾ 3 – Drawing
- Create a drawing to illustrate content of a lesson.
- Meyer suggests that drawing might be: decorative, representational, organisational, explanative. (first two do not generate learning)
- use SOI process and insists that students engage with it.
- Links to dual coding theory.
- Drawing a map of the island in Lord of the Flies is not going to generate a deeper understanding of text. Plottng characters’ movements around island does.
- “One consideration for the classroom is where the learner’s attention is directed during this activity. If ‘memory is the residue of thought’, then we need to ensure they are thinking about the information at hand rather than the process of drawing it. It is too easy to be distracted by the ephemera of the process, such as the colouring pencils used or the level of accuracy in the representation. In addition, the act of drawing places an extraneous load on the learning process if the learner is having to focus on the ‘tedious mechanics of drawing?’ For this reason, it may be advantageous to use drawings where the outline is already provided so that the learner can focus on selecting and organising the relevant information rather than on their skill at drawing.”
- Examples across the curriculum: Science, English (representations of imagery), Art.
- Case study: Ben Newmark (History) – example of cartoon used to illustrate evidence,
- Limitation is frustration of students who feel they can’t draw well. Also, too much time drawing rather than thinking about the information.
- ▾ 4 – Imagining
- Form internal images to illustrate the content of a lesson.
- Students are asked to create: static images, steps in a process, animated sequences.
- Students do need some prior knowledge of the topic in order to be most successful.
- Memorise key facts with images, landscapes, stories, varied details.
- “The research indicates that imagining can have specific relevance to comprehension in reading, with visualisation a frequently used strategy in theteaching of reading. In research conducted by Pressley in 1976, it was found that learners who were given instruction to picture elements of a 950-word story in their head significantly outperformed the control group in a subsequent comprehension test.”
- Needs explicit instruction to achieve highest effects. Also supporting students in selecting and organising information.
- Gives example of A Christmas Carol opening. (I understand this as creating additional characters/settings etc based on knowledge of text. Also suggests looking at different pictoral representations of Scrooge.)
- Gives examples from PE, History and Maths.
- Case study: Tim Taylor (“freelance teacher”) refers to Kieran Egan’s Teaching as Story telling and Dorothy Heathcote (use of drama in classrooms).
- Limitations: some students with low prior attainment or developmental delay may not be able to sustain mental images, Imagining can place extra load on working memory. Students need to have good motivation for this strategy as no concrete outcome.
- ▾ 5 – Self-Testing
- Self-testing of previously-studied content by answering practice quesions.
- “Self-testing in generative learning is the process in which students recall information from a learning episode, using questions or activities which require them to retrieve either specific detail or broader recollections, such as with ‘brain dump’ activities. So, for example, students read over a chapter of a science revision guide and then complete the questions at the end of the chapter and use the materials to check their understanding.”
- Allen, Mahler and Estes (1969) found practiced retrieval improved retention.
- Roediger ans Karpicke (2006) self-testing and low-stakes quizzing can have significant impact on memory and learning process.
- Positive impact of “the testing effect” (Edwina Abbot).
- Rohrer and Taylor (2007) found that students who revised material in a distributed way scored significantly higher than those who crammed revision into short time period.
- Argarwel and Roediger (2011) – students in closed book test conditions were likely to perform better as they put more effort into initial study of material. “Further research from Roediger et al. found myriad benefits of testing, including improved organisation ofmaterial in the minds of the students and a better awareness of gaps in knowledge and metacognitive processes, all central to the ideas of generative learning.”
- Ask students to write down what they have learned at the end of a topic. Short-answer quizzes, flashcards, online platforms.
- Students do need quick access to corrective feedback.
- “We can also utilise the testing effect by employing spaced practice strategies. Giving students a chance to retrieve information at key points, allowing time for forgetting to take place, will increase the storage and retrieval strength.Therefore, plan for opportunities for self-testing on previous topics.”
- Curriculum examples given for MFL, History and Drama.
- Case Study: Emma Smith (History) and Mark Enser (Geography) – raised issue of the complexity of questioning.
- Limitations: students need good-quality materials in initial learning episode; multiple-choice questions approached cautiously. Students need motivation not to use learning materials (becomes copying).
- ▾ 6 – Self-Explaining
- Explaining the content of a lesson by elaborating on material covered.
- Involves explaining a text or diagram to themselves.
- To generate learning then this technique should go beyond simple comprehension of the text.
- Aim is to develop more independent learners who can use self-explaining to interrogate something new without direct input from a teacher.
- ▾ Needs to go further and ask learner how they arrived at the answer they reached. Use Socratic questioning:
- Classify their thinking – ‘What do you already know about this topic?’
- Probe assumptions – ‘What additional evidence would lead to youreaching a different answer to this question?’
- Demand evidence – ‘What evidence have you got for the conclusion that you have reached?’
- Alternative viewpoints – ‘Who would disagree with the conclusion you have reached?’
- Explore implications – ’What are the implications for your conclusion? What would need to happen or change?’
- Question the question – ‘Why do you think this was an important question to ask?’
- Fiorella and Mayer found it better to give a learner a menu of options for explanation to select from rather than an open choice (avoids misconceptions).
- Gives examples: Maths, Science, RE.
- Case study: Ceridwen Eccles (primary) – self-explanation during reading sessions.
- Limitations: time to fully train learners in using technique; time taken for self-explanations rather than other forms of learning; some contradictions in research.
- ▾ 7 – Teaching
- Teaching other students about previously-studied material.
- Lower evidence base for this approach.
- ▾ Bargh and Schul (1980) found three phases of teaching:
- – preparation stage (S)
- – act of engaging encourages students to actively engage in the materials (S)
- – deep questioning by tutor encourages metacognitive processes (advises use of Socratic questioning).
- Students who taught materials significantly outperformed others (including those who prepared but didn’t actually teach).
- Gives suggestions for teaching poetry and Shakespeare (need high-quality study materials).
- Simple form = think, pair, share
- Research suggests that well-designed peer tutoring programmes can have significant impacts for tutor and tutee.
- Examples given for Drama, Maths and Art.
- Case Study: Freya Odell (English) – uses a “jigsaw” method to give structure to how students teach each other newly learnt material. Gives the example of different groups studying an aspect of a poem then teaching to others. Teacher uses discussion at end of activity.
- Limitations: potential to embed false conceptions; not all students able to construct effective explanations of some topics; questions and follow-up interactions have potential to misdirect; stress and anxiety caused by task.
- ▾ 8 – Enacting
- Engaging in task-related movements during learning.
- Involves students making gestures or manipulating objects linked to the thing they are learning.
- “One thing that enacting does is to make the abstract more concrete in the mind of the learner.”
- Enactment helps younger children (2-7).
- Sweller and Paas suggest that use of gestures might help offset some of the cognitive load of the task (CLT).
- Fiorella and Mayer find evidence base for enacting weaker than others.
- Enacting approaches “only really apply to younger learners who struggle with moving from the concrete to the abstract.”
- Gives examples from: Music, MFL (use gesture when repeating key phrase or vocabulary), Maths.
- Case study: Tarjinder Gill (primary) – use of gestures to support recall of key vocabulary and storytelling (for sequencing of story or process).
- Limitations: really for younger children only; poor evidence base; make sure enacting doesn’t become a distraction from the thing being learned; danger that actions create “primarily episodic memories rather than semantic ones”.
- ▾ Conclusion: What Have We Learnt?
- “It has also become apparent to us that generative learning fits well alongside theories of effective instruction.”
- Instruction “of new information needs to be crystal clear and deploy the kind of strategies of instruction given by Rosenshine, with opportunities for retrieval, modelling, practice after small steps and regular reviews.”
- “It should not be seen as a replacement for effective instruction but provide guidance to what pupils do after effective instruction.”
- Pitfalls: remove instruction and rely on generative learning. Also: strategies require students to be trained in strategies differently in each subject. Need to be used consistently and frequently.
- Important to ask: is the time invested in teaching generative learning approaches worth it in the context of the whole-school?
- GL “dovetails neatly” with effective instruction strategies such as Rosenshine’s and”closely aligns” with cognitive load and schema theory.
- David Weston and Bridget Clay – Unleashing Great Teaching.
- David Kolb – experiential learning cycle.
- Strategies can be used for revision.
- Implications for remote learning (writers don’t feel this is possible or, at least, desired).
- Final implication: “Generative learning strategies are based on the SOI model and integration with prior knowledge is key. For these strategies to be successful, it is critically important that the curriculum is well sequenced so that there are explicit links made to what pupils have already learnt and they are made in a way that allows pupils to make the connections to what they are now learning.”