Miller - Magic number 7 and chunking... referred to himself as being ‘plagued by an integer’!

George Miller is one of the founders of cognitive psychology, which was to eclipse behaviourism. He is best known for his paper ‘The magic number 7 plus and minus 2’ (1956) which focused attention (literally) on a problem that plagues teaching and learning, the limits of working memory and the danger of ‘cognitive overload’. Although he felt as though this work had been overvalued and he jokingly referred to himself as being ‘plagued by an integer’!

Miller was one of the first to bring in cognitive information theory, famously saying the mind had been brought back into psychology. He also contributed to the then fledgling subject of psycholinguistics, working with Chomsky, among others.

Chunks

Miller started by identifying relevant studies that show we remember ‘chunks’ of information. But what is a ‘chunk’? 

Chunking depends on what is being chunked. We can remember, on average, seven randomised numbers but only sixletters and four or five words. Miller’s solution was to posit clusters of chunks e.g. a word is a cluster of letters, a phrase a cluster of words. But Miller’s definition proved inadequate. This problem, that we remember different types of content differently giving variable sizes of chunks, was to open up a whole field of experimental work that began to reveal how short-term, or working, memory works. The concept of a ‘chunk’ has proved enduring and useful in memory research.

Encoding methods

Research by Bower (1970, 1972) showed that meaning also played a role in chunking so that, for example, TVFBIJFKCIAIBM is far more difficult to remember than TV FBI JFK CIA IBM. Mnemonics, as chunks, can also be used to unpack larger amounts of information. Gardener (1988) then showed the power of the ‘generative effort’, where creating your own chunks, is even more effective.

Of course, chunking is just one form of encoding that increases retention, the organisation of material is another, chunking categories logically and with relatively few numbers of items in each category, increases recall, as do outlines. Imagery is another useful form of encoding, which is why loci and peg-word systems have been used for centuries to improve memory. In particular, movement and interaction between images so that they do things to each other, were shown by Wollen, Weber & Lowry (1972) to be particularly effective. Craig & Lockhart (1972) popularised the word ‘elaboration’ for techniques such as paraphrasing, summarising, highlighting and note taking, all of which improve recall.

Miller’s work also led to research on the time we are able to hold attention in short-term memory. Waugh and Norman (1965) found, in recall experiments, that this was around 18 seconds. However, if recognition, and not recall, is used this can stretch out to 90 seconds (Lutz & Wuencsch 1989).

Online learning and chunks

Online learning, with its learning objects, microlearning and page-based delivery has long sought to chunk material to make it more digestible and to increase encoding and retention. Chunked media objects such as short videos, animations, examples and so on, have also been part of the online learning landscape. Attempts have even been made to determine standards around reusable learning objects that can be re-assembled in different orders and in different contexts and courses, like LEGO. The problem is the conflict between usability and the flow and context of the learning experience on one hand, and the fragmented nature of chunked learning objects on the other. This has also led to standards such as SCORM, that some believe traps designers in to producing learning objects and does not cope with more complex entities such as simulation and games pedagogy.

Influence

Miller, a pioneer in the information processing or computational model of memory, set us on the path of defining the limits of short-term memory, the gate through which attentive learning must pass. It proved to be limited in time, capacity and the way we encode information. This has opened up research on cognitive overload and a deeper exploration of what we now call ‘working’ memory and encoding through elaboration. The magic number ‘7’ may have proved misleading, as our short-term register often operates at a lower 3-4 register level, but the principle of cognitive overload remains a limit on learning.

Bibliography

Miller, G.A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63, 81-97. 

Miller, G.A., Galanter, E., & Pribram, K.H. (1960) Plans and the Structure of Behavior. New York: Holt, Rinehart & Winston.

Miller G Psychology Penguin

Bower G. H. (1970) Organisational factors in memory Cognitive Psychology, 1, 18-46

Bower G. H. (1972) Mental imagery in associative learning in Gregg L,W. Cognition in learning and memory New York, Wiley

Gardener (1988) Generation and priming effects in word fragment completion Journal of Experimental Psychology: Learning, Memory and Cognition 14, 495-501

Wollen, Weber & Lowry (1972) Bizarreness versus interaction of mental images as determinants of learning Cognitive Psychology, 3, 518-523

Craig F. I. M. & Lockhart R.S. (1972) Levels of processing: A framework for memory research, Journal of Verbal learning and Behaviour, 11, 671-684.

Waugh N. C. and Norman D. A. (1965) Primary Memory, Psychological Review, 72, 89-104.(1965)

Lutz J. & Wuencsch K.L. (1989) Acoustic interference in a recognition task. The Journal of General Psychology, 116(4), 371-384

Study on retention using Video plus AI-generated retrieval practice

Abstract

The aim of this trial was to test the effectiveness of chunking video and placing effortful, retrieval practice after each chunk of video. Chunking is the slicing of video content down into several, separate video segments or chuinks, so that there is less cognitive load and forgetting. Retrieval practice is making the learner recall what they think they know to reinforce therefore increasing retention and subsequent recall. Two groups were compared. One was shown only a training video on Equality & Diversity produced for a large company, the other the same video, chunked into smaller segments, with AI generated practice at the end of each short segment. Both groups were tested immediately after the learning experience. The results showed a 61.5% greater score in the Video + AI generated practice group over the Video only group. This study shows that video significantly benefits from enhancement through chunking, reinforcement, retention and recall by adding Video plus AI-generated retrieval practice.

Introduction

Video has become commonplace in learning, through YouTube and Vimeo in both the public domain and on private channels. It has also become common to deliver learning video content from a VLE (Virtual Learning Environment), LMS (Learning Management System) or LXP (Learning eXperience Platform). Other video specific platforms use Netflix-style carousel and other interfaces to deliver learning video content.

Yet little attention has been paid to the research that suggests video should be enhanced with active learning. Research into the use of video for learning recommends several techniques to enhance the watching of video on its own, (Reeves & Nass, 1996; Zhang, 2006; Mayer, 2008: Brame, 2016; Chaohua, 2019).

Method

Twenty-six participants were selected. The first group of thirteen watched the video only. The second group of thirteen watched the same video but chunked down into four meaningful segments, edited to match separate topics, interspersed with AI-generated, retrieval practice. The AI-generated, retrieval practice group required the learner to recall key ideas and concepts and type them in. This involved acts of recall and writing created by the AI tool, that reinforce learning, where the learner was required to recall concepts as well as type in those concepts. Any items that were not correct had to be repeatedly input until all were correct. A separate and identical written, recall test was completed immediately after the learning experience for both groups.

Note that the retrieval practice tool used was WildFire. It creates online learning from the chunks of video, applying AI to the automatically generated,video transcript , using the AI to identify the key learning points, create questions, as well as generate links to external content to enhance the learning experience. If the learner has not been able to retrieve the relevant concepts, it provides remedial practice until that concept is known. On input it accepts spelling variants, as well as British and American English. 

Results

The Video + AI group scored significantly higher than the video group.

Figure 1 shows that Video + AI group had a 61.5% increase in mean retention, from a mean value of 9.00 to 14.54. 

Figure 2 shows that Video + AI group had a 61.5% increase in mean retention, from a mean value of 9.00 to 14.54.

In Figure 3, histograms of the two groups are compared showing that the Video + AI group has a higher mean and users scored higher more frequently.

In Figure 4, a box and whisker plot gives more insight into the respective distributions. The Video only group had a lower median value of 8 and a smaller range than the Video + AI group. The Video + AI group had a 75% increase in median score over the Video group.

Discussion

We know from (Guo, 2014), from a large data set of learning video data gathered from MOOCs (Massive Open Online Courses), that learners drop out in large numbers at around six minutes. This drops dramatically down to 50% at 9-12 minutes and 20% beyond this. Evidence from other studies on attention, using eye-tracking, confirm this rapid drop in arousal (Risko, 2012). The suggestion is that learning videos should be 6 minutes or less. Chunking video down into smaller and meaningful segments achieves this aim and relieves the load on working memory.

Many can recall scenes from films and videos but far fewer can remember what was actually said. That is because our episodic memory is strong and video appeals to that form of visual memory and recall but video is poor for semantic memory and semantic knowledge, what we need to know in terms of language. One remembers the scene and can literally play it back in one’s mind but it is more difficult to remember facts and speech. This is why video is not so good at imparting detail and knowledge. There is a big difference between recalling episodes and knowing things.

Learning is a lasting change in long-term memory and video suffers from the lack of opportunity to encode and consolidate memories. Your working memory lasts about 20 seconds and can only hold three or four things in the mind at one time. Without the time to encode, these things can be quickly forgotten through cognitive overload or the failure to consolidate into long-term memory (Sweller, 1988). Our minds move through video at the pace of the narrator but like a shooting star, the memories burn up behind us, as we have not had the opportunity to encode them into long-term memory. Without additional active, effortful learning, we forget. An additional researched problem is that people often ‘think’ or ‘feel’ they have learnt from video but as (Bjork, 2013) and others have shown, this can be ‘illusory’ learning. The learner mistakes the feeling that they have learnt things for actual learning. When tested they are shown to have learned less than they thought they had learned.

How do we reduce cognitive load in video for learning? (Mayer, 2003) and others have shown that text plus audio plus video on the screen, commonly seen in lecture capture, actually inhibits learning. One should not put captions, text or scripts on the screen while the narrator or person on the screen is talking. (Florella 2019) proposes that learning improves when there are “visual rests” and memory is enhanced when ”people have a chance to stop and think about the information presented". Chunking video down to smaller, meaningful segments and providing the opportunity for active, effortful learning will both enhance learning by reducing cognitive load and increasing reinforcement, retention and recall.  

But what exactly should learners do after and between these video chunks? (MacHardy, 2015) shows that the relationship between video and the active learning must be meaningful and closely related. In a large data mining exercise, they showed that if the two are too loosely related, it inhibits student attainment. To increase reinforcement, retention and recall, (Szpunar, 2013; Roediger, 2006; Vural, 2013) suggest that retrieving key concepts is a powerful learning technique. This was the aim of this study, to test the hypothesis that chunked video with video plus AI-generated retrieval practice increases reinforcement, retention and recall.

Practical applications

There are several possible applications of this form of enhanced video learning:

1.     Existing video learning libraries can be made into far more effective learning experiences 

2.     New videos for learning can be made into far more effective learning experiences

Note that additional design recommendations identified during the study include:

1.     Scripting the videos into a more ‘chaptered’ structure

2.     Clear edit points on visuals and audio at the end of each planned chunk of video 

3.     Close relationship between the video and the retrieval practice

Conclusion

This trial provides evidence that the use of both chunked videos and AI-generated retrieval practice, in combination, significantly increases retention and recall and can be strongly recommended for both existing and new video learning content.

Bibliography

Bjork, R.A., Dunlosky, J. and Kornell, N., 2013. Self-regulated learning: Beliefs, techniques, and illusions. Annual review of psychology, 64, pp.417-444.

Brame, C.J., 2016. Effective educational videos: Principles and guidelines for maximizing student learning from video content. CBE—Life Sciences Education, 15(4), p.es6.

Chaohua, O, Joyner, D, Goel, A., 2019. Developing Videos for Online Learning: A 7-Principle Model. Online Learning

Fiorella, L., van Gog, T., Hoogerheide, V. and Mayer, R.E., 2017. It’s all a matter of perspective: Viewing first-person video modeling examples promotes learning of an assembly task. Journal of Educational Psychology, 109(5), p.653.

Fiorella, L., Stull, A.T., Kuhlmann, S. and Mayer, R.E., 2019. Fostering generative learning from video lessons: Benefits of instructor-generated drawings and learner-generated explanations. Journal of Educational Psychology.

Guo PJ, Kim J, Robin R.  L@S’14 Proceedings of the First ACM Conference on Learning at Scale.New York: ACM; 2014. How video production affects student engagement: an empirical study of MOOC videos; pp. 41–50.

MacHardy Z, Pardos ZA., 2015 Evaluating the relevance of educational videos using BKT and big data. In: Santos OC, Boticario JG, Romero C, Pechenizkiy M, Merceron A, Mitros P, Luna JM, Mihaescu C, Moreno P, Hershkovitz A, Ventura S, Desmarais M, editors. Proceedings of the 8th International Conference on Educational Data Mining, Madrid, Spain.

Mayer, R.E. and Moreno, R., 2003. Nine ways to reduce cognitive load in multimedia learning. Educational psychologist, 38(1), pp.43-52.

Mayer, R.E., 2008. Applying the science of learning: Evidence-based principles for the design of multimedia instruction. American psychologist, 63(8), p.760.

Reeves, B. and Nass, C.I., 1996. The media equation: How people treat computers, television, and new media like real people and places. Cambridge university press.

Risko, E.F., Anderson, N., Sarwal, A., Engelhardt, M. and Kingstone, A., 2012. Everyday attention: Variation in mind wandering and memory in a lecture. Applied Cognitive Psychology, 26(2), pp.234-242.

Roediger III, H.L. and Karpicke, J.D., 2006. The power of testing memory: Basic research and implications for educational practice. Perspectives on psychological science, 1(3), pp.181-210.

Sweller, J., 1988. Cognitive load during problem solving: Effects on learning. Cognitive science, 12(2), pp.257-285

Szpunar, K.K., Khan, N.Y. and Schacter, D.L., 2013. Interpolated memory tests reduce mind wandering and improve learning of online lectures. Proceedings of the National Academy of Sciences, 110(16), pp.6313-6317.

Vural, O.F., 2013. The Impact of a Question-Embedded Video-based Learning Tool on E-learning. Educational Sciences: Theory and Practice, 13(2), pp.1315-1323.

WildFire www.wildfirelearning.co.uk

Zhang, D., Zhou, L., Briggs, R.O. and Nunamaker Jr, J.F., 2006. Instructional video in e-learning: Assessing the impact of interactive video on learning effectiveness. Information & management, 43(1), pp.15-27.

10 important things AI teaches us about ‘learning’

Alvin Toffler, who died this week, said, “The illiterate of the 21st century will not be those who cannot read and write, but those who cannot learn, unlearn, and relearn.” But he had a better quote, “If you don't have a strategy, you're part of someone else's strategy.” There’s one species of this argument that really matters in the learning game - if you don’t have an AI strategy, you’ll be part of someone else’s strategy.

Most pedagogic change now comes through the use of technology. The internet is a huge Darwinian machine that selects ‘fit for purpose’ services, which millions, sometimes billions, want and use. As a result, in the learning game, we have seen more pedagogic change in the last 20 years, than the last 2000 years. This process is accelerating.

There are some new kids on the learning block, like blockchain. But at a deeper level there’s something far more significant that’s happening; AI is re-shaping the learning landscape. Intriguingly, AI now draws its inspiration from evidence of how we actually learn. This should be a wake up call for those of us who work in the field. The AI community now shows us what works, practically, in learning theory, and what doesn’t. They do this by building ‘learners’, software that learns. Pedagogic change no longer comes from educational research (not sure that it ever did), it comes from insights in cognitive science and, increasingly, through this form of technological innovation. AI is the latest manifestation of digital pedagogy and the one that is now giving us confirmation about what works and doesn’t work.

Learning embedded in AI

We’ve recently seen some super successes in AI, with Deepmind’s spectacular win against one of the World's greatest GO players, the Todai project passing the Tokyo University entrance exam. Deepmind used a layered neural network, with an executive layer, initially trained on 30m human moves, which then played itself using a trial-and-error process (reinforcement learning), with the huge processing power of the Google Cloud Platform – and it won.

This is where it gets fascinating for those of us interested in learning theory. Cognitive scientists, and cross-discipline minds, like Demis Hassabis, have taken principles from cognitive learning theory, such as clustering, attention, learning by doing, reinforcement learning, chunking and practice, embodied these in AI, and are using them to great effect in getting software to learn and problem solve. These learning machines are showing us the way. So what can WE learn from this?

1. Search

Access to knowledge and skills has long been a problem in education and training. The traditional model has been slow, scarce and costly supply through expensive institutions such as Universities and libraries, with rising costs and debt. Then along came Google, a great pedagogic leap, and reduced that time and cost to almost nothing. This has been a huge pedagogic leap, one that completely reshaped the learning landscape at all levels. It is all down to AI. Google is nothing but AI.

2. Feedback

We know that improved and detailed, personal feedback accelerates learning. Yet traditional teaching, especially in the classroom and lectures, make this very difficult. Using data to dynamically adapt, in real time, what is taught next, is a common technique in AI. Much of what you see online is determined by algorithms that constantly monitor your needs. Software (AI) driven feedback is the only way to provide such detailed feedback, personally, on scale. AI thrives on new data to update what it thinks it knows, whether from the individual learner or aggregated learners. This is how Bayes and dozens of other species of AI algorithms work. We need to recognise that the brain has exactly the same needs. Everyone has unique learning needs and everyone need to be educated uniquely.

3. Less is more

The principle of clustering, indeed most refinement of algorithms, reduces what has to be learnt to a minimum, looking for optimal ways forward, while retaining efficacy. This is an important ‘less is more’ principle that is all too often ignored in real world teaching and learning. We still teach at too general a level, for example, teaching how to write essays by repeated essay writing, rather than the more detailed components of good writing and analysis. For a complete breakdown of this error see this excellent video by Daisy Christodoulou. AI has applied and developed a battery of mathematical techniques to optimize learning. They select, reduce, optimize, judge and create recommendations. It has 2500 years of philosophy, logic, probability theory and mathematics behind it and now that we have an abundance of data and cheap computers, is bearing astonishing and exotic fruit.

4. Chunking

We have known about this for decades. Chunking, meaningfully, accelerates learning. AI does this often with data structures, pre-processing and choice of efficient algorithms. Sorting algorithms are a great example of chunking (binary sort) for efficieny. Skills acquisition is not about just, say playing golf, but practising how to putt, drive etc. No golfer becomes great by simply playing golf – they chunk down and practice. AI has shown that, to learn effectively, you chunk problems down into their constituent parts, practice those, then build up your skills. AI is lots of little skills that add up to something big – like the self-driving car. Who saw that coming? Teaching general skills is still all too common, especially in schools, colleges and universities. But it’s operating at the wrong level. AI has taught us that we need to focus on the detail.

5. Reinforcement learning

Another insight from learning theory, that the AI folk have picked up on, is that people learn by DOING things. Children don’t learn much by sitting and listening. They do stuff. This reinforcement learning, the idea that every state has a value, beyond the simple binary ‘win’ or ‘lose’ positions, is powerful. Reinforcement learners (software) rehearse learning a huge number of times, trying and trying again, sometimes the best way, sometimes randomly. It’s a turbo-charged learner. So having learnt what to do from human data, it plays itself, an enormous number of times in a short period of time, to get super-smart. This is how Deepmind beat the GO champion. They have been hugely successful in all sorts of AI, real world tasks. It tries all sorts of habits but selectively chooses those that are successful. It learns how to learn. This effortful learning, learn by doing theory, is at the heart of AI.

6. Deliberate practice

AI has embodied the idea of ‘deliberate practice’, from Ericsson, and built algorithms and methods of propagation that do exactly that. They practice with intent, that intent being improved performance. This is what backpropagation in neural networks and many other machine learning approaches do – they automate and optimize deliberate practice and improvement. They embody what education and training has ignored for too long – deliberate practice.

7. Spaced-practice

Spaced-practice tools are largely driven by algorithms that deliver the pacing, interleaving and load balance for spaced practice. We have known since Ebbinghaus, since 1885, that learning suffers from massive forgetting. We also know that the solution is deliberate, spaced-practice. This can be effected online with smart algorithms that determine how this should be delivered. They do what no teacher can do, identify what needs to be reinforced, how it needs to be reinforced (interleaved etc) and when it needs to be delivered (load balanced etc), related to that individual’s needs. It is all down to AI.

8. Data driven

We are no longer data poor, we are data rich. It is no accident that slow burning AI suddenly had its Cambrian explosion, with thousands of practical examples, from speech recognition to self-driving cars, breakthroughs that are reshaping global industries. The ability to manage, read and interpret this data gives you the radar you need to keep ahead of the game. That’s exactly what AI does applying logic, probability and computer power to the problem of prediction. It has been fashionable to ignore ‘knowledge’ as just ‘data’ in education, but AI has shown that knowledge (data) really does matter. It is an integral part of learning, not something to be abandoned or tritely classed as ‘rote learning’. It is, in fact, what enables deep learning.

9. Socratic learning

Socrates learning theory is often called a ‘theory of ignorance’. It was a process of excising what you think you know, stripping things back until the real knowledge was exposed. He taught us humility in learning. This is also true of AI. It knows what it knows but also knows what the probabilities are in its outputs. In this sense it is free from the cognitive biases, even gender, race and socio-economic biases that teachers, as they are human, almost always have. We have a chance here, to both recognize our limitations and embrace technology enhanced teaching. A little Socratic humility, through the use of AI, recognizing where we are good, and not so good, in teaching (and learning), could go a long way.

10. Learning to learn

An interesting insight comes from what AI folk call ‘learners’. These are machine learning algorithms, that learn themselves. This fundamental point, that the new landscape is not the old one of ‘human teachers’ and ‘human learners’, but also one of’ machine teachers’, and importantly, ‘machine learners’. These new entities cope with complexity; time complexity, space complexity and errors. The machine learners learn to learn and now learn how to teach to learn. The automatons automate automation. This has profound implications for productivity, employment and politics. It will, in time, become an existential issue, first for the professions, even for our species.

Conclusion

AI is all about learning. It is software that learns. It is also software that can create good learning content. We have to pay attention to what is happening here. AI tells us that learning is a process, one that can be unpacked and copied - reverse engineered. They are achieving things that were unimaginable just a few years ago. They are learning about learning by creating effective learners. The fact that AI is having so much success by following the lessons that cognitive science has to teach, is surely a wake up call for learning theorists, especially those still stuck in foggy world of social constructivism.

Miller (1920 - ) Magic number 7 plus & minus 2 - chunking

George Miller is best known for his paper ‘The magic number 7 plus and minus 2’ (1956) which focused attention (literally) on a problem that plagues teaching and learning, the danger of ‘cognitive overload’. For this reason alone, he deserves mention. Short-term memory has a limited capacity and this capacity can be seen as being dependent on ‘attention’ i.e. our ability to attend to things in short-term memory. More than this Miller was one of the first to bring in cognitive information theory, famously saying the the mind had been brought back into psychology on the back of the machine.

Chunks

Miller started by identifying relevant studies that show we remember ‘chunks’ of information. But what is a ‘chunk’? We can remember, on average, seven randomised numbers but only six letters and four or five words. Miller’s solution was to posit clusters of chunks e.g. a word is a cluster of letters, a phrase a cluster of words. But Miller’s definition proved inadequate. This problem, that we remember different types of content differently giving variable sizes of chunks, was to open up a whole field of experimental work that began to reveal how short-term memory works. The concept of a ‘chunk’ has proved enduring and useful in memory research.

Encoding methods

Research by Bower (1970, 1972) showed that meaning also played a role in chunking so that, for example, TVFBIJFKCIAIBM is far more difficult to remember than TV FBI JFK CIA IBM. Mnemonics, as chunks, can also be used to unpack larger amounts of information. Gardener (1988) then showed the power of the ‘generative effort’, where creating your own chunks, is even more effective.

Encoding

Of course, chunking is just one form of encoding that increases retention, the organisation of material is another, chunking categories logically and with relatively few numbers of items in each category, increases recall, as do outlines. Imagery is another useful form of encoding, which is why loci and peg-word systems have been used for centuries to improve memory. In particular, movement and interaction between images so that they do things to each other, were shown by Wollen, Weber & Lowry (1972) to be particularly effective. Craig & Lockhart (1972) popularised the word ‘elaboration’ for techniques such as paraphrasing, summarising, highlighting and note taking, all of which improve recall.

Duration

Miller’s work also led to research on the time we are able to hold attention in short-term memory. Waugh and Norman (1965) found, in recall experiments, that this was around 18 seconds. However, if recognition, and not recall, is used this can stretch out to 90 seconds (Lutz & Wuencsch 1989).

E-learning and chunks

E-learning, with its learning objects, nuggets and page-based delivery has long sought to chunk material to make it more digestible and to increase encoding and retention. Chunked media objects such as short videos, animations, examples and so on, have also been part of the e-learning landscape. Attempts have even been made to determine standards around reusable learning objects that can be re-assembled in different orders and in different contexts and courses, like LEGO. The problem is the conflict between usability and the flow and context of the learning experience on one hand, and the fragmented nature of chunked learning objects on the other. This has also led to standards such as SCORM, that some believe traps designers in to producing learning objects and does not cope with more complex entities such as simulation and games pedagogy.
Conclusion

Miller, a pioneer in the information processing or computational model of memory, set us on the path of defining the limits of short-term memory, the gate through which attentive learning must pass. It proved to be limited in time, capacity and the way we encode information. This has opened up research on cognitive overload and a deeper exploration of what we now call ‘working’ memory and encoding through elaboration. The magic number ‘7’ may have proved misleading, as our short-term register often operates at a lower 3-4 register level, but the principle of cognitive overload remains a limit on learning.

Bibliography

Miller, G.A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63, 81-97. http://www.well.com/user/smalin/miller.html

Miller, G.A., Galanter, E., & Pribram, K.H. (1960) Plans and the Structure of Behavior. New York: Holt, Rinehart & Winston.

Miller G Psychology Penguin

Bower G. H. (1970) Organisational factors in memory Cognitive Psychology, 1, 18-46

Bower G. H. (1972) Mental imagery in associative learning in Gregg L,W. Cognition in learning and memory New York, Wiley

Gardener (1988) Generation and priming effects in word fragment completion Journal of Experimental Psychology: Learning, Memory and Cognition 14, 495-501

Wollen, Weber & Lowry (1972) Bizarreness versus interaction of mental images as determinants of learning Cognitive Psychology, 3, 518-523

Craig F. I. M. & Lockhart R.S. (1972) Levels of processing: A framework for memory research, Journal of Verbal learning and Behaviour, 11, 671-684.

Waugh N. C. and Norman D. A. (1965) Primary Memory, Psychological Review, 72, 89-104.(1965)

Lutz J. & Wuencsch K.L. (1989) Acoustic interference in a recognition task. The Journal of General Psychology, 116(4), 371-384

10 top online learning writing rules and the psychology behind them

Don’t waste your reader’s time. There is a difference between reading from the printed page and reading from a screen. Reading from paper makes use of reflected light, whereas the screen is lit from behind. This makes it more difficult to read on screen, particularly when large amounts of text are involved, and where the user is required to read for long period of time. It is almost always a mistake to simply transfer text from the printed page or workbook to the screen, without taking into account this fundamental difference.

In all reading, we do not simply scan letter-by-letter, word by word, sentence by sentence. We scan, skip and look for anchors. When we read text on screens, these behaviours become more extreme. More than this, in online learning, we need to be aware of chunking, meaning, order, cues and retention. The art of writing text for online learning is a skill, especially when that text will be read across a whole range of devices and different screen sizes. Ten quick suggestions:

1. Cut until it bleeds

Several pieces of research point towards overlong text as a destructive force in learning. Mayer, Bove, Bryman, Mars and Tapangco (1986) showed that too much text can reduce learning by up to 50%. Edit out unnecessary words, use shorter words, break down long sentences. Give yourself a maximum word length and rewrite to within that limit. Cut it til it bleeds - then cut it again.

2. Chunk text

When we read to learn, text must be chunked to increase retention. This is a well researched area. Chunking text means more paragraphs. Break down the text into more paragraphs than you would on the written page. Ebbinghaus through to Miller and many other psychologists of learning showed that chunking is a proven technique to increase retention.

3. More structure and meaningful headings

Bruner researched how structure (content must be structured so that it can be grasped by the learner), sequence (material must be presented in the most effective sequences) and generation (encourage extrapolation, manipulation and a filling in the gaps, just beyond the learners existing knowledge) all aid learning. Write to these rules. Like this blog post, screen text benefits from more headings. These headings act as structural cues for meaning, understanding and encoding, all of which aid retention. Don’t use headings like ‘Introduction’ ‘Section 3’ and so on. Make headings short, meaningful cues. Start with a question, capture main learning point in the heading.

4. Highlight key words

Highlighting ‘cue’ words within text in bold or italics. These act as cues for retention but don’t overuse, one or two at most per paragraph. Tulving explored cues in learning showing that they are vital in the process of encoding, storage and recall, and therefore retention. Another useful technique for emphasis is to pull out quotes or key sentences. You often see this in good magazine articles, quotes and main points pulled out in a separate or much larger font, even placed separately at the side of the text. This is a useful learning device that can be used to reinforce the main learning point. Again, these act as cues, which Tulving thought were essential in memory and retention.

5. Don't 'centre' text

It is far harder to read centred text, as you lose the left adjusted visual cueing, and people tend to read the lines, not the sentences - so don't do it.

Graphic artists may want your text to 'look good' but that it not the point in learning. Above all, it must be readable. This means left justifying the text.

6. Use stab/bullet points

Long paragraphs, that are in fact lists, work better on screens as stab points. These also increase encoding and retention. In one study, Mayer, et al. (1996) presented 600 pieces of scientific learning and found that briefer versions, which were concise, coherent and co-ordinated, resulted in far more effective learning. They are precise in their recommendations, ‘There is a clear pattern in which the more words added to the core verbal explanation, the more poorly the student does in producing the core explanative idea units. These results are consistent with the idea that the additional words overload verbal working memory, drawing limited attentional and comprehension resources away from the core verbal explanation.’

7. Personalise

Formal and conversational styles (using terms such as I, me, my, you, your and we) of text have been compared by Moreno and Mayer (2000) showing that the conversational style resulted in 20-46% better learning. Nass and Reeves also produced 35 studies showing that learners see technology as a human and therefore needs dialogue on a personal level.

8. Keep text close to image.

Mayer (1989), Mayer Steinhoff Bower (1995) and Moreno and Mayer (1999) in five separate studies compared graphics with text close to the graphics, and graphics with text below the graphics, at the foot of the screen. In all five studies, learners who used the co-located text and graphics improved their problem solving by between 43-89%. Similar results have been found by Chandler and Sweller (1991), Pass and Van Merrienboer, (1994). Making the learner’s eye jump from one part of the screen to another is disruptive and reduces the effectiveness of the learning.

9. Don’t do audio and text on screen

Words, in both text and accompanying audio narration, can hurt learning. This is interesting, as it is often assumed that one needs both to cover accessibility issues. Clark and Mayer (2003) argue that ‘audio or text on their own’ are better than ‘text and audio together’. This is confirmed by another study by Kalyuga, Chandler and Sweller (1999) where the group with audio scored 64% better than the group with both text and audio. They claim that one or other is redundant and will overload the visual and aural channels.

10. Don’t use unnecessary graphics

A review of studies around this concept, known as the redundancy effect, by Sweller et al. (1998) cites a list of research studies that all point to the damage done to learning when redundant material interferes with the efficacy of the learning. For example; they illustrate a point about leaving out extraneous or distracting graphics in media with an experiment, conducted by Harp and Mayer (1997), in which students were given a text to read on lightning strikes. Students who read the passage accompanied by elaborate colour photos with additional captions - as opposed to the text with simple graphics - showed 73% less retention of knowledge and 52% fewer solutions on a transfer test.

Conclusion

The problem, in many online learning courses and resources, is not that they are in text but that there is too much text and that it is badly written and presented. A good editor can often take subject matter text, book text or initial documents and use these techniques to make the text screen-ready. When I hired interactive designers, I gave them all literacy skill tests and was keen to see evidence of their ability to write well and edit. In many ways this is a primary skill with screen friendly writing something that had to be acquired. A good editor will reduce text like a good sauce and the learner invariably benefits in terms of understanding, encoding, retention and perseverance.