OK, so you're an experienced teacher of 30 pupils, only slightly jaded by the constant trickle of government edicts about what you should do and how you should do it, and still basically happy with the career you chose all those years ago. It's Friday afternoon and you're whistling cheerily in anticipation of the weekend ahead as you pack your bag and do some last-minute tidying up in the classroom. And then the Principal pops in. Alarm bells ring immediately; the Principal never just “pops in.” This time is no exception. She wants to let you know that she's agreed to trial a new approach in which you and your colleagues will be regularly observed and assessed on the basis of how effectively you personalize the teaching and learning environment in your class.
She smiles as though she's presenting you with a large bouquet of orchids as she tells you that she would like to see your “personalization improvement plan” on her desk by half term, which is in just two weeks. You smile and you nod and you say “No problem. Have a great weekend!” all the while unleashing a torrent of abuse in your head.
The next half hour passes in a blur of hair pulling (your own), moaning with the teacher in the next room, and eating the biscuits you'd managed to ignore all day. You're already doing your best. What on earth do they want from you? You appreciate that it probably would be great if you could teach 30 different lessons at the same time, but how on earth are you supposed to do that? What can you actually do about the fact that Danny Hardcastle is bored to tears by how easy he's finding math but can't take on anything harder until you find time to show him some new techniques; that Millie Bracken still can't read much and is getting teased by the other kids for reading baby books; and that blue table are winding each other up to the extent that none of them have made any progress in anything for weeks. Every time you focus on one problem it seems to create several more. You trudge out of school with a heavy heart, already dreading Monday.
The most obvious solution currently available to us is computers. Using technology to personalize teaching and learning is an approach that has proved somewhat contentious, and its benefits have not been proven, but in terms of practical potential computers are pretty hard to beat. It's certainly enticing to think that there could be technology out there capable of drawing out children's skills and understanding at a precisely calibrated pace and with a precisely calibrated method of address.
Some people don't like the idea of pupils using computers too regularly. They see them as a necessary modern evil, acceptable for background research—“google” is a verb even for very young children these days—but not much more than that. The fear seems to be that classrooms kitted out too enthusiastically with computers will become dull, lifeless places with rows of automata silently staring at screens. As one well-respected UK education commentator, Phil Beadle, put it: “An inspiring education is a sensory joy, and the idiotic preeminence of the computer a denial of this.” He fears that: “…if a computer can be used to personalize education, then there will eventually be no need for learning support in human form.” (Beadle, 2008). We agree that an inspiring education is a sensory joy, but we disagree that using computers necessarily precludes this and that computers diminish the need for human learning support. Computerized personalization should support, not replace, school personnel (although some misguided political number-crunchers might think otherwise).
A more serious challenge is that computer-based teaching, even when highly individualized, has not yet been proven to increase achievement scores. In fact, a 2010 US Federal Review found that the computer-based instruction programs they assessed showed “no discernible effect” on students' SAT scores. It is important to education that new interventions are supported by evidence, and computer-based personalized learning methods have not yet satisfied the high standards of science. Nonetheless, there is cause for optimism that they might. For instance, there are scores of case studies in which teachers, schools, and even whole school districts report improvements among their pupils in terms of understanding, enjoyment, and ability, making it possible that the studies are not yet getting at what the teachers see on the ground.
It is disappointing that the software developed so far is not supported by scientific evidence, but we think that this should provide a spur to make such programs, and their implementation, more effective rather than ditch the approach as a whole. Until such approaches gain a strong scientific evidence base we would advise cautious enthusiasm and funding to improve the software, rather than a gung-ho spending spree. Personalization technology is a work in progress. It is an approach that has more potential for truly personalized education than any other currently available to us, and therefore it should be given the time and resources needed to improve.
Computers and software, when used well, should free up teachers and teaching assistants to support individual pupils more effectively. This approach is likely to be most successful in objectively assessed subjects such as mathematics. We do not for one moment suggest that all lessons should be computer-based. The most highly developed math instruction program we are aware of was created by Carnegie Learning, in conjunction with scientists at Pittsburgh's Carnegie Mellon University. The US Federal Review mentioned above did not find the Carnegie program to have statistically significant effects on achievement. However, many of the schools to have used it describe benefits. For example, Louisiana math teacher Krista Majors describes struggling students exposed to the program on Carnegie Learning's website:
Students who were enrolled in the pilot class “caught up” with their peers enrolled in regular classes during the first year. They maintained that gain into the second year even though they were no longer enrolled in Carnegie classes. Most students in the control group, who were mixed in with the regular population, either stayed the same or fell further behind.
There is obviously work still to be done here to establish whether empirical support for this approach is forthcoming. However, it seems to us that the existing program has merits and may in time become a model for a teaching and learning strategy that will improve as a result of the research process. Its ability to respond to individual development is intriguing and it is not unreasonable to predict that such software will also eventually be able to interact with the Learning Chip technology that we discussed in Chapters 1 and 2. There's work to be done but we believe it to be work worth doing.
Carnegie Learning was founded by a team of cognitive scientists from Carnegie Mellon University in conjunction with a team of veteran math teachers. One of the founders was Professor John Anderson, best known for his development of a model of how the mind works, known as ACT-R. The aim of any cognitive architecture such as this is to define the fundamental cognitive and perceptual processes that make the mind work. ACT-R has done this successfully, and has been validated by hundreds of studies. This model of the mind underpins Carnegie Learning's Math program by using Cognitive Tutors with an internal ACT-R model that can mimic the behavior of any pupil using the software. Cognitive Tutors can therefore personalize materials and “predict” the difficulties a particular pupil might have.
By recognizing individual differences, the fact that pupils develop, learn, and master math at different paces, this artificial intelligence-based approach can personalize the math learning environment of every pupil in a class. It can identify when a pupil is struggling or has not fully understood something and then customize prompts to focus on the area of weakness, providing new problems until the idea has been understood or the skill has been learned. There's no putting your hand up and waiting around when you're stuck, or giving up because “I just can't do it”; the software is capable of gently guiding pupils out of ruts, and supporting and encouraging them as they do so. It works like Personal Training—a source of expertise and encouragement by your side as you work to achieve your goals and your potential.
The Carnegie math program is rooted in findings from cognitive science, as described above, but it also draws on the body of research into mindset and motivation led by Professor Carol Dweck of Stanford University. In Chapter 7, we discussed the implications of this “mindset” research on the way that we praise and encourage our children in order to build their motivation and self-confidence. The Carnegie software does something similar by harnessing Dweck's findings to tailor the feedback that Cognitive Tutors provide to pupils.
In her books, talks, and papers Professor Dweck describes two types of “mindset”: a fixed mindset and a growth mindset. Over the course of scores of experiments she and her colleagues and collaborators have shown how a growth mindset yields better results for everybody and, importantly, how the growth mindset can be taught. In light of this research the feedback given to pupils by Carnegie Learning's program is specifically designed to foster a growth mindset as a spur to mathematical development.
People with a fixed mindset believe that intelligence and talent are innate and cannot be changed. This leads to beliefs along the lines of “naturally clever or talented people shouldn't have to try” and “If I fail, people will think less of me.” Dweck and her team have shown time and again that adults and children with this mindset shy away from challenges because they don't believe they can learn to do what doesn't come naturally to them; they don't want to make an effort because having to do so undermines their self-worth; and moving from their comfort zone puts them at risk of what they see as failure, and this is intolerable to them. This perhaps explains why some pupils who perform very well in English write themselves off in math, and vice versa—their self-concept can't cope with the fact that success doesn't come effortlessly. The generalist genes hypothesis suggests that a pupil who is extremely able in English is likely to be at least reasonably good at math, and yet such pupils, if they have fixed mindsets, are likely to write themselves off as hopeless just because they fall short of exceptional. As a result these people, even when highly able and talented, often plateau and live a life in which they operate well below their full potential.
Many pupils have fixed mindsets, some from very young ages. Dweck argues that they mostly acquire these beliefs from the people around them—their parents and, later, their teachers. However, we suspect that such beliefs are also a marker for genetically influenced temperament: we hope to explore this in our own future research. We think it is likely that, for genetic as well as environmental reasons, it will be harder for some people to develop a growth mindset than others. However, Dweck has lots of good ideas about how to help a child with a fixed mindset to develop a growth mindset. Her suggestions form the basis of an educational software program she has developed, called Brainology. The Brainology software is another means of connecting with individual pupils to raise achievement and, again, shows the advantages of learning from a computer at least some of the time. A computer, for example, talks to you alone. You can pause or repeat lessons whenever you like and as often as you like. You're not having to keep pace with 29 other pupils.
A child with a growth mindset loves a challenge. Dweck first became interested in this whole area when she was researching different people's responses to failure. She was surprised to see that some children, when confronted with a puzzle that was very difficult for them, didn't see that as failure. She saw from their responses that rather than feeling as though they were failing when they couldn't do something straight away, they felt as though they were learning. She describes her own initial (fixed mindset) reaction to these kids: “What's wrong with them? I wondered. I always thought you coped with failure or you didn't cope with failure. I never thought anyone loved failure. Were these alien children or were they on to something?” Children (and adults) with a growth mindset know that hard work pays off. Dweck and her team asked people of all ages a simple question: “When do you feel smart?” The fixed mindsetters said it was when they didn't make any mistakes; when they finished something fast and it was perfect; or when they found something easy that other people couldn't do. The responses of those with a growth mindset were very different. They said they felt clever when they tried really hard and managed to do something they couldn't do before. Dweck's findings are, we think, hugely important to education and parenting. The programmers at Carnegie Learning were well advised to take this research on board in designing their personalized math software, as “mindset” appears to provide a unique perspective on pupils' learning triggers.
It turns out that lots of parents and lots of teachers, encouraged by the self-esteem movement that bulldozed its way through child development in the late twentieth century, are getting it wrong. Every time we tell a child “I can't believe you got another A—you're so clever!”; “You're a natural—you're going to go far!”; or “You were robbed—you were easily the best and you should have won,” we encourage a fixed mindset. What starts out as a simple attempt to make the child feel good and bolster their self-confidence inadvertently harms their ability to achieve their full potential. If we praise them for ability they won't want to risk failure. And this is not just opinion—Dweck has a whole series of compelling studies which support her advice. Instead we should praise children for effort, or for trying different approaches to solving a problem, identifying strategies for overcoming hurdles. If a child completes a task fast and perfectly they learn nothing from it—the task is simply too easy for them. It happens all the time, but Dweck suggests that rather than giving the child a sticker or a certificate and telling them how great they are the teacher or parent should apologize to them for wasting their time and promise to find a more suitable task next time. Carnegie's software automatically guides such a child to the next level in mathematics and offers some of the early support they need in learning the new skills involved. Equally, it encourages and supports pupils who are struggling in ways that praise small steps forward and encourage perseverance.
Children do best when they are working just outside of their comfort zone, having to scrabble a little to reach the next level. Children with a fixed mindset will find this desperately uncomfortable and want to back away, so the teachers and parents around them should respond by praising effort, concentration, persistence, and all of the other qualities that keep them at it until they get the buzz of achieving something they didn't know how to do before. The Carnegie program promotes a growth mindset by giving feedback that focuses on effort and progress and via “Messages of the Day” that include facts about how the brain changes and grows as students learn. They show pupils that the brain is a muscle that can be strengthened via exercise, namely hard work and perseverance. Perhaps one of the ways the Carnegie software could be improved to meet the high standards of scientific testing would be to further personalize these messages, to trigger positive genotype–environment correlations by giving individual kids more precisely targeted praise and encouragement. The fixed mindsetters, for instance, are likely to need a different approach to those who already have a growth mindset. Some will be tougher nuts to crack than others, and software that can recognize this and respond appropriately may represent progress for both the child and the teaching method. Dweck's “mindset” research has widespread implications for education.
Carnegie Learning's ACT-R-based math instruction program is just one example of the personalization software already available to educators. We do not recommend that all schools rush out and buy it, or current alternatives to it, because it is not at all unreasonable to expect such a program to prove its worth in the form of objectively assessed improved achievement. It has not yet accomplished this. It is important, though, that some schools trial these approaches, so that they can be refined and assessed. The fact remains that software such as this can potentially help to guide a child through the math curriculum at their own pace, and can facilitate progress in a way that a teacher alone cannot, by literally providing 30 different lessons to 30 different pupils. In short, it is capable of much of what personalized instruction should be, and therefore capable of making personalized teaching practically possible. In an ideal world, educational decision makers at a national level would support such software developers. In order to assess their effectiveness for example, it would help if they were linked to national tests. It would also help if adequate funding was provided for professional development so that teachers are properly taught how to integrate such software into their lessons.
Computers offer personalization technology but they also offer choice and, importantly, access to education. Stanford University professor, Sebastian Thrun, recently bemoaned the fact that his Artificial Intelligence classes were only reaching the 200 or so students that were enrolled on his course. Not being the sort of person to sigh and shrug and do nothing he developed an online version of the course. Since its inception he has enrolled 160,000 students and now, through Udacity, a private company he has founded, he offers 11 courses to students from, according to Thrun, every country in the world except North Korea. He says his costs, roughly, are $1 per student per class and, at this point Udacity's courses, examinations, and certification are offered free of charge. Courses such as Thrun's are known as MOOCs (Massive Open Online Courses) and are increasing in number. 2012 was designated “The Year of the MOOC.” At a time when organizations like UNESCO are spearheading the spread of education throughout the world, initiatives such as this one have to be of interest as a means of providing a university-level education to interested students who could never otherwise afford one. It is interesting that of the students who sat Thrun's first examination the top 410 exam performers were not Stanford students but simply interested people who had signed up online. In the democratization of education it seems highly likely that computers will play a major part. The work ahead involves finessing products and processes, and establishing validity and proof of effectiveness.
Of course personalized learning cannot begin and end with computers, and if we were privileged with a bird's-eye view of what goes on in every classroom every minute of every day we would see many, many examples of personalization in practice. There are wonderful, sensitive, and highly skilled teachers out there, in great numbers, who draw out the best from individual children all the time. The difficulty lies in doing it for all children at the same time. We need to focus on identifying what works for individual children and testing whether it could work more widely and whether it can stand up to the rigors of scientific proof. When we find initiatives that meet these criteria they should be rolled out to everybody, so that all schools, teachers, and pupils can benefit from good practice. In our own experience—both personal and professional—we have come across approaches to personalization that really do appear to work, approaches on which we would be prepared to structure testable hypotheses.
Joined-up thinking is sorely needed here if we are to pool all of the personalization practices out there and scientifically test their effectiveness. The US-based “What Works Clearinghouse” attempts to achieve exactly this. Created in 2002, it gathers educational research together in one place and clearly shows the degree to which particular educational methods are supported by good scientific evidence. The University of York's Institute for Effective Education represents a related initiative in the United Kingdom, and there is a “What Works” agenda gaining momentum in government too. However, systems such as these can only be as good as the research that's out there, and there is a responsibility to fund well-designed studies of any education intervention that shows promise. It will be important that at least some of these studies are genetically sensitive. Seeking scientific evidence of effectiveness is a step that is too often missed out because it takes time and money, and a government with a four-year term wants to make as big and quick a splash as possible, for as little cash as possible. This is short-sighted, and is the reason that education should be a cross-party domain that cannot be disrupted by changes of government. If we put time into finding what works, implementing it, and then allowing it to bed down and flourish over as many years as it needs, we will benefit as individuals and as a society. Taking your best guess, spending taxpayers' money, irritating teachers, and then withdrawing the intervention because of teething problems really doesn't work.
And so, let's not abandon our frustrated teacher, dragging her feet as she walks out to the school car park racking her brain for something to put in her “personalization improvement plan.” We would suggest one simple step: turn the page. In the next chapter there are 11 ideas that take into account everything that we have discussed in this book so far—all genetically sensitive and all as practical as possible. Together they explore ways—some old, some new—of introducing personalization to every classroom in the country. How about that for a personalization improvement plan?
Beadle, P. (2008). A step too far. The Guardian, 1 April 2008.
http://www.carnegielearning.com/ (accessed 17 June 2013).
For an interesting approach to personalized learning, using technology among other methods, read this article about the “School of One”: http://www.theatlantic.com/magazine/archive/2010/07/the-littlest-schoolhouse/308132/1/ (accessed 17 June 2013); and watch this video: http://schoolofone.org/concept_introvideos.html?playVideo (accessed 17 June 2013).