• Print

Does digital text create a cognitive gap?

A study finds electronic text may disrupt learning techniques.

Kindle DX Pilot Project montage.jpgEreaders are changing the face of reading across the board, and experiments in creating more economic-friendly textbooks for students are increasing. The results, however, are not all positive.

As students attempt to incorporate electronic text into their studies, issues with e-textbooks are starting to emerge — and the problems go beyond poor annotation and sharing tools.

A study at the University of Washington and six other universities in the US involving the use of the larger-format Kindle DX indicated a disconnect between digital text and the way students learn. In a post for Fast Company, Ariel Schwartz cited from the study results:

The digital text also disrupted a technique called cognitive mapping, in which readers used physical cues such as the location on the page and the position in the book to go back and find a section of text or even to help retain and recall the information they had read.

More results from the study are discussed here and here. All seem to point to an opportunity to create different kinds of ebooks and ereaders for use in academia that better accommodate cognition. If the study results hold, companies creating smartphone apps for e-textbooks may want to rethink their efforts.

Photo: From the University of Washington Kindle DX pilot website.


tags: , , , ,

Comments: 15

  1. This is not so much a flaw in the devices… I think e-readers are simply exposing the underlying assumption of most textbook content creators: that the material is meant to be re-read, re-referenced, etc. *by going back*. The “cognitive gap” reflects a failure in the design of the materials to support forward-only reading.

    When I hear “works for novels, not for text books” I’d rather ask the question, “So, how can we redesign the way we author text books (and other forms of study materials) so that the learner does not NEED to keep going back?”

    I do not believe that what we are seeing in these results is just The Way People Study, but it does reflect the way most of us are forced to study *because* these materials were designed knowing that we could and would go back.

    But what if we thought of it as a creative constraint and asked, “Imagine that after completing each page, that page simply vanished, forever. What could we change in the book to support continous forward-only movement (in a way that is at LEAST as effective as the way they’re used to studying-by-referencing?”

    This opens up a world of possible answers that do not involve changes to the devices, hardware or software, but do require changing the way the content is created and presented. This is no different than how many of are thinking of video learning which in some ways (to do it right) is even MORE constrained because the material moves always-forward, and at a pre-determined speed.

    There are many interesting and exciting answers to these questions. Many of us have spent much time considering the possibilities, and consider the forward-only constraint one that will make is better as designers of learning.

    (There is some very old research that I think applies here, though I will need to dig it out. My co-authors and I refer to it ourselves as “over-authoring”, and we use it in our books, sometimes well, sometimes not. The idea is extremely useful here even if our own implementations have been clunky.)

  2. Interesting idea on adapting the material for forward use only. I won’t expect too much from it though. Learning is fundamentally non-linear. Most people probably err on no going back and reviewing enough. Even when listening to podcast, after the show, I often have to go back to the computer to do further study or write up something to solidify my learning.

    For now I think school should go slow on ebook and regard it as interim device for niche use only. Ten year from now the next generation device will be very different from the primitive Kindle or iPad we have today. The ebook you purchased today may not even be usable then, either because the company behind them is no longer around, or the format is obsoleted by some vastly different future format.

  3. @wai yip tung — I agree that people often do not go back enough, and that they need to, but again, that is in part because most of today’s study materials were built with the assumption that you would/could. I am suggesting that this is an assumption we can challenge. We do not need non-linear learning experiences if we can at least patly simulate them, even in a forward-only experience with no branching.

    An assumption of forward-only means we must anticipate the cognitive leaks that require learners to go back for review. There is some research done back in the earlier days of artificial intelligence for learning (Intelligent Tutoring Systems) that suggests we can, and I can of course point to a few books that already use “forward-only” thinking in their design.

    I think it is a lovely, motivating, and do-able challenge. And one that we can do now, vs. waiting for devices and reader apps to improve to the point where they simulate the way we study from physical books today. This is a real problem, though, because it is not a matter of just getting people “used to studying from digital text books”. There’s a fundamental problem that won’t go away until we create digital textbook “experiences” that better support learning and studying. I am saying we can improve the digital content to help compensate for the current limitations of e-readers.

  4. Interesting – so surely what we should be doing is teaching the appropriate study skills and reading techniques to allow the students to use cognitive mapping with electronic texts.

    I teach low level regional literacy students entirely online with virtual classroom and website resources/links. My students would struggle to map cognitively with paper based and don’t seem to experience any additional difficulty with online resources.

    We do try to expose our students to many different types of online resource with the intention that this will slowly enable them to extract relevant info from different “text types”

    Jo Hart

  5. I read these comments with great interest; I’m the person who helped run the study and was first author on the paper. 🙂 It’s tricky to get a press release to tell the whole story since space is so limited, so let me say a bit more here.

    I like the idea of considering the linear-only model as a way to redesign reading materials; surely that design constraint would produce some interesting design thinking as a result. The thing is, students are reading for some other reason than simply to read: They have broader goals that motivate their need to read for academic purposes, and they also have a richer context of academic work into which reading fits. In other words, reading is only part of the picture.

    I’d suggest that the area where we need to shift the design thinking is toward students’ workflows and away from isolated pieces of those workflows. Reading is just a piece of the puzzle, and it’s hard to design to support academic reading when that puzzle piece is considered in isolation.

  6. Thanks Alex, and I agree. I have been trying to shift my language more around “experiences” and “learners” vs. “readers” to help reflect the broader context in which these “texts” are used.

    I do think the most interesting and useful questions are around what the students/learners are trying to accomplish and how we can help them do it. We must revisit the assumptions we used in designing and creating most text books and self-study materials. And we might consider areas where the workflow is not based on what learners WANT to do, but on what they have been forced to do by the choices we (book/study materials developers) have made.

    It’s an exciting time, and I think your study, Alex, is going to
    have a positive impact. It has certainly helped motivate *me*.

  7. This is an interesting debate and I am unsure which is the better solution: to enable e-textbooks that enable non-linear reading, skimming, marginalia, notes, etc with ease vs. adjusting our paradigm to accept that from now on linear textbook reading will be the way we learn.

    Is it either or?

    Personally, I’m a little uncomfortable with the idea of abandoning non-linear reading and learning, especially given that one of the early promises of technology was seamless hypertext. Feels like a step backwards if we abandon non-linear reading.

  8. @Kathy:

    If you’re old enough (not you, I expect) you will remember “Tutor Text” textbooks. I used a few in the 1960’s; don’t even recall the subjects. They were designed as forward only. They didn’t work. (So, I decided to check WikiPedia, expecting at least a paragraph, but no. OTOH, I did find an Amazon listing for one: http://www.amazon.com/Trigonometry-Practical-Course-Tutor-Text/dp/B000NOCWZO )

    The idea: present some material, then quiz the reader. The answer(s) were on some, quasi, random page; the reader thus traversed the book completely non-monotonically. If you expand the image from the Amazon link, you’ll find that my memory (not that you need to believe I actually remembered this) is pretty close.

    The notion that deeply difficult material, not just another Harry Potter pot boiler, can be assimilated in however small spoonfuls, one at a time (and only once) is just silly. That’s not how people think, and it’s why non-fiction of all kinds have indexes.

  9. S Falcon Soeder

    Thank you for offering challenging concepts. If your not confused or frustrated at some point, you’re probably not learning. I think it would be wonderful if we could understand human learning processes to the point where we could evolve teaching materials that adapt to wide learning styles instead of expecting the student to ALWAYS adapt.

    I think it is more likely that humans who can succeed with standardized learning tools will always redefine the narrow spectrum of “norm”. Man changes his environment then adapts to the results of those changes.

    Who will succeed in the future? ADHD, Aspergers, linear thinkers, intuitive learners, L/R brain… The most successful persons are not always the gifted, the strong or the persevering. USUALLY it is the one who can best navigate within their society. (Kudos to Temple Grandin.)

    In some professions, tactile learning is indispensable. CGI and CAD designers and engineers are much more successful if they have completed simple physical skeching/spatial/perspective classes.

    For a very long period, we sat in groups of two or many. Passing knowledge on through repetition and review. Striving alone we make mistakes, learn how to repeat mistakes that improve an outcome, avoid those that do not and value the difference. It is hard for me to imagine successful linear learning for the average person.

    I had total recall for most of my life and do well with tactile learning. Combining the two worked well for me. But that’s just one person.

    We need to understand HOW we learn first. More importantly, how can we pass on WISDOM vs KNOWLEDGE?

    ps I hate concussions. After 15 years I still feel like I’m working with half a brain and 1/3 of my vocab.

  10. “The notion that deeply difficult material, not just another Harry Potter pot boiler, can be assimilated in however small spoonfuls, one at a time (and only once) is just silly. That’s not how people think, and it’s why non-fiction of all kinds have indexes.”

    I do not agree. Or rather, I agree that the approach you are talking about IS silly, but not a useful example of forward-only. The idea of “only once” is in no way a requirement of a forward-only experience. The best learning is an iterative experience (especially deeply difficult material). Forward-only does not mean “only once”. It means only once *for that iteration*. And the argument of “tried that, didn’t work” is only relevant if what was tried was a good and useful implementation of what the research supports.

    Forward-only experiences map most naturally to how people do learn best… with each additional look layering on new bits within an ever-growing context.

    Not suggesting that learners will/should NEVER go back, but that the point is to design materials that are not based so heavily on the assumption that they will. And I also do not want to confuse reference materials with learning experiences. They are often smushed together into books, but making extremely clear distinctions is another part of this discussion.

    I will say that the top-selling technical books today on programming and design patterns were largely designed using a forward-only approach. The market has definitely spoken on this over the last 8 years. However, these were not designed as university/academic text books, but simply as learning experiences. And only one of a zillion possible implementations of forward-only. (which I should really call forward-MOSTLY; we certainly did not nail it 100%).

  11. @s falcon seder
    Some of the research we used was based on exactly that question: how can we use the power of this new (at that time, late 80’s) interactive/hyper linkable tool to create learning experiences that adapt to the individual’s learning styles? At the time, most of what they found was that people did not learn best when presented with a specific learning style that was either selected by the learner or inferred by the program (and the whole “learning styles” concept has been debunked at this point).

    However, what they did find was the juicy part: when multiple representations of a topic were inserted, linearly, nearly everyone learned better, and learners appeared to just skip or skim the seemingly redundant “here is another way to look at this” content.

    Another way to consider forward-only is to try to model a human tutoring session. We asked ourselves, what can a learner do in a session with a human tutor that they cannot do with a book? Two simple examples are “raise their hand” and “look confused”. In both instances, the human tutor would then DO something as a result… answer the question or, in the case of “confused look”, say something like, “well let me try THIS way…”. At least part of the answer to crafting better learning materials is to just assume the questions and confused looks, where appropriate, and deal with them just as a human would. The downside (it appeared) was that you end up adding extra stuff that might get in the way of forward flow for those who ARE moving forward and do not have questions and confusion.

    However, it turned out that if presented in a particular way, learners were not held back by the extra representations, but on the contrary seemed to learn even more effectively and efficiently. Our answer was, “when in doubt, add the extra way of looking at it”.

    Of course the big key to this is to be able to anticipate the cognitive leaks. It would be just as bad to spend too much effort on three different ways of representing a fairy simple and gettable concept while rushing through a tough one too consicely. The research on “explanatory dialogues” was helpful for us… It demonstrated that you can capture more than 80% of the kinds of questions people have when trying to learn computer programming, so you do not need to have an adaptive system or a natural language processing engine in order to allow learners to “ask” a question. Just anticipate the question and answer it. Imagine they, or someone in their virtual classroom asked a question, then answer it, right there, just in time.

    Again, when I advocate for a forward-only approach, I do not mean that learners do not go back to a topic, concept, example, etc. I mean that the revisits are embedded in the forward experience. No different than a real world classroom where nobody goes back in *time* even if they revisit a formerly-discussed topic.

  12. Maybe I just should have said “spiral”.

    I still consider this “linear” because it does not go back… It maintains forward flow while iterating through topics, examples, concepts, graphics, ideas, etc. Each iteration becomes deeper and richer. If we believe “linear is not how people learn”, we can expand the way we are defining “linear”. To me, a spiral design model is still linear.

  13. Specific examples/tips for forward-flow for a programming book:

    * Repeat the context for a snippet of code, a minimum of three times (more if needed). Where books typically show a code listing once then after that switch to snippets, this forces the learner to go back. Use a graphics treatment so that the part of the code being discussed is not competing with the rest of the context. For example, make the full listing grey while the currently-discussed snippet is bold, larger font, etc.

    * Make complex graphics build over time, across many iterations of the graphic. Start with just a box here with a server there, then keep redrawing the graphic adding and discussing new pieces. Use a graphic treatment to highlight the part you are currently discussing while maintaining the rest of the graphic for context.

    * Anticipate areas of confusion that create “cognitive leaks” if not addressed. If someone has a burning question, they are effectively stuck. Answer the question at the earliest possible moment, either as part of the material, or, if it’s not a question EVERYONE would have, then use a side bar or other graphic representation that suggests, “this MIGHT matter to you, but if not then just keep going.”

    * Reflect the most likely state of mind by the typical reader for each topic and sub-topic. People often go back not because they MUST, but because they do not have any way to know what level of comfort they should have with what is being presented. In other words, TELL them at each stage what level of certainty or memorization they should have. For example, “do not try to memorize these eight things… Right now all that matters is that you know there are several pieces, etc…” One of the worst cognitive leaks is when we think we ought to know something better than we currently do, and we’re afraid we’ll be hopelessly lost if we keep pressing forward. Simply *telling* them what is and is not needed let’s them feel confident that it is OK to keep moving forward. (of course, this assumes the materials WILL revisit and build on prior exposures. It it was designed with a “listen up and listen good because I am only going to say it ONCE” approach, then it doesn’t matter.

    * Use pacing and weighting of topics so that the learner has a place in the material to catch their breath, reinforce what they learned, reflect, digest, etc. If they will keep moving forward, then the “pause” must be inserted. It can look like new content while actually being more subtle ways of considering what they have already done. NO, comprehension or simple memory quizzing is NOT usually the answer here.

    * Choose graphics over words whenever possible. For difficult topics, especially tricky concepts, graphics are usually much harder to misinterpret. If the A is connected to the B via the C, the chances of the learning getting it exactly as intended go up with a graphic vs.trying to describe it in words. Again, the goal is to reduce the learner’s *feeling* that they might not have gotten it…

    * use every other shred of research about reducing cognitive load, including the work of Ruth Clark. Forward-flow is another way of saying “well-managed cognitive load.” Using evidence-based approaches like favoring conversational language (improvements in retention and transfer, etc.), embedding annotations *within* the image rather than below it or (worse) described somewhere in the page text, not using examples that require domain-specific knowledge beyond the current topic, and on and on…

    It seems that every possible attempt to reduce cognitive load. — necessary for a successful forward-only approach — becomes that much more important when using a digital text. But really, we are talking about more efficient learning. These are only a tiny handful of the techniques that (in my opinion) can and should be considered.

  14. If anyone’s still following, I offer this post from My Heeero, Nick Carr:


  15. Thanks for offering expanded information. I agree that “spiral” is still feels linear to me, but it seems to better convey the spirit of constructive reiteration vs exact repetition. Anticipating questions is a positive. I get the feel that part of what is under discussion here is a simplification of the presentation of information. I wonder if it would be possible to build in a “hover” at the end of paragraphs offering a dropdown list of associated info linked with the portion of the “presentation” in which it appears? Thank you again, it is doubtful that I will have time to return here. I run a family farm and it is difficult to check back on discussions in a timely matter.