Monday, February 7, 2011

E-books and the reading marriage

Those who read the recent posts (one, two) about my attempts to recall novels I'd read on an e-book reader will be interested in some little recursive reading I've been doing lately on the neurology of reading.

Pretty much what I said

Recently Jonah Lehrer at Wired has written a variety of short pieces around e-book readers and reading. He points to two pieces of research that seem to mesh with his own feeling that he doesn't remember things as well when he reads them on his Kindle.

First there is the research of the neuroscientist Stanislas Dehaene. I've been reading Dehaene's book Reading in the Brain (recommended). The neurological work of Dehaene and (many) others has found that there are two reading pathways in the brain: the lexical (or ventral) route, in which groups of letters are linked directly to dictionaries of meaning, and the phonological (or dorsal) route, in which the sound processing system plays a part. The lexical route dominates for words that are either familiar (like hat) or strangely spelled, thus more difficult to sound out (like women). The phonological route dominates for words we have seen only infrequently (like hatter) or do not know but can sound out (like coproxynaprionate). The phonological route is also dominant when words are difficult to read, as with handwriting or smudged or distorted text.

Lehrer points to research that claims the lexical route is faster and more automatic than the phonological route, which causes people to read on auto-pilot and thus forget more of what they have read. So the extreme predictability and unnatural clarity of reading on e-book readers creates an imbalance toward the lexical route. Hence we forget.

The odd couple

However, as usual, when you peel off the outer layers of exports from science to the popular press, you find things are not as simple as they seem. From what I've read so far (part of Dehaene's book and scattered papers on the web) I have learned some things. To begin with, the two routes are not mutually exclusive but work simultaneously on the same words and even trade information back and forth.

And whether the lexical route is indeed faster and more "automatic" is not fixed but a matter of debate. The dual-route theory is not the only theory of how lexical and phonological processes combine to support reading. Another theory is put forth mainly by Guy Van Orden at the University of Cincinnati. I think he used to call it something to do with resonance, but lately it goes by the name of the covariant learning hypothesis. According to this theory, well, I'll let this 1998 paper (by Luo, Johnson and Gallo, in Memory and Cognition, which by the way claims the phonological route is faster) describe it better than I can:
[A] word recognition system involves three interconnected components (subsymbols): visual subsymbols, phonological subsymbols, and semantic subsymbols. Presentation of a printed word initiates a massive spread of activation from visual features to linguistic features. After the initial spread of activation, cooperative-competitive dynamics begin among all subsymbols, and coherent structures emerge as relatively stable feedback loops. Eventually, visual–phonologic–semantic dynamics will all settle into a coherent global resonance.
That makes more sense to me than an either/or switch at the start. (Three components? Hold on, I'll get to that later.) Dehaene also says that children vary in which route they rely on most as they learn to read, most people gradually coming to rely on both. Elsewhere I read that adults vary in how much their lexical or phonological routes habitually dominate (possibly based on the conditions under which they learned to read). In the face of such uncertainty, claiming that e-book readers make us forget because they privilege the automatic route to reading ... seems attractive, but simplistic.

The collaborative, emergent view of reading makes more sense intuitively as well. I often find myself policing my own reading. Have you ever caught yourself, while reading, thinking about something else, and realize your eyes have made their way through a paragraph or even a page without you actually knowing what you had read? You fell into eye-scan autopilot, and you have to go back and find out what you had just passed over? Sometimes I scold my eyes for getting ahead of me (and they scold me right back for drifting off the task at hand).

Another trick we play is - I'll bet you've done this - I want to read something that looks enticing in the next section (great heading, or interesting word that jumped out, or weird pattern or figure in the text). But I also want to read the whole text so I don't miss anything important (or so I can say I read the whole thing, if somebody asked me to read it). So I pretend to read, and shuttle my eyes over the intervening text very quickly, essentially paying lip service to reading it. In this case it is often my eyes that do the chastising - hey, go back and read that, cheater. So I don't think it's as simple as one mode being automatic and one requiring conscious effort, and e-books pushing our one gleaming red auto-pilot button. Like old married people, our two reading modes intermingle and interrupt, finishing each other's sentences - literally.

Jumping to conclusions

The second reference Lehrer makes is to a 2010 paper in Cognition (by Connor Diemand-Yauman, Daniel M. Oppenheimer and Erikka B. Vaughan) that has been making the rounds on the web. In this paper the researchers recount how they showed people information about fictional alien creatures in either an easy-to-read "fluid" font or a messy "disfluent" font. They said:
On average, participants in the fluent condition successfully answered 72.8% of the questions [about the aliens]. Meanwhile, participants in the disfluent conditions were successful on average 86.5% of the time. This difference was statistically significant .... In sum, after a 15-min delay, participants in the disfluent condition recalled 14 percentage points more information than those in the fluent condition.
My favorite part of the article:
[I]n some cases making material harder to learn can improve long-term learning and retention... More cognitive engagement leads to deeper processing, which facilitates encoding and subsequently better retrieval... Aptly named "desirable difficulties" capitalize on this by creating additional cognitive burdens that improve learning.
I have mixed reactions to these results. On the one hand, the idea of making things harder leading to more learning and retention makes sense: the squeaky wheel gets the grease, right? Nobody remembers what the auto-pilot does; that is the point of the auto-pilot. But on the other hand, this also sounds a lot like the Hawthorne effect. Maybe the novelty of the change, or the attention implied, in itself increased attention, participation and memory. Also, this is a very narrowly defined experiment, and I would hesitate to believe any broad claims about reading and fonts based on it.

One thing that surprised me about the Cognition paper was how so many online news vendors picked it up and ran with it, with superlative headlines about "e-books make our brains lazy" and the like. (I had some links to show you, but googling it will do as well.) What I can't figure out is this. The paper makes no mention of e-books, or even screens, versus paper. It just talks about fonts and legibility. In fact, at the end of the paper the authors say:
... fluency interventions are extremely cost-effective, and font manipulations could be easily integrated into new printed and electronic educational materials at no additional cost to teachers, school systems, or distributors.
They seem to make no distinction between "printed and electronic" materials. So why has research about fonts regardless of location been co-opted to promote an anti-e-book stance? That's the phenomenon I'm most interested in.

Usually when people say that "research shows" something when it actually doesn't, they either want it to show it, or they are worried it shows it, or both. Why would people want to believe reading on an e-book reduces retention? Why would they be worried about it? Why would they see a research result in a paper when it wasn't there? My guess is that there is something else people are looking for. So I don't think the auto-pilot explanation is necessarily wrong. It's just insufficient. It's too narrow a focus. Maybe people need to study reading not just at the words-and-sentences level but at several levels above it as well. Reading ripples up all the way from the simple act of making sense of marks on a page to making decisions about how we will live our lives. If e-books make us read in a new way, shouldn't we study its effect on those ripples all the way up? And shouldn't the higher ripples impact the design of e-reading devices? Makes sense to me.

Lehrer does say one thing that connects to what I've written about e-books and context:
... I do have a nagging problem with the merger of screens and sentences. My problem is that consumer technology moves in a single direction: It’s constantly making it easier for us to perceive the content. ...I worry that this same impulse – making content easier and easier to see – could actually backfire with books. We will trade away understanding for perception. The words will shimmer on the screen, but the sentences will be quickly forgotten.
I would add that while technology keeps making it easier to perceive content, it keeps making it harder to perceive context.

Three's company

As a final note to this little rumination, I will as usual represent the under-represented when I say that I have failed to read as prescribed in every single exercise I have encountered so far in Dehaene's book.

One example I remember was that he offered a string of words and gave his reader the task of picking out which ones referred to parts of the human body. One of the words was "hare." He said, "I'll bet you took longer to decide about that word, because it sounds like "hair." I actually burst out laughing. To a synaesthete, the words "hare" and "hair" are like - well, to choose a pair that springs to mind, "bicycle" and "fish." I could not possibly mix those words together in my mind, nor take a millisecond to decide which is which.

Actually, reading this book about reading has solved a lifelong mystery for me. I have always read about twice as fast as everyone I know. I knew this as early as fourth grade when a reading specialist came into the classrom one day. He set up his apparatus, then asked us all to watch as words flashed on the screen. We were to hold up our hands, then put them down when we could no longer follow the increasingly rapid word changes. I got absorbed in the words - I think they told a story - then was disappointed when the machine suddenly broke down and stopped showing words. Still holding up my hand, I looked around to find (to my horror) that everyone in the room was staring at me. The machine hadn't broken down. The test was over. It couldn't go any faster than that. This was one of the more traumatic moments of fourth grade, surpassing being sent to stand in the corner for an entire hour for whispering in class.

Well, now I know why that broken-projector thing happened. It is the same reason I failed Dehaene's tests. I don't have two reading spouses, I have three. Synaesthesia is the third spouse. (Rather than start a polygyny polyandry discussion let's just say there are three genders, okay?) This, I think, must be the system of "visual subsymbols" referred to by the Luo et al. paper I quoted from above. At least that's the only reference I've seen to three possible reading partners.

But the third spouse in my reading marriage is choosy. When the colors/sounds/shapes/tastes/emotions of the words and sentences don't cohere, she stomps off angrily and abandons the effort. I call this sort of writing "cardboard" writing and read it slowly and painfully. I have even avoided whole books because they tasted like cardboard. Ralph Stacey is one author with excellent ideas unfortunately clothed in cardboard words. My favorite writing is done by fellow synaesthetes, but it has the opposite problem. The works of Nabokov are so engaging to the third spouse that she becomes overwhelmed with emotion and has to take breaks to calm herself. With coherent but not overwhelming writing, the three spouses collaborate splendidly, helping each other with every little problem.

The three spouses also collaborate when I write. I constantly make little jokes in my word choices of which most people are probably not aware, or only subliminally so. Once in a letter to my brother I wrote that I was having a "rapid spasm of panic" - a joke because in fact that series of words, in my synaesthetic system, is a paragon of calm. (I of course chose the word "paragon" because it is a paragon of calm itself. If I had wanted to make a joke, I would have said it was the epitome of calm.) I've seen other people make these jokes too - there are some real howlers in Nabokov's writing.

So, what is the reaction of the third spouse to e-books? I can tell you instantly that she does not like them. Looking through that little window puts a tinge on all the words that saps the saturation out of their colors and the expression out of their emotions. The window may be translucent, but it is not transparent. The third spouse does not leave the picture entirely when I read e-books, but she does mutter under her breath a lot. And for me, the third spouse is a big part of memory. When she does not participate fully, I have a much harder time remembering what I have read. I often remember stories, articles, names, as much for their physical natures as for their meanings. If you ask me the name of a researcher who studied something, I am more likely to tell you the color of the name than anything else about it (though often I don't actually say the color because it means nothing outside of my head). Some books are easy to call to mind because they smell like freshly baked bread, and so on. With this understanding, brought on by my failure to read as Dehaene expects, I begin to understand much better why I want to abandon e-books.

What is your point? Asks the impatient reader. Synaethesia, while mildly interesting to most, has a brief interest life-span, after which it becomes, apparently, excruciatingly boring. My husband at this point practically runs away when I mention it, poor man. Imagine somebody listening to you drone on about the many ways in which your lawn mower works, or doesn't, and you will understand. [Update: Said husband read that and said No I don't, in an inaccurately-represented sort of way, so I take it back. .ti noitnem I nehw yawa snur eH] In one of those little joke-paradoxes the universe loves so much, the joining of the senses that is synaesthesia has a socially isolating effect on those who have it. And there is also the problem that whenever you try to talk about it you seem strange, thus unreliable. Nobody listens to what the talking dog has to say.

So, that point? Yes. It is that as people study synaesthesia more and more, they find that more and more people have it. To someone who has been following the field (even contributing to it, as an occasional subject) that means something. Sure, some people have it in spades, as I do, but many others have it in thimbles. It may influence far more people than researchers now understand. I think the prevalence has to be higher than reported, because otherwise the percentage of cardboard writing would be much higher. (I put it at about twenty-five percent.) And I have a suspicion that at least some people might be getting my word jokes. So the probability that you have a third spouse in your reading marriage is most definitely not zero, and that third member is probably not enthusiastic about e-books. (Sadly, Dehaene touches on synaesthesia in his book only in the typical, talking-dog way, saying only that some strange people are "convinced" they "see" colors when they read. Opportunity squandered. Anybody else like to take this up?)

Now, being a software designer of sorts, I immediately begin to wonder how e-book readers could improve on this situation. My e-book reader (the Sony Reader) has what is called e-ink, as do most. It looks astonishingly like a printed page ... but somehow it doesn't either. The third spouse is not content. Because of the complexity of the brain, I'm not sure it is possible to find first principles with which to design for sense mixing. I think it must have to be done on an experimental basis. Because sense-mixing happens below the conscious level we are not fully aware of its rules. We may just have to do what parents do with their babies: keep trying things until it stops bawling and starts cooing. And different babies coo at different things, so we may well see e-book readers disperse into a multi-dimensional array of what-works-for-some approaches. Given what I've seen in the way people read and write, that is what I'd be least surprised to see if I jumped forward in time twenty years.

By the way, I made up coproxynaprionate. You googled it, didn't you? Is that healthy? I'm just asking.

Wednesday, February 2, 2011

Whose truths are these?

I keep seeing people arguing and speculating about the creation, history, authorship and ownership of the Cynefin framework, and I feel increasingly obliged to say something about it. To keep quiet seems an injustice to all parties. So I'll answer a few questions I keep seeing come up, once and for all, to the best of my recollection. Warning: these are my memories, this is my perspective. It cannot be the same as anyone else's, even if I try. (To those who have no idea what this means, hold on; a new post about e-book readers is simmering and should be done soon ;)

So, am I the co-creator of Cynefin or a contributor? That depends on what you mean by Cynefin. If you mean the form of Cynefin that existed before I started working with Dave Snowden at the IBM Insitute for Knowledge Management (IKM) in April of 2001, no, I am certainly not its co-creator. I believe Dave showed a version of these early Cynefins in his "history of Cynefin" paper. In the earliest versions of Cynefin it related mainly to learning, KM and communities. This is why in my post about Cynefin last summer I said I had helped Dave with "his" Cynefin model. To be honest, I had a very hard time writing that word "his." It took me months to do it, but I'm glad I did. It was right. Dave did have something he called the Cynefin model before I started working with him, and it would be churlish to pretend what he had before I came along didn't matter. It was not what it is now, but it was something, it was something good, and it was his something.

Now, if by Cynefin you mean the version of Cynefin in the 2002 "Complex Acts of Knowing" paper, I consider myself in between co-creator and contributor. Dave and I had been working on the model together for a year before he wrote the paper, and I can see signs of my influence there; but I can also see that much of it is Dave's own thinking. Our worlds were beginning to leak into each other, you could say, at that point. I first drew my seeing-eye pyramids in April or May of 2001, and that as well as many discussions and experiments had an impact on how the model evolved. But we each had our own separate thoughts too. (I had been working in the areas of narrative and complexity for some time, in graduate school and later at IBM Research, and was interested in where these worlds came together.) As I recall it, our first presentation of the framework in its known, knowable, complex, chaotic form, with the bubble domains, was at an IKM member workshop in June or July of 2001. Dave and I collaborated heavily before, during and after that workshop and several others in the fall of 2001, and much of what became the stable Cynefin arose during that collaboration. But I did not write any of the 2002 paper. At the time it never occurred to me to ask to be included as an author (as a contractor I was far too low in the social hierarchy at IBM to even consider it). There is a reference in the paper to a "Snowden and Kurtz" paper on social network stimulation, which I did write but which never got published.

Finally, if by Cynefin you mean the form of the framework that appears in the 2003 IBM Systems Journal paper, yes, I definitely consider myself its co-creator. What is in that paper is substantially the same as what we presented and used in 2001, but there were small refinements during those two years as we began to collaborate more and more strongly. The 2003 paper was actually cut down from double the size for publication and it involved a pretty equal contribution of ideas - and one of the best back-and-forth collaborations I've ever experienced - over months. I did write most of the text, but that's only because I had more time to write because I wasn't flying around giving keynote speeches, which I hate doing.

People have told me they wondered whether it means anything that I'm first author on that 2003 paper. It does to me. My recollection is that Dave offered me first authorship to make up for the fact that he was able to make many keynote speeches and present the framework in person to many audiences, who naturally assumed it was his own. I wasn't very comfortable with that imbalance, and said so, so we agreed that I would be first in print and he would be first in presentation. Over the years I have consoled myself for the many mentions of "Snowden's Cynefin framework" with the fact that I am the first author on the most often cited paper about it. It is silly of me to care about it, I know, but I do care, and I appreciate it when people get the author order right.

As I wrote about last summer, at some point I decided to take my toys and go home, and my old/new confluence framework stands alone. It's happy that way. The collaboration was fruitful for a while, but it ended, and Dave went on to collaborate with other people, as I did.

So, take your pick. If you want me to be either "just a contributor" or a full co-creator I hereby give you permission to believe either you like. You can of course follow this exercise through for every other person who has had an impact on the Cynefin framework, including maybe some I don't even know about. This is probably true for any framework that has ever been published. Several people have already helped me with "my" confluence framework, in discussion and feedback. Are they collaborators, contributors or co-creators? Sure, yes, whatever you like. The distinctions only make sense if you believe in owning ideas at all, which I don't.

To repeat myself for the tenth time, my view is that ideas are like whales that let us swim along if we behave ourselves. Part of behaving ourselves is not putting ropes on them. They might seem tame for a while but that's just an illusion. Wait a few hundred years and see if that rope is still there. Did Darwin discover natural selection all by himself? Was Bell the inventor of the telephone? What did Tesla do? Who built the first computer? Depends on who you ask, doesn't it? So why ask? I'm more interested in the whale.

Now, having handled the trivial issues, here is the only question I really care about. Who is the true first creator of the Cynefin framework? Again, it depends. If by Cynefin you mean the thing you draw that way with those names to it, it is Dave Snowden and nobody else. But if by Cynefin you mean the essential truths it conveys, it is not Dave or anybody born this side of Tutankhamun, but countless and nameless ancient people who saw what human life was all about and passed it on for thousands of years. And maybe it goes even deeper than that ...

I wasn't quite ready to write about this yet, but I've been reading an excellent book called The Master and His Emissary by Iain McGilchrist. It is about the way our left and right brains process the world. It is Cynefin all over, or confluence all over, or the medicine wheel or anything you like. The left-brain right-brain stuff has settled down from its first outlandish claims to a more nuanced understanding, and McGilchrist summarises the research very well. To put it briefly: our two brain hemispheres have different responsibilities, approaches, personalities almost when it comes to dealing with reality. The right hemisphere (left hand) is vigilant and handles breadth. It watches for predators and conspecifics and large patterns in the environment. It watches things emerge, seeds desired patterns, disrupts undesired patterns, thinks out of the box, considers all angles - get it? The left hemisphere (right hand) is intent and handles focus. It picks out which grains of dirt are food. It categorises and sorts, finds best practices, repeats experiments, tests, proves. These two ways of looking at the world work together to help us survive and thrive, and they go back all the way to the birds and fish.

A few tantalizing quotes from the introduction (only because it's the only thing I can find online to copy and paste from - much more meat in later chapters):
One of the more durable generalisations about the hemispheres has been the finding that the left hemisphere tends to deal more with pieces of information in isolation, and the right hemisphere with the entity as a whole, the so-called Gestalt...
It seems that they [the two hemispheres] coexist together on a daily basis, but have fundamentally different sets of values, and therefore priorities, which means that over the long term they are likely to come into conflict. Although each is crucially important, and delivers valuable aspects of the human condition, and though each needs the other for different purposes, they seem destined to pull apart.
...the relationship between the hemispheres does not appear to be symmetrical, in that the left hemisphere is ultimately dependent on, one might almost say parasitic on, the right, though it seems to have no awareness of this fact. Indeed it is filled with an alarming self-confidence.
"Alarming self-confidence" seems an apt way to describe the rigid structures of best practices and all of the business acronyms we love to hate.

So what I'm thinking now is, maybe this is what ties the grandmother medicine wheel and all her grandchildren together. Maybe our truths are not truths about what is out there but truths about what is in here. Maybe this is what we are, and what we are is what makes our world what it is. I have yet to finish reading this (long but wonderful) book, and will write about it when I have something to say about it, but I wonder if the grandmother medicine wheel may have her own grandparents who came long before her and started the whole thing going. Now that's ownership worth exploring.