Tuesday, August 26, 2008

Giving Yourself to the Algorithm

My mind is going in a thousand places at once at the moment, but I feel like I should put down a little of this on paper/electronic bits.

So, as I've said before, my undergrad background is in philosophy & religion and psychology. I'm in grad school now for a masters in religious studies, but am splitting my time between that and cognitive sciences, as I'm pursuing a Ph.D. in cognitive sciences, probably cognitive psychology, but I'm looking at places where most professors are in the field of neuroscience, as that's the area I'm interested in, and cognitive psychology and neuroscience are trending towards merger anyway.

The areas I'm primarily interested in are perception and memory. These are intimately linked fields, and also tend to go hand-in-hand with issues of learning. Cognitive psychologists since Hermann Ebbinghaus have studied memory (often by rigorous self experimentation), and have come to some surprising conclusions - the spacing effect being one that interests me a great deal. Basically, this is an important combination of experimental knowledge about the learning curve and forgetting curve. This tells us something very important about how best we learn. Combined with the insights of other cognitive psychologists like Elizabeth Loftus and Robert A. Bjork, we should by now have actually implemented some of what we have learned in our methods of teaching.

Have we? Oh, hell no.

You see, cognitive psychologists, historically, have lost out. We haven't been listened to all that much. Part of the problem is that while we have excellent experimental evidence of how best to learn and retain information (remember just before you forget something), it is impractical (or at least incredibly hard) to do this outside of a laboratory setting. "Just before you forget it" sounds like a very vague statement, and is indeed difficult, at first, to quantify. Ebbinghaus paved the way, and we've refined it since then. The answer to our problem, as it has been often before, is to use computers. A complex algorithm can calculate this curve for an individual relatively easily. At that point, it is merely a problem of sticking to a regimented schedule.

Now, briefly, I want to clarify what I mean by "learning" and "knowledge". Knowledge is a thorny topic, as I'm sure The Rooster can tell you. Here, I'm going to avoid the problems of the word itself and simply refer to any facts/procedures/information that you take in. Thus, in this case, knowledge need not be "true." It is simply information which you are interested in taking in, storing, and recalling. In case you're wondering, that's basically what I mean by learning here - effectively taking in information, storing it for as long as need be, and then being able to recall it at will. Now, let us think about this...

Most people have little problem with taking in information. In the absence of learning disabilities, such as dyslexia, we have hardly any problem with taking in information like addition tables, multiplication rules, grammar rules, etc. Storing information was thought for a while to be possibly a problem...however, we know are getting increasing evidence that the actual storage capacity of the brain is not something we need to be concerned about. It is not infinite, by any means, but large enough so that we needed be concerned with it. This has largely to do with the way that the brain forms connections between neurons - something that we call long-term potentiation. Increasingly, we are seeing that there is an interaction between patterns of firing and specific neurons firing. This is incredibly over-simplified, as I don't want to get into the neuroscience of memory or recognition at this moment, and would rather stay at the more abstract psychological level.

So, our problem is really recalling information (isn't it always?) It is unlikely that once something is actually committed to long-term memory (this is a psychological short hand for a lot of complicated neurobiological activities) that we lose that storage. Instead, we lose the ability to recall the information (again, psychological shorthand. This is why in the modern day, you really have to study both to speak with any kind of knowledge about the subject, and design some very interesting studies). Now, long-term potentiation depends upon the repeated simultaneous activation of communication between neurons, so, obviously, repetition helps. We know, however, that pure repetition is not enough. Constant repetition is merely keeping the information in your working memory (this usually involves a different brain region), the way you repeat a phone number to yourself before you dial it. If anything interrupts you, for even a second, you may completely lose the number. Long-term memory is different, though teasing the two apart, and how exactly something gets from short-term memory to long-term memory is quite a complicated process. Look up the history of this type of work, it's fascinating.

So...we have been almost blasted fools in our teaching methods for years. We know that cramming doesn't work. We know that with a careful schedule, we can use the spacing effect to our benefit, but it's hard. It takes dedicated and individualized effort, but it's possible. Most teachers do not do this, because it is so impractical and hard to implement. And it takes a lot of individual effort on the part of the student. Also, there is a huge movement to say that "mere memorization" is nothing. Here's where you can take an interesting tack - memorization is learning in quite a few situations, and forms the bedrock upon which more important theories and creativity can be built. For example, you have to memorize the alphabet and a certain amount of words in your vocabulary to be able to read. If you did not, you would have to look up every letter and every word to process this post. Let's be thankful for that memorization, at least. When you memorize something in this form, it becomes almost automatic in recall. It is important to differentiate between spoken and written language in this case - different neuroanatomy comes into play. Likewise, mathematics becomes an automatic skill. The further you progress in math, and the more you use it (the same with reading), the more automatic it becomes. It is the same with just about any body of knowledge. The less you use it, the more it fades into a state that is difficult to retrieve.

For instance, I know Latin, Hebrew, a little Greek, and am learning German. I find if I take a month off from reading anything in these languages, it takes quite a while to get back into any sort of fluency. If I take a summer off, then I have to go back to remind myself of some basic grammar rules before I can read them. A year, and they seem quite impenetrable without further reflection. Wouldn't it be wonderful to just be able to recall anything at will within a few months.

Well, I think that this is slowly becoming possible. Check out this site, SuperMemo, a program created by Piotr Wozniak. Wozniak has undertaken an extreme form of self-experimentation, in the form of Ebbinghaus. And does it work? If you stick to the schedule, yes...with amazing results. Check out the May 2008 (16.05) edition of Wired Magazine for a treatment of the subject.

Wozniak has made a breakthrough of sorts - the realization that the brain operates in a way similar to computer networks (vastly more complicated, and neurons are not really like semiconductors, but they do perform some complicated integration functions and can operate like complex networks of boolean gates). By breaking down the learning process to an algorithm, and using a computerized helper, you can train yourself to learn better. By giving yourself up to the algorithm, you can take use the mechanical interactions of your brain to your advantage. This sounds mildly religious (in the John Dewey sense), and in that way, it probably is. You have to give significant time and ritual to it, if you want it to work. The problem is that this will significantly destroy your chance of ever having a "normal" social life.

This is a departure from what most people thought computer-aided learning would be. The idea was that we would use machines to store all the information that was trivial to us and free up space in our brains for real learning. The problem is - what are we using that space for now? Not much, in most people's estimation. Computers have freed up our time, but we haven't progressed much in learning, unfortunately. Wozniak wants to turn this around - use the computer to boost our own natural abilities, our own brains. Once you accept that the mind and the brain are really the same thing in basically every "conscious mind" process you want to talk about, then you can learn to use this to your advantage - all the messy, mechanical brain processes that contribute to our consciousness and learning.

It takes dedication and effort, but it is possible...if you're willing to give up on other elements of your life.

Any thoughts?


Okada said...

What do you know about the learning process of children? Obviously we use the English language on a daily basis so it is probably the best example of repetition and using that information to retain it. But I find it hard to believe that if I go a month without using it I would lose any of it at all.

Going on the premise that I wouldn't lose any of the language after a month of zero use.

It is commonly thought that children soak up information like sponges, hence the learning a language gets harder as your get older phrases.

Do you think that would be due to me learning it as a child or just having it so ingrained from previous repetition and use?

If children are better at learning than adults do you think science will be able to harness that and turn it on for people still learning in school?(I know it must be a more complex issue than a switch on and off but I hope you understand what I am asking.)

Overall a very informative post. The brain has always fascinated me.

DirtyGaijin said...

I'm no expert, so this is just my take on your English idea. Maybe the quantity of repetition is related to the time retained. A month may not be enough time to forget it, but with a little more time it certainly can dwindle. Remember Yoshiko? She began forgetting Japanese after a year or two in the US.

But yeah, the post is interesting. Curious to check out this program a little bit.

Jason said...

I've been very keen about how I learn because of all of the god-forsaken exams that I had to 1. memorize, (2. integrate?), 3. spit out/(reflect on?) later. For me, brute memorization does work when it comes to lists, facts, vocabulary and other atomic pieces of information that can be put into multiple choice or fill in the blank answers. I've gone back to much of the material I've put into my long-term memory before and do have a difficult time recalling it at level worthy of taking a test.

Overall, I find that the more reliable way for me to retain, recall, and integrate new information is not through memorization but through narrative and explanation - long trains of thought with facts, interpretations, opinions, etc. that give my memory a complex body of connections with other things that I know.

Now, having put all of that down for the past months, I haven't so much forgotten it as can't recall all of the pieces in as much depth on the fly. I do have to go over my notes, reread texts, and just spend time thinking about it again. It's much the same with German, though a bit more difficult because it's been a year. Long-term memory recall seems to be less about losing the information than about not being able to recall it as easily, quickly, or accurately as when you were in the middle of learning and using it.

I think I'll try the memory program if there's an accessible copy of it. I do have a social life as Ragoth knows, but I'll see what I can get out of it and report back.

Jason said...

I downloaded the supermemo 98 (freeware!). It has a pretty easy learning curve (just start by hitting the "learn" button and a tutorial starts). Unfortunately, a lot of the pre-programmed topics (language, vocab, grammar, economics, etc.) that are usually available through the site are being re-uploaded. So, I've given up today on trying to do anything more with it.

Ragoth said...

Let me try to offer a few brief comments now, I'll fill in more later.

First, I wanted to avoid language, as it's a messy problem that makes this whole discussion even worse. Writing and reading are better examples, because they are obviously acquired behaviors. No one knows how to read straight from the womb. This is not to say that you know how to speak from the womb either, but there is good evidence that there are deep structures of the brain that aid in language acquisition. Basically, you come primed to learn a basic grammar and vocabulary, which you then practice and expand constantly, every day of your life. But, you're pretty much hard-wired to pick up a language. It's similar, but not identical, to learning in the way that I've been discussing it. A better example would be odd words that you don't use every day. SAT vocabulary words, for example. If you don't use these words frequently, you do forget them.

(side note - the process of learning, as I stated, is not fully understood at this time. However, we have evidence that long-term potentiation [increase in neuron synaptic connections] and probably mylenation [increase in the fatty tissues that surround neuron axons] contribute to learning in significant ways. Brain plasticity decreases as you age to some degree, and major culling of synapses begins in adolescence, so the greatest number of synaptic connections [which may be your best chance of learning something - may be] is just before then. So, for a lot of people in this country, you learn only one language in this time period, making the acquisition of a second language a much harder process. To have true fluency, you probably want to be exposed to and learn a language before that stage. Of course, there's also the problem of the brain not fully developing until a bit later...oh, it's a really complex problem. If you want, I'll try to do a brief write up of some of issues in memory. Short answer is - we don't have it all figured out yet, but we've got some really interesting stuff going on right now.)

Gaijin basically has the right of it - the forgetting curve never goes away. You can only hope to level it off to such a point that it would take a very long time for you to forget it. This is the point of the spacing effect - you do have to constantly re-present the information, but the interval becomes longer and longer. Any trigger after a certain point will probably allow you to recall the information, but you want it to be automatic.

Jason, you likewise have a very good idea. Rote memorization is not helpful for more complex ideas. There are all sorts of ways that cognitive psychologists have discovered that we can help along our own learning. One of them is through active encoding, which is basically what you describe. Create a story, a visual scene, anything to aid you in identifying and deeply embedding the information in as many ways as possible. The forgetting curve levels off further with these added techniques, but, unfortunately, is still always present, as you likewise agree. So - an ideal technique would be to use the spacing effect to your advantage, while also using active encoding of the information. The more trails and trains of thought you can leave yourself, the better your chance at recall in the end, which is always the goal. Note, this is still considered memorization in the cognitive sense, just not how we often use the word, by which we typically mean the rote-repetition of facts.

Also, let me know how SuperMemo works out. Can you link me to the freeware version you found?

Finally, Okada, yes, we are now able to implement better learning techniques in training children. Most important of all would be to teach them how to think, and how to learn. These are two very important issues that most people have no education in whatsoever, which leads to a lot of people bombing out of more advanced course work. They have the capability to learn, they just haven't trained themselves how yet.