Super (Memo) Man

Dr. Woźniak about memory and remembering.

Dr. Piotr Woźniak is the developer of one of the leading programs for “Spaced Repetition” — a learning technique that incorporates increasing intervals of time between subsequent review of previously learned material in order to exploit the psychological spacing effect. In short: Maximize memory retention. That program — SuperMemo – utilizes computers and algorithms to achieve that.

Dr. Woźniak followed the classic path of an inventor, from noticing a necessity (was frustrated by his inability to retain newly learned knowledge while studying molecular biology at age 20) to experiencing rejection (psychology journals would not publish his findings due to the inclusion of computer algorithms, whereas mathematics journals refused on the grounds that the results relate to physiology). The journey was long, but SuperMemo now has tens of thousands of users improving their educational pursuits with the help of Woźniak’s hard work and ingenuity.

Dr. Woźniak and I corresponded a while back about some of my thoughts and ideas to do with memory. 7 years later he contacted me to remark about one of our discussions. Talk about good memory! (Though I suspect it has more to do with his filing and time management system.) He was letting me know that a relevant new feature was implemented in SuperMemo, and I was letting him know that SuperMemo is very relevant to my search for technology aided learning. I really wanted to ask him some questions and he was really excited to answer them.

What is your earliest memory?

Pretending to be asleep in the kindergarten when forced to nap at a wrong time.

If you had to give just one advice, one tip or suggestion, to anyone who wants to improve their memory performance, what would it be?

I thought about it and it is super-hard to run away from just recommending SuperMemo. I thought “read” is more general than “install”, hence the original proposition. I might go for “learn mnemonic techniques”, or “never use an alarm clock”, or even “exercise”, but this would not be accurate. As SuperMemo is a commercial product, telling people to just get it might sound like a suggestion with an agenda. Perhaps a good alternative would be “learn a lot” because all learning boosts learning capacity. Small print could say “learn a lot with SuperMemo”, but those who just read a lot of books, also benefit tremendously.

How did you discover the existence of the spaced learning phenomenon? How did you start developing a method of putting it into practice?

It was an intuition. It was an intuition that MANY PEOPLE have, had, and will have. The intuition is: Once you review, you can review less often. So I went on just to measure how often we need to review. Once I measured optimum intervals for review, I started using them in real life and it was just an avalanche of ideas for improvement since then. You can guess that once I started using a computer, the acceleration was incredible. Synergies with the web, Google, wikipedia, YouTube, etc. were just adding fuel to the fire.

How much knowledge existed before you came into the scene and how much did your research contribute to the science behind it?

SuperMemo is original from scratch. However, many of its contributing ideas were developed by others in parallel, before, or later on. The most striking ones:

  1. Bjork: two variables of memory
  2. Ebbinghaus: plotting the forgetting curve
  3. Leitner: review priorities

It was fun to slowly discover all those parallels. I still keep discovering undiscovered contributions.

When do you come up with new theories? Where is your favorite Eureka place (mine is the shower, by the way)?

Ha ha. Shower is good indeed. Shower contradicts lots of my statements on creativity. Why is the sleepy mind with closed eyes still productive? Probably it consumes the work of the day while tiredness washes out all “creative pollution”. Jogging is great too. Walk-talking cum brainstorming with others is productive. But nothing beats (1) work hard on a problem, (2) get some sleep, and (3) think about it in bed in the morning (and if it does not work, Go to (1) again).

Common denominator? Lack of distraction! (shower — tiredness depollute; jogging — tiredness depollute; morning bed — sleep depollute).

This reminds me some learning advice: isolation and stresslessness.

The benefits and applications of spaced learning have been known long ago and established over many years. Are there any new findings or theories that you’re excited about that might lead to new developments and techniques?

We are working on proving that creativity can be improved through “Incremental Recall”:Imagine you had a very eventful day. You might want to sit down write about your experiences. You can sketch an outline, but you will quickly notice you cannot recall many of the details. This is normal. However, these details will later be popping up in your head in different contexts, once you get up and carry on. An image, a smell, or an article in press may all contribute to unexpected recall. If you happen to execute incremental learning in the meantime, you will notice that the process of recall is intensified. The richer the learned material is, the more associations will be produced as more memories of the day come back to you. What you witness here is Incremental Recall, and it’s a short step away from Incremental Creativity. Both are based on incrementally building a mental structure in your mind. In case of recall, you will recover past memories. In case of creativity, you will additionally add new thoughts and memories you have not experienced before, and ideas you have not thought of before. Finally, in problem solving, you will direct this creative process towards accomplishing a specific task: solving a problem or answering a question.

Can you give us a glimpse into what you’re currently working on that has the most promise?

Another exciting concept we are working on at the very moment is “Neural creativity” (if you got a better name in mind, please let me know). Neural creativity employs the incremental learning process to produce new ideas in a neural fashion. In neural creativity, closely related concepts are more likely to show up when executing a neural review of a studied subject. For example, if you are a doctor who is making a diagnosis for a specific patient, you can set up a diagnostic concept and link chief suspects in a concept map. Go Neural to see what new ideas come to your mind while learning new topics at the same time.I am super-excited about the whole idea because it feels like equipping the users with Einstein’s brain in slow motion. You can build a brain of knowledge as big as you wish. You can make your networks of ideas as complex as you wish. And then you can press just one button to start your genius train of thought at your own pace while watching all new ideas form in front of your eyes.

What is the one thing that we do not (yet) know about spaced learning, and the science behind it, that you would most like to be discovered?

Ha ha … if I knew, I would go there. I think sleep-vs-learning data is most precious and there will be a new discovery.

These days we have so many options and devices to help us retrieve information. What do you see as the remaining advantages of using “actual memory” in the age of Google and Siri?

Google only increases the need for long-term memory. Previously, scarcity of information was slowing down learning. Now, the volume is astronomical and knowledge at Google is useless without being transferred to human memory, esp. long-term associative memory. I write about this in many places. I will gladly write again because your question implies an expectation of a different answer.I have a micro-blog idea “Why SuperMemo is better than Google?” — provocative statement on why knowledge in SuperMemo is superior over that in Google when doing research (in abstraction from learning, i.e. when focusing on short-term problem solving!)

Are there different “types” of memories? For example, is the difference between remembering units, such as a string of numbers — 2397701, and associations, such as an answer to a question — the second president of the United States was John Adams?

Memories differ primarily by complexity of their mapping in the brain. Numbers are complex, presidents are simple (once you get a basic course in history).

Can you categorize the different types of memories? Are there any fundamental elements out of which memories are made?

You can find many classifications, but most are rather useless for a student, who only needs to know which knowledge matters, how to represent it well for learning, and how to remember it for years and use it well.

From the perspective of the way our brain operates and develops, is there a difference between being exposed to a piece of information for the first time and encountering that piece for the second and third times?

Yes. First you need a find the right synapses to represent a memory. Later you just fix the information in place.

There’s a saying that “the brain is like a muscle”. How much, and in what ways, do you think that is true?

The brain is a super-muscle and every piece of information learned works like a legal anabolic 🙂

The SM algorithm — let’s get under the hood… Can you briefly list the main variable at the heart of the underlying formula behind the algorithm? As simply as you can, can you explain the relationships between them?

The two components memory model — we have demonstrated the existence of two independent components of long-term memory and have established theoretically that the two variables — R and S — are sufficient to describe the status of memory in spaced repetition.

R (Retrievability) — the expected probability of recall at any time.

S (Stability) — how long a memory can last if undisturbed and if not retrieved.

SuperMemo plays on R and S to optimize learning and Algorithm SM-17, implemented in SuperMemo 17 Beta, is the ultimate practical proof for the correctness of the two component model.

You are constantly improving the algorithm, tweaking the formula. How much of the revising is made based on new data and how much of it is based on theories, speculations and understandings?

The new algorithm (SM-17) is 99% data, and 1% old theories. No “speculations” come to mind, everything is measurable and at worst, some things have not been measured yet, for scarcity of time.

What do you do with new theories and how do they grow old and find themselves in new versions of the algorithm?

The best method: dump it all in SuperMemo and let it germinate. Nothing disappears in SuperMemo, it keeps coming back unpredictably until the resolution, and sleep can work its magic in the meantime. SuperMemo is particularly precious when you work on things over decades (e.g. 2 variables of memory: proposed 1988, proven 1990, published 1995, investigated computationally 2005, implemented in the algorithm 2015). Drip drip drip, small steps, nothing revolutionary and then … new quality.

Each person is different. How much weight do you think that has for achieving success and how much does SuperMemo account for individual factors? What method can you use to assess an individual’s capacity over another’s?

There are huge individual differences, but it is super-hard to figure out which might be genetic and hard to overcome and which come from bad training. It is possible that once the “critical window” is missed, it is very hard to bring an average man to expert/genius level. Neglect may occur in early childhood (e.g. nutrition, health, abuse), or at teen years (e.g. bad math teacher). Personality makes a world of difference. I spend too much time with pretty smart people to truly understand this problem. Speaking of genetics: my own brother was a very bad SuperMemo user. However, he was a prankster and I could never figure out why. My sister was even worse, she had serious surgery as a newborn. This might have handicapped her for life. I talk to hundreds of people and there is one huge wall that 99% of people face: POOR UNDERSTANDING of how important knowledge is in life. In short: MOTIVATION!!!

You are someone who is not just promoting a product. You have a strong vision of knowledge and its part in our society and selves. Where do you think people get knowledge wrong?

Perhaps too much focus on short-term? Too little focus on abstraction? Too much cramming (vs. learning)? Too much pop learning? Too little effort to step out from the comfort zone? Insufficient methodology? Last but not least, and perhaps the worst: insufficient appreciation of the power of knowledge?Teens do not care about school enough. Students care about degrees too much. Adults do not invest enough in lifelong learning. Despite all the talk “knowledge is power”, “education is the future”, I do not see much serious commitment around. Parents will tell their kids about the importance of learning but do not bother to shine a good example or try to do some work with the kids themselves?

Which brings me to my next topic — you are also someone who is trying to champion positive change in education. What do you think about the current state of formal education?

I am a wrong person to ask. I spent 22 years in the system and live with an impression that most of the value I got from self-learning and SuperMemo, not from school. In other words, I love the idea of homeschooling, self-learning, self-paced learning, lifelong education, etc. I am sorry for all kids who go to school with great enthusiasm only to lose it in a couple of years. However, if parents cannot afford homeschooling, the lesser evil must continue.

How excited are you about the advancements, both recent and upcoming, in learning technologies?

I like all progress in all fields. As for learning technologies, I am mostly excited about new SuperMemo. It just takes 98% of my focus, so I am too biased to make a statement.

Which technologies would you bet on as going to have the most impact on how we acquire knowledge in the future — near and far?

Incremental learning! (Part of SuperMemo.)

I experimented with SuperMemo — I downloaded the iPhone app and used it to memorize the lyrics for “Don’t You Forget About Me” by Simple Minds, which seemed to me like the perfect choice. About two months later, not only was I still able to recite it accurately, it was hard to get it out of my head during the whole period (an acceptable side effect as I like the song).

SuperMemo is not a tool that will help you cram for the exam. It is for those who truly wish to learn. Use it if you are really interested in knowing stuff and retaining information for a lifetime. It should not disappoint.