Write to Super Memory Guru |
Do people with higher IQ benefit more from SuperMemo?
Answer:
People with higher IQ are more likely to find clever uses for SuperMemo and they are
usually faster to grasp the principles of the program. They are also more likely to become
addicted to SuperMemo as one of their most important applications. However, recent research indicates no significant correlation between
IQ and any of 30 major studied learning parameters used in SuperMemo
It is well known that people can be divided into late sleepers, owls, and
early sleepers, larks. Has there been any research to indicate what sleeping
type has better memory?
Lifestyle and
personality are critical here. Owls may show lots of
excitement for learning as this excitement is the main factor that
makes them
owls. On the other hand, larks can make better use of early morning
hours where
they can study in quiet at the time when their brains are most
refreshed. The
formula for best learning is then (1) to go to sleep in accordance with
one's own body clock (i.e. when actually sleepy), (2) to get up
naturally (i.e. without an alarm clock) and (3) learn at the peak of
one's alertness (which in a natural rhythm should come shortly
after awakening).
It is harder to learn foreign languages after forty
Question:
I
have been learning English for several decades, but I still speak
poorly. My IQ is well above average, I am not lazy, and otherwise I
have a fluent command of the language. I learned Esperanto in a few
weeks and speak good German. But speaking English is always a problem.
Any suggestions?
Answer:
Over the years, you have certainly acquired
phonetic, lexical and mnemonic representation patterns that deprive you
of rudimentary building blocks that enable fast learning of languages
other than your native one! If you
play with the Polish word "start" in your memory, you use a
very scanty synaptic representation that makes processing fast and
retention easy. If you want to use the word
"start" in English, suddenly you are faced with a burden of
connotations that make learning quite frustrating. The mere
pronunciation pattern may require weeks of effort devoted to a single
vowel. You need specialized circuitry to process this single vowel,
extract it in the flow of speech, pronounce it in different contexts,
skip it when possible, and emphasize when necessary. Your language
toolkit has already been molded to fit your native language. It is not
true that you learn slower than young kids. Your main obstacle is to
unlearn hundreds of neural dogmas about the language. Then you need to
develop anew specialized circuits for processing building blocks of
speech. The shortest path to resolving this problem is to find people
to who you will be able to speak English and use the language as often
as possible. Fill up the gaps in knowledge with SuperMemo. Each time
you pause on a word or phrase while its Polish equivalent is ready to
serve in your memory, store the appropriate association in SuperMemo
(e.g. Polish word
vs. English word, and later, English word definition vs. English word).
Only with months of intense training will you be able to overcome
mental obstacles that also grow in your mind with each frustrating year
of limited progress. All in all, it may appear easier to develop a
fluent command of spoken English that to learn to decode various
dialects of English from all corners of the world. Again untold hours
of training are needed to achieve proficiency. You may
greatly simplify your task if you definitely decide not to struggle to
get rid of your Polish accent. International communication needs to
strive to never discriminate between individuals on the basis of their
accent
There is a physical limitation on how much we can learn per day
Question:
Alan
Baddeley's 1976 book "The Psychology of Memory" (Chapter 2
"Input Limitations")
cited an experiment in which
postmen were taught to type on a variety of schedules. The results
strongly indicated that
there is a definite limitation, probably physical concerning the
consolidation of the memory
trace, to how much can be learned in a day and that too much effort led
to negative returns.
Are you aware of any research quantifying this limitation, which is
distinct from the more pragmatic
SuperMemo limit on the amount of review necessary for too large an
input. How many items, inputted for the first time, outstrips the
brains
capacity to consolidate them as memories?
The postmen experiment was described in
"The Psychology of Memory" and again in his more recent "Your
Memory, a User's Guide".
Baddeley's experiment showed that of 4 groups of postmen who learned to
type
over 60 hours, the one hour a day group attained a far higher skill
level than
those who practised 4 hours per day, 2 hours per day in one session and
those
who did 2 one hour sessions. In fact, the data shows that the 4 hour
group must
have experienced negative returns at some point. He describes a similar
result in a wartime experiment in groups learning Morse code. This is
odd, in that in "Genius Explained" by Rowe a
common finding in individuals of very high ability is that they have put in
typically 10,000 practise hours before virtuoso performance is achieved. They do
not work "little and often" as the experiments would suggest.
Baddeley hypothesizes a limit to the daily learning rate and suggests it is
based on time need to physically consolidate memory traces. In practical terms,
it is very important to be able to quantify this limit in order to have
efficient learning and practise.
Answer:
There is indeed a limit on how much we can learn per day.
However, with good strategy, there does not seem to be a limit beyond which we
should stop learning.
First we need to draw a distinction between procedural learning and declarative learning. Procedural learning is used to acquiring a skill such as riding a bike or typing the keyboard. In procedural learning, we do not tell the brain exactly how it should perform. The brain provides "the answer" on its own by trial and error, while we only "approve or disapprove" of its performance. In declarative learning, as in memorizing a textbook, we tell the brain exactly what to learn, and expect it to encode information in memory. Procedural learning, by definition, is highly repetitive (you repeat the same moves again and again, only with a slightly improved precision). With declarative learning, we want to minimize the repetition. For those reasons we need to discuss the two types of learning separately.
Procedural learning
In procedural learning, as in the typing postmen experiment, the following factors may result in the need to stop the practise at some point:
If you see virtuosos practicing all day long and getting excellent returns,
it all comes from the mastery of the learning procedure itself. A top-class
violinist will counter-act all factors that put the limit on the recommended
amount of procedural learning. She will switch the routine to avoid diminishing
returns. She will keep her entire upper limb motor system in excellent shape to
effectively avoid overuse injury. Most of all, she will master the art of attention and
capitalize on her passion for the violin to avoid the effects of un-learning.
Consequently, there will always be a natural limit on her daily progress, but there
may be no limit on the amount of training other than the need to take breaks
(e.g. for lunch) and the need to sleep (to consolidate the results of
learning).
Declarative learning
As in procedural learning, there are factors that will limit the efficiency with the progress of learning on a given day:
In declarative learning, unlike in procedural learning, we can accomplish maximum speed of learning if we minimize the review of the material with a view to a selected level of knowledge retention. For each selected retention level, there is an optimum review schedule that can easily be computed (as in SuperMemo). Maximum speed of learning is reached if we allow 20-25% of the material to be forgotten by the time of review. For higher retention levels, knowledge acquisition rate is slowed by the rapidly increasing frequency of review. For lower retention levels, knowledge acquisition is slowed by forgetting. In practise, for excellent long-term results we elect to remember anywhere between 80 to 98% of the learned material. For most important material, we choose higher retention at the cost of lower knowledge acquisition rate (i.e. learning speed).
The existence of the optimum schedule of review and the optimum knowledge representation makes it possible to determine the natural limit on the amount of information acquired in unit time over lifetime. This limit has been variously estimated in the range of millions of individual pieces of information (e.g. flashcards, questions&answers, items in SuperMemo, etc.).
As it was the case with procedural learning, a skilled student will also be able to spend long-hours on declarative learning without suffering negative effects. He will use spaced repetition to optimize the learning schedule and avoid spacing effect (SuperMemo). He will formulate knowledge for learning in the way that will minimize interference, minimize fatigue, and maximize consolidation (see: 20 rules of formulating knowledge). Finally, he will take the break from learning with increasing fatigue and never neglect sufficiently long and sound sleep that is the best warranty that the day-long learning is optimally consolidated for long-term objectives.
Conclusion: the key to overcoming the limits on the learning rate and the daily amount of learning is the learning skills. Those vital skills include: understanding optimum knowledge representation (how to formulate questions, what training procedure to choose, etc.), understanding optimum repetition spacing (when to make the review), understanding fatigue and sleep, passion (love of learning), overall fitness, time-management, stress-management, etc. All these skills are discussed in detail at supermemo.com
Question:
We are 7-th graders and work on a project related to memory. We have a question to Dr
Wozniak: "Why do we better remember pictures than word combinations?"
Answer:
In the course of evolution, humans practiced visual memory a lot. They did not deal
much with math or abstractions. That is why there are parts of our brain built specially to serve
visual memory. As you know, evolution gives better adapted individuals a better chance for
survival. Those who could remember better, e.g. shape of the prey or enemy, could survive
better, and pass their "good" genes to the next generation. Calculating a
differential was not needed in apes or early humans. That is why evolution did not built a
specialized calculator into our brain. It has, however, built a calculator for processing
visual data. You "type in" the picture, and get a short answer:
"danger!" or "food!". Those simple signals are easier to remember than
... streams of bits of a complex image. Evolution and memory are fun, aren't they.
Forgetting is a decay process that can wipe out memories for good
Question:
Some psychologists believe that forgetting is not a decay process. Although it
may seem a memory cannot be recalled, it is still there. For example, you can
recognize an item in a group. Or your re-learning time is shorter. With disuse
memories become inaccessible but not gone. Does SuperMemo account for this fact?
Answer:
We need to differentiate between memory stability and memory retrievability
to show that memories do fade entirely. Retrievability case is simpler and
obvious. Exponential decay is a universal phenomenon in biology and physics.
This is how memory retrievability declines. This can be seen in SuperMemo in
forgetting curve graphs. Decline in retrievability is responsible for making
memories inaccessible at some point. The harder question is what happens with
memory stability. The problem is that with decline in retrievability, stability
can no longer be measured. Once we lose access to memories, we have no way to
test for stability leftovers at the synaptic level. Those leftovers are
definitely there. We can see that in spaced repetition. Stability of memories
keeps increasing with each review. Even and especially then when the memory is
on the verge of being irretrievable. However, the evolutionary purpose of
forgetting is to free the storage for new memory traces. Forgetting and spacing
effect work together in optimizing the memory storage. If synapses sustained
their stability after "forgetting" (i.e. when memories become
irretrievable), the brain would gradually lose its original optimization
capacity. New memories might become very easy to establish if participating
synapses had already been stabilized. That would make the brain a progressively
more conservative learning machine. If this "becoming conservative" actually happens in reality, it is
more because of previously established memories (i.e. states that involve both
retrievability and stability). Not because of the previously established
synaptic stabilities alone.
If there was no decline in stability, you should be able to observe the fact in SuperMemo using the first interval graph (this graphs shows both newly memorized items as well as items that have been re-learned). In a well executed learning process, the length of the first interval keeps decreasing for items with each successive memory lapse. Forgotten items are in no way favored in the process of re-learning. In a vast majority of cases, in a massive learning process, forgetting a simple item will require re-learning as if from scratch. For more complex items, this is less visible, as these are often "partly forgotten" making facilitated re-learning possible.
Passive recognition is easily explained at the network level. While memories decay, individual synapses may no longer provide access routes to a specific network. In other words, a desired configuration of activity cannot be evoked in active recall. However, passive recognition can still be possible as it may require a tiny subset of original activity to trigger recognition. After all, the recognition stimulus itself provides activations that are able to plug the gaps in memory. Similarly, re-learning of partially forgotten material is explainable by the same mechanism. As long as the portion of the potentiated network is retained, the original configuration of activity can be restored and produce re-learning that capitalizes on previously produced stability of memory traces in individual synapses taking part in the network. However, once the decay process proceeds far enough, neither recognition nor facilitated re-learning will be possible. Memories will be gone for good.
How does SuperMemo deal with re-learning? It modifies the estimated difficulty of the forgotten item and enters it into the learning process anew. As if never learned before. Measurements show it very clearly, for a simple well-formulated item, forgetting nearly wipes out the benefits of previous learning. If you look at the forgetting curves for re-learned items, they are as steep as for newly learned items. Those items cannot capitalize on previously earned stability. Each time they are re-learned, they form a new pattern of activation. The process of building stability must then begin from scratch.
Keeping things in memory is important! Memorization allows of employing associational thinking
Question:
What's
the difference between persons A and B: Person A can recall thousands
of facts in 5 seconds from memory. Person B can look up those facts in
30 seconds from the net. Clearly, person B is more effective, because
he/she hasn't invested 30 minutes every day in appending the facts and
then repeating them with SuperMemo
Answer:
Person A can produce instant associations between all those thousands ideas. If the number is 10,000, he or she can produce
10,000 x (10,000-1) instant associations of which many may appear useful (depending on the actual content of memory)
Person B can produce only as many associations as his short-term memory can hold while browsing the sources (say
50 x 49 in case he can keep 50 things at once in short-term memory)
(Dawid Calinski, Poland, Apr 6, 2001)
Question:
Why does this site not mention learning in the relaxed state? What is your
opinion about alpha waves in learning? Do you recommend products like BrainWave
Generator, Hemi-Sync, Holosync or Polish Sita biofeedback?
Answer:
We do realize that the proper cognitive environment is paramount for
learning. However, for clarity we use the term concentration instead of
an all-inclusive relaxation. It is highly recommended that you maximize
your concentration by taking into account the following factors:
The concept of relaxation is often associated with alpha wave learning which has attracted lots of companies that are more interested in their bottom line than their customers actual success in learning. EEG measurements can be used to generally conclude on the current state of the brain in the same way as you could detect bustling activity in a major city by scanning the surrounding electromagnetic field. The usefulness of alpha wave scanning in learning can be compared to the usefulness of electromagnetic field scanning for social life of a city. You need to focus on the causes rather than on symptoms. Alpha waves appear primarily in the absence of visual processing and other intense mental processes. This is why they cannot dogmatically be considered as a desired learning state. After all, the drowsy alpha state that precedes falling asleep is exactly the worst moment for learning during your day.
In evaluating the "relaxation products" you need to differentiate between the relaxation effect and the actual learning effect. The number of companies producing false claims in this field is astounding. It is very easy to fall for a simple solution to a learning problem (e.g. get 10Hz binaural beat difference and your learning problem will go away for life, and perhaps your sex drive will improve at the same time, you will sleep better and you will look younger). The easy learning solution explains why false claims related to "learning in relaxation" are so hard to extinguish.
At the same time, if you need to cope with stress or insomnia, many products in the field may have a legitimate application. Customers of the Polish Sita system jokingly claim that the company would do better if they marketed their product as a napping system. A worthy application on its own. If you know of relaxation products with legitimate claims and proven results, please let us know. We will gladly write about the subject or provide links from this website
Question:
What value of the forgetting
index ensures the optimum ratio of (retention)/(time
spent per day)?
Answer:
Paradoxically, the highest speed of learning can be accomplished ... without SuperMemo! In our daily life we pick up lots of facts that stay in our memory for long with few repetitions in lifetime! The problem is that these are usually not exactly the facts or rules that are critical to our goals. In other words, not the speed of acquiring new items counts but the speed of acquiring new items bearing a given content.
It is difficult to determine exactly what forgetting index brings the highest acquisition rate. Simulation experiments have consistently pointed to the value of 25-30%. You can even plot speed-vs.-forgetting graph using your own actual learning material in SuperMemo 98 or later using Tools : Statistics : Simulation. You will probably also arrive to similar results
As you perhaps know, SuperMemo disallows of the forgetting index above 20%. This comes from the fact that you should aim at achieving high speed of learning combined with high retention of the learned material. Setting the forgetting index above 20% would be like giving up SuperMemo altogether and coming back to remembering only that what is easy to remember. In highly interlinked material where new knowledge depends on the previously acquired knowledge, high forgetting rate can even be more harmful
Nevertheless, if you want to maximize the speed of learning with little control over what actually stays in your memory, set the forgetting index to 20%
Reading also involves active recall
Question:
I think your website overestimates the
importance of active recall, esp. in learning languages. If I read 20 books
(purely passive review), my result will be far from zero. On the contrary, my English skills will shoot up
Answer:
Active recall is needed to guarantee the high retention as defined by the
forgetting index (even 99%). Depending on
volume, structure, delay, etc. passive review may leave as little as
less than a percent of recall. However, reading books for the sake of
learning English is not just passive review. Each time you encounter a
problematic word, the need for comprehension will automatically trigger
an active recognition test in which the stimulus is the word in
question and the response is its semantic association.
This is active recall
Putting things in the same place is a good idea
Question:
Some
memory tip websites recommend that we always put things in the same
place. This way we never forget where they are. I thought that we
should do just the opposite. Should we not train our memory as opposed
to making things easy?
Answer:
Life is a process of
perpetual learning. If you put things always in the same place, you
will economize more time for learning more useful things. You will
never run out of the material for training your memory. It seems far
more useful to memorize with SuperMemo
the geography of South-East Asia than to spend time on a key
hunt
Question:
What techniques can you use to remember someone's name after you meet them? For
example, some sort of word-association; if the person's name is Smith,
associating a blacksmith with them might help. If you could give me two or three
tips for our readers that would be really helpful
Associating Smith with blacksmith and providing a mental image of
Smith hammering his colleagues at work is likely to leave a strong memory
imprint (the more gruesome or shocking the association, the better it works).
The solutions range from low-tech and less efficient to high-tech and
"memory-leak-proof":
Mnemonic techniques: The Smith-blacksmith approach falls into the category of mnemonic techniques, which use the power of the visual cortex of the brain to represent memories that humans are poor at remembering (e.g. complex numbers). We easily remember images and can often recognize faces after very long intervals without repetition in the meantime. However, an intelligent person will often find it difficult to recall seven consecutive digits presented one after another. Humans are inherently poor at numbers! With mnemonic training, the same person may quickly arrive at remembering sequences of 50 digits, cards, or objects. The Guinness Record drives this to unimaginable heights (thousands of digits or playing cards) -- far surpassing what we need in our daily lives. One of the most useful techniques, for remembering numbers (e.g. phone numbers) is the so-called peg list. If you memorize 100 so-called pegs that associate numbers with pictures (e.g. one is tree, two is a switch, three is a tripod, etc.), you will quickly learn to "see" numbers as "scenes". For example, 911 might be a cat ("cat has nine lives") stretched between two trees (assuming tree is one). Memorizing 100 pegs is a substantial effort and not many will want to undertake it. However, remembering a 10-digit number can then be reduced to remembering a 5-event sequence (e.g. my grandmother is chasing the postman who hides in a manhole and is eaten by a giant rat). Such memories tend to be manifold easier to retain. If you need to remember more than 30 numbers, the investment in memorizing the peg list will pay back. Beyond 50 (e.g. your personal phone book), peg-list is a pure saving.
Spaced repetition: The above mnemonic approach converts "difficult memories" to "easy memories" and is a subject of many books and self-improvement tapes worldwide. However, there is a substantial flaw. "Easier memories" are also subject to forgetting! They tend to be lost at a slower pace but they do get lost in the end. Forgetting also affects the peg-list if it is not rehearsed often enough. In other words, we cannot remember without a review/repetition. Spaced repetition can rationalize and simplify the review of study material, and ensure nearly perfect recall. This may not be as easy to present to your readers as the Smith-blacksmith association, however. The major drawbacks of spaced repetition are:
Herbert Simon guess on the expert memory power reflects his genius
Question:
Nobel
Prize winner Herbert Simon claims that it takes 100,000 to 2 million
"memory patterns" to make an expert. How does it square with your
estimations based on SuperMemo?
Answer:
Simon's estimate
is surprisingly accurate considering the fact that he did not use any
mathematical formulas for estimating the forgetting rate. He started
off with a very rough estimation of time needed to establish a single
memory pattern and used his "10 years of hard work" prescription for an
expert. However, the number 100,000 is probably closer to the real
memory power of an expert. Anything beyond that will hit the wall of
knowledge management that can only be overcome with tools like
SuperMemo
(Mariusz
Zmuda, Poland, March 22, 2000)
Is it true that it is better to get shorter sleep in the night and then
take a few naps during the day?
Poor memory is usually a self-imposed limitation
Question:
I was tested some years back for my memory, I always had problems learning at
school, they told me I had a genetic short term memory loss, possibly inherited
from my father, my children show some of the same signs that I have
Answer:
Please consult another specialist! Genetic factors affecting short-term
memory are quite unlikely. In addition, it is the long-term memory that makes you succeed
in life, while short-term memory can be honed easily with mnemonic training.
Hopefully, SuperMemo should help you figure out that your abilities are not much
different from others, and that your future rests in your hands (and your brain).
Dozens of people claim they have poor memory only to find out, in measurement,
that they do not differ much from others (except perhaps for some trainable
mnemonic skills). Among those above forty, the claims of poor memory are
epidemic
Spacing effect results from reduced potential for increasing synaptic strength
Question:
Why is the term
Spacing Effect used with negative connotations at supermemo.com? Spacing Effect
refers to the beneficial effects of spaced repetition (which I prefer
to call spaced recall since active recall, not mere repetition or
review, is key)
Answer:
Forgetting and the spacing effect
are beneficial from the evolutionary point of view as these both
evolved to prevent memory overload. However, in the context of
learning, both phenomena need to be struggled
against by the student. Spacing effect may be interpreted as "shorter
intervals
- harder learning" or as "longer intervals - easier learning". In other
words this might be a glass half-empty case. However, once we
understand the molecular origins of the spacing effect, we
quickly come to see it as a mechanism directed against remembering. In
the same way, once we try to fill a glass with water, we will call it
half-full in the middle of the way. Spaced repetition is effective
exactly because it goes around the spacing effect, i.e. around the
reduced ability to reinforce memories at high levels of retrievability
Automaticity does not correlate with the probability of forgetting
Question:
How highly correlated is low latency in answering a question (high automaticity) to the probability of forgetting?
Answer:
There is no correlation between automaticity and the probability of forgetting in the period that follows the measurement.
In the terminology of the two variables of long-term
memory, automaticity is correlated with retrievability, while probability of forgetting in a given period (assuming the same known retrievability) is correlated with
memory stability (also confusingly called memory strength).
In simple terms, if you remember something very well, you do not know
how fast you are likely to forget it. It depends on the history of
previous exposure. For example, in SuperMemo, the automaticity of
recall is similar after the fifth and after the tenth repetition.
However, memories built in ten repetitions are likely to last far
longer. Perhaps even a lifetime.
Mathematically speaking, retrievability R(t) corresponds with the probability of retrieval of a memory trace at the time
t. Probability of forgetting in the time d will then be R(t)-R(t+d). However,
R is the function of memory stability S: R=e(-t/S). Hence we have the probability of forgetting as
P(t,d)=e(-t/S)-e(-(t+d)/S). Predictably,
the stronger the stability, the less forgetting you will see. The
starting point retrievability is not included in the equation, because
it is the steepness of the forgetting curve, not the starting point,
that determines the speed of forgetting.
It is worth noting, that with biased statistical sampling, it might be possible to erroneously prove the false hypothesis of the negative correlation between automaticity and probability of forgetting. This comes from the fact that very low retrievability will always produce very low probability of forgetting nearly independent of memory stability (at the bottom of the forgetting curve). No wonder then, that in "traditional learning" (i.e. not based on spaced repetition), where both retention and stability are much lower, it is easy to falsely assume that if you know something well, you will remember it far longer than if you show recall problems
Mixing declarative and procedural learning may be good for you
Question:
Has
an
interference effect been noted between declarative and procedural
learning? If I immediately follow a Spanish lesson with a martial arts
lesson, will consolidation of either be adversely effected?
Answer:
Declarative
and procedural learning are governed by different brain circuits and
even their consolidation in sleep seems to be separated in time. For
that reason, interference between both is highly unlikely. At the same
time, there could be positive trophic feedback between the two (i.e.
learning in one circuit may have some positive effect on another). In
other words, mixing Spanish and martial arts might be a very good
strategy for enhanced learning. As for the optimum sequence, it is very
difficult to determine. If your martial arts training is very intense,
it might better follow Spanish due to possible physical exhaustion. On
the other hand, if it is modestly stimulating, it could equally well be
executed before. You could also split your Spanish into two portions
and execute part of it before and part of it after the training. That
could prevent the overload effect (mental tiring in prolonged
learning). Your best strategy for determining the learning sequence
would probably be to try what seems to work best for you
Ages old techniques are still valid
Question:
Do you recommend techniques such as positive attitude, visualization, concentration, etc.
Answer:
Yes.
Positive attitude works in all walks of life. Visualization makes
things easier to remember. Concentration is vital for encoding memories
(the sharper the laser light, the better the etch). Those things are
all listed in
SuperMemo Decalog
Your own personalized peg-list would work best
Question:
I was trying to search for a pre-done peg words set for memorizing
numbers. I could not find it in SuperMemo Library
Answer:
If
you want to memorize numbers, e.g. phone numbers, a peg-list from 0 to
100 is priceless. However, nothing works better than your own peg-list.
For example, if you were born in 1976, you could associate 76 with
yourself. That's far easier than associating the peg with Jimmy Carter
who was elected in 1976. Choose your own unique pictorials for all pegs
and then learn how to build funny scenes of two-digit combinations
I have heard that baroque music is the best for improving memory; particularly Bach and
Vivaldi. Does it also work in association with SuperMemo?
Answer:
The claim that any particular kind of music is best for memory is unlikely to be true.
Indeed, music can have powerful impact on our emotions and, consequently, on remembering.
It has been found that it is similar in effects to caffeine. However, a study that measures
impact of a particular kind of music on recall in a group of people can be compared to
trying to find the optimum size of a shoe for an average citizen. Depending on the musical
education of an individual, the same kind of music may bring a variety of emotion from
relaxation, through agitation to aggression. While Four Seasons may have a positive impact
on the mind of majority of the population, the best bet would be that everyone should
stick with the music he or she likes. Be it punk or funk. The truth is that all that is
good for the mind and health is good for remembering.
Why do synapses get weaker during sleep?
Question:
NYT has reported on a research
(http://www.nytimes.com/2008/01/29/science/29obslee.html)
that sleep weakens the synaptic strength. At the same time, synaptic
strength increases during waking. This seems to go against your own
findings that you published on your website that indicate that during
the day one's ability to recall facts seems to be waning
(http://www.supermemo.com/articles/sleep-research-2007.htm)
Answer:
There is no contradiction between the fact that synapses get strengthened during waking and the ability to recall things drops at the same time.
First we need to differentiate between:
Secondly, we need to look at the most likely explanation for the weakening recall during waking. The most coherent, attractive and best-supported hypothesis says that the overload of short-term low-interference networks is responsible for a declining capacity of memory during a waking day. This decline cripples the working memory, and in consequence, it affects the entire spectrum of human cognitive capabilities. The main function of sleep would then be to redistribute, reconsolidate and optimize those short-term memories that slow down further learning.
To put it metaphorically, the brain is like a computer that keeps loading chunks of data to its memory during the day. As the memory fills up, the computer slows down, and all applications crawl into a halt. However, if you test individual memory cells, you will notice that they strongly cling to their new data. In the night, the computer will gradually organize the chunks of data, remove discrepancies and duplicates, write down memories to the hard disk, and run a defragmentation process for easy and fast access.
Both the increase in synaptic conductivity in wakefulness, and the decline of learning capacity during the day are well documented.
As for the decline in synaptic strengths during sleep, it also fits well with the present models of sleep and learning. One of the main functions of sleep should be to optimize the memory storage. This entails representing memories in most efficient way, i.e. so that they are most abstract, consume least space, generate minimum interference, and so on. That process should indeed result in reducing the overall cost of memories, and result in weakening of redundant synaptic connections.
There is no point when you can say that a given piece of information is remembered for ever
Question:
From your experience, and I know it is different with different people, how many
repetitions (how far out on the scale in the future) does one need to make to have
mastered in information, and what is the standard deviation for that number?
There is no point when you can say the material is remembered for
ever! Even after 20 years of repetitions there is a slight risk of forgetting.
If you want to always remember a piece of information with a defined degree of
probability, you should continue repetitions indefinitely.
We do not take an official stand on memory enhancing drugs and advise caution and consulting your physician
Question:
What do you think of so called memory-enhancing smartdrugs: Piracetamum, GABA,
Vinpocetin, Takrin, Diapid, Vasopressin, Fipexid, Deprenyl, Hydergin, etc.?
We do not take an official stand on memory enhancing drugs. Please consult
other sources. Note the following though:
Memory "overflow" is not "fatal"
Question:
One some website related to SuperMemo I found that without forgetting
"the brain would fill up ... with ... fatal results". This sounds quite silly. Does research support this?
Answer:
No.
This statement is unfortunate. We are unable to control all
what is being written about SuperMemo.
The corrected statement might be: Forgetting is needed to clean up room
for new memories in the limited space of the human
brain. For example, neocortical representations cannot be reconstructed
by the
hippocampus due to the limited size of its network; hence the constant
flow of
memories. The reports of patient confusion in cases of reduced
forgetting are sporadic and difficult to verify. Genetically engineered
mice that exhibit reduced forgetting rate have actually been shown to
outsmart normal mice in tasks such as navigating a maze
What are the youngest users of SuperMemo?
(anonymous, Germany, Jan 21, 2013)
Question:
What is your position on teaching a toddler with SuperMemo? What is the earliest age of kids who start SuperMemo successfully? Will the algorithm work in the young brain as well as in an adult? Could there be some side effects?
Answer:
Theoretically, there is no age limit for SuperMemo, however, the skills needed to efficiently use the program develop slowly over the first 10-15 years of life. Most of the kids, below the age of 10 can only use SuperMemo with parental supervision and help, unless they use a simple ready-made collection, e.g. for learning a foreign language vocabulary.
The algorithm for spacing repetitions will be less efficient in the first years of life due to a rapid growth and the rewiring of the hippocampus and other structures involved in memory. This neural growth is almost certainly responsible for
childhood amnesia and may last a few years. In most cases, kids cannot recall episodic memories from below the age of 3. However, this does not invalidate the repetition spacing algorithm. It only makes it pretty inefficient. Spaced repetition will work up to a point when memory becomes inaccessible due to the interference caused by network growth or new knowledge that streams into the young brain at the speed of light. Our knowledge of spaced repetition in kids is limited due to the lack of research and even lack of anecdotal evidence. Up to this point (Jan 2013), we have not received any reports of the use of SuperMemo in kids at an extremely young age.
The main side effect that you might experience is the possibility of discouraging the kid. With any degree of coercion or monotony, the kid may grow to dislike SuperMemo or learning in general. The litmus test, as always, should be smiles, enthusiasm and the fun of learning. Without it, you would really do better without SuperMemo by just following kid's natural learning instincts. If the kid prefers to stack bricks instead of learning Japanese words, bricks should get the priority. The young brain is programmed to seek novelty. SuperMemo, by definition, is a destroyer of novelty and goes against a toddler's natural instinct. However, it is conceivable that you might instill some passion by interleaving items with some YouTube songs, home videos, pictures, etc. Most of all, you need to show the enthusiasm yourself, praise the tiniest sign of progress, and be restrained in criticism and complaints. You can program the proportion of topics to items, you can manually choose priorities and intervals, etc. In essence, your assessments might act as a better substitute for SuperMemo algorithms with the program playing essentially the role of the knowledge base and knowledge manager.
Remember to choose the right material for learning. Your choice of items should be pragmatic. Memorizing capitals of countries is totally abstract and not too useful to a very young child. Body parts, foods, and the names of daily activities are far more pertinent. The young brain is exposed to new knowledge all the time. This means that it keeps learning even if you do not try too hard. SuperMemo could assist you in breaking through some bottlenecks that make daily life difficult, however, natural learning is based on extracting patterns from the environment and is definitely more valuable than the ready-made servings from SuperMemo. If your kids like SuperMemo and you end up using it, please remember to report back so that this theoretical FAQ could be updated with real-life experience.
What is faster: learning or forgetting?
(Martin Mich, USA, Feb 3, 2014)
Question:
What is faster: learning or forgetting?
Answer:
Interesting question! In 30 years of SuperMemo, nobody seems to have ever asked!
Comparing the learning and forgetting rates is like comparing apples with oranges. Both processes vary in speed and both are faster at the beginning of the learning experience. The are strongly connected: the more you learn, the more you forget.
The speed of learning depends on many factors: the individual, difficulty of the material, the way the material is formulated, and the learning method used. If we neglect individual differences and compare the speed of learning and forgetting of the same material of the same difficulty, and employ optimum learning based on spaced repetition then the speed of learning will mostly depend on the forgetting index. Paradoxically, for higher values of the forgetting index (e.g. 30%), we forget more, and yet learn faster. This is because we save a lot of time on less frequent review.
For the same individual and the same material, the speed of forgetting will depend on the stage of learning. Forgetting is negatively exponential, and is fastest directly after learning. For a regular inflow of knowledge, new knowledge is forgotten faster, however, with passing time, we learn less and, as a result, we forget less.
Even though the speeds of learning and forgetting seem hardly comparable, you can get a big clue from an important observation: your knowledge keeps increasing throughout your life, and may peak at retirement age when aging, cognitive decline, or disease might reverse the balance.
The ultimate answer would therefore be: for a healthy individual, learning is faster than forgetting!
Comparing learning with forgetting is like looking at the river. It gains some water from precipitation and from its tributaries, but it loses roughly the same amount by the flow of the water to the sea. However, in the course of the life, we move closer to the delta, and keep carrying larger and larger volumes. More importantly, that entire mass of knowledge becomes more coherent and better structured. Hence older people are known for their wisdom (usually). The same knowledge that makes them wise, makes them conservative too. This is why old wise people are less adaptable and rarely make great inventors.
In conclusion: Learning wins most of the time! But he battle is hard
Who invented spaced repetition?
(Martin Mich, USA, Feb 03, 2014, 22:37:58)
Question:
Who invented spaced repetition? Not Ebbinghaus?
Answer:
Nobody should ever take credit for discovering the spaced repetition.
This is because of the fact this concept is pretty intuitive, and will be
re-discovered in the mind of every attentive student sooner or later. In fact,
we asked teenagers a set of questions about how their memory works. A large
proportion can come with pretty good guesses without ever making any
measurements. In particular they often correctly guess that the first optimum
inter-repetition interval might be a week long, and that successive intervals
will increase. Moreover, many could guess that the second interval might be a
month long, and that successive intervals might double. In other words,
spaced repetition is a common intuition.
Publication
Many authors over the centuries published their own guesses and intuitions about optimum intervals in learning. In particular, practitioners Tony Buzan and Paul Pimsler managed to popularize their own variants of spaced repetition. An oft mentioned Sebastian Leitner system was rather a flashcard prioritization procedure that did not produce a specific interval scheme, but could still be used to improve retention in learning. Those early recommendations were relatively popular in privy circles of memory artists and language students. However, they still remained largely unknown to the general population for years to come.
Computers
SuperMemo can take credit for pioneering computational spaced repetition. First experiments that made it possible to compute optimum intervals were made by a student of biology, Piotr Wozniak in 1984-1985. Those were later computed more precisely with his software that evolved into SuperMemo (1987). At the same time, we heard of similar efforts by other researchers and programmers using fuzzy logic or set theory, but have no confirmation if these attempts were successful. Since then, we have only heard from a user of SuperMemo, Dawid Calinski from Poland, that his Full Recall program, based on neural networks, was also able to compute a good approximation of optimum intervals (2003). Those results have probably not been published yet. In the 1990s we heard of many attempts to investigate or implement spaced repetition that were abandoned upon the encounter with SuperMemo. In the new millennium, we see a transition from experimentation to implementation. We hear of many programs implementing SuperMemo algorithms, while very little of other experiments using various computational methods to replicate SuperMemo results.
To illustrate the omnipresence of spaced repetition see if this "Harvard patent story" does not resemble "SuperMemo on paper, 20 years later": http://harvardmagazine.com/2009/11/spaced-education-boosts-learning
Name
In the early days of SuperMemo World, we actively searched for a well-recognized scientific term to use when referring to "the SuperMemo method" in scientific contexts. The company's marketing strategy was to move away from a "program developed by a student" to a "program based on a scientific method". Unfortunately, memory and learning literature was scant on publications other than short-term studies of the spacing effect (with a notable exception of H Bahrick's research on the retention of Spanish vocabulary). In the end, we moved from our own term "repetition scheduling" to a rarely used term "repetition spacing" (around 1992) that was later replaced with "spaced repetition" (around 1999). Over time, the term "spaced repetition" became more and more popular. In that light, SuperMemo can also take credit for making this generic scientific term take root in public's mind with a specific association to optimum intervals in learning.
Definition
For the sake of computing optimum intervals in learning, we need a definition of the optimum interval. This definition was provided by SuperMemo only in 1991 with the introduction of the concept of the forgetting index. No other definition has been proposed and accepted until now even though different criteria might also be used for defining optimum intervals (e.g. with a focus on speed of learning rather than the level of retention).
Ebbinghaus
A popular myth about spaced repetition was initiated by our file on the history of SuperMemo. Our statements on the contribution of Hermann Ebbinghaus to the experimental study of memory was misunderstood as "Ebbinghaus invented spaced repetition". Ironically, the distorted version found its place even in official marketing materials distributed by SuperMemo World. In actuality, like many scholars before him, Ebbinghaus expressed his intuitions about spaced repetition in his ground-breaking work on memory: Memory: A Contribution to Experimental Psychology (1885). He stated that: "with any considerable number of repetitions a suitable distribution of them over a space of time is decidedly more advantageous than the massing of them at a single time" (source).
Summary
Spaced repetition is a common intuition. It has been written about for centuries by various authors. The credit for computing optimum learning intervals goes to Dr Wozniak. The credit for the computational definition of the optimization procedures goes to SuperMemo. So does the ultimate choice of the term "spaced repetition" to attach to the process. In the end, the credit for the rising popularity of spaced repetition goes to countless users, students, and program authors that continue learning, implementing new applications, and spreading the word about this amazing technique and associated technologies.
Advanced students should become blind to their environment
(SuperMemoUser, Aug 1, 2014, Fri, 23:38)
Question:
In an interview with Dr Bjork I read:
"Studying in only one location is great as long as you’ll only be required to recall the information in the same location. If you want information to be accessible outside your dorm room, or office, or nook on the second floor of the library, Bjork recommends varying your study location".
Answer:
It is true that when learning an association between two concepts, the brain often picks up irrelevant information from the environment: both external (e.g. location) and internal (e.g. emotional state). This is why the same information may be retrieved in certain contexts, and unavailable in other situations. For this reasons, students may be
advised to look for a change in the context in which the same information is learned to make sure neural networks of the brain filter out the variable component and retain only the key association that is to be formed.
However, this advice is not universal. There are differences between declarative and procedural learning. Moreover, there are different levels of advancement in the studying art that will affect the optimum strategy.
In procedural learning, the change of context may be particularly important. A tennis player who sticks with a single court will be a master of that court but his neural networks will never produce generalizations that can help the same player excel on clay, grass, in Paris, and London.
On the other hand, the context change advice is not applicable to incremental learning at the advanced level. SuperMemo students should be able to isolate the learned information and become context-blind. This refers to the learning environment, not the context of information (e.g. other memory links leading to and from the learned knowledge). Paradoxically, the lesser the change in the environment and the mental status, the easier it is to achieve the state of "blindness", i.e. arriving at the full focus on the learned information only.
This is why, in the context of incremental learning, for advanced students, the advice is the opposite: always learn in the same "boring" place, on the same "boring" computer, in the same peak mental state (i.e. "boring" unemotional peak of alertness).
In the end, for an expert in incremental learning, the environment will matter
less and less due to the ability to immerse oneself in the studied subject.
A creative wandering state of mind is welcome in the process of reading. However, while making item repetitions, the whole world of students
consciousness must be focused on only two entities: the stimulus (question), and the
response (answer). The brain will do the rest by creating or reinforcing the association between the two. The less noise arrives at this binary world, the better the association. Even the relevant information, e.g. illustrations in an article, must be chosen carefully to make sure it does not provide additional cues that might lead to a false sense of remembering. All hint-carrying references, examples, illustrations, etc. must be removed from items.
For an incremental reader, the ultimate test on his or her learning progress is the applicability of the acquired knowledge in real life. The context of the learned association should ideally be filtered out at learning, review and retrieval states. All that should ever matter is how the formed association contributes to the complete fabric of knowledge of the individual.
Without review, all memories are vulnerable to permanent forgetting
(SuperMemoUser, Aug 12, 2014, Tue, 19:56)
Question:
Dr Robert Bjork says in this
interview:
Once you learn something, you never actually forget it. Do you remember your
childhood best friend’s phone number? No? Well, Dr. Bjork showed that if you
were reminded,you would retain it much more quickly and strongly than if you
were asked to memorize a fresh seven-digit number. So this old phone number is
not forgotten—it lives somewhere in you—only, recall can be a bit tricky.
This is contradicted by your claims that everything can be forgotten and that re-learning isn't much easier than learning from scratch.
Answer:
Even though it really sounds like a different interpretation of memory
science, the actual difference in views isn't that significant. Let us consider
two separate subjects: the ease of re-learning and the possibility of never
forgetting things.
Re-learning
We have known for ages that learning the same subject again is easier than learning it from scratch. However, in SuperMemo, we work using minimum information principle where each item forms a maximally simplified link between two concepts (e.g. the same word in two different languages). The more complex the memory, the more complex the process of forgetting. In complex items, you may forget a few memory traces, and leave others still remembered. Re-learning such a complex item will often be easier. Part of the learning job will have already been done due to the traces that have not been forgotten. At the macro level, it is as if learning calculus again after a 5 year break. Re-learning makes a world of difference. It is often just like brushing up the old memories. In SuperMemo, on the other hand, forgetting is much closer to all-or-none in nature. The memory status is more binary. Either you remember, or not; with fewer shades of gray. This is why, once you lose access to your item's memory, you will need to build the whole memory anew from scratch. This means the benefit of re-learning is negligible. You can best see it in your own collection in Tools : Statistics : Analysis : Graphs : First interval. After a few years of learning you will observe that the first interval for forgotten items is shorter than the one for newly memorized items (see example). In short, re-learning of whole subjects is easy, re-learning of well-structured items is as hard as learning the same items for the first time. In SuperMemo, re-learning appears even harder, because only difficult items get forgotten multiple times and their first interval post re-learning must be particularly short.
Deleting memories
Do memories just get misplaced or do they get permanently deleted? To our best knowledge (as of 2013), we do not have a direct answer to what happens to memories once we permanently lose access to them. It is simply not easy or possible to locate neurons or synapses that once formed a part of a meaningful memory and have later been orphaned in the process of forgetting.
To understand the process of forgetting, it is helpful to distinguish between memory retrievability (i.e. how easy the memory is to be recalled) and memory stability (i.e. how long it can last)(see: Two components of memory for a detailed explanation). In Bjork's terminology, retrieval strength and storage strength stand for retrievability and stability respectively. In the case of an old friend's phone number, we would have built up solid memory stability over a couple of years of use of that memory. However, 20-40 years later, retrievability may drop enough for us to be unable to actively recall the number. We might still be able to retrieve that number with a help of a hint, multiple choice question, or in passive recognition. All those props help fill out the missing pieces of the jigsaw puzzle where the retrievability was not high enough. Bjork's example thus shows that memories can last for years, however, it says nothing about what happens to stability once the memory is permanently inaccessible. This means a situation where no amount of help is sufficient to reactivate the same memory circuits. In SuperMemo, given a sufficiently long learning process, we can detect hundreds of cases in which relatively stable memories disappear after a sufficiently long interval. Those cases all say the same story: some memories are gone for good and re-learning is hard as if learning for the first time.
We have come up with the most accurate formula for changes in memory stability in this publication Building memory stability (formula 7.1). You will notice however that the formula can only describe the situation in which the repetition was successful. In case of a memory lapse, we are not able to measure memory stability. Re-learning in SuperMemo occurs all the time, however, as explained earlier, it is highly binary and items do not capitalize on past stability. In other words, there is little difference between learning and re-learning (as seen in the first interval graph above). This is why, after forgetting, in SuperMemo, items build their stability from scratch.
Molecular memory stability is likely associated with synaptic proteins. Do we have any evidence that these proteins get gradually degraded, re-located (e.g. internalized), or modified (e.g. de-phosphorylated) after the synapse stops playing a part in memory recall? We can probably rely on indirect evidence only:
The ultimate fate of memory stability in unused synapses is not clear. If Dr Bjork was right, we might grow conservative brains that gradually lose their learning capacity in addition to the natural process of aging. Degree of that loss is hard to estimate as the information capacity of the human brain is indeed enormous. On the other hand, the more likely scenario is the one in which unused memories get cleared up over time. This scenario is biologically and computationally more sound and more optimistic. One way or another, this should have little bearing on our learning strategies.