Thursday, July 30, 2015

Put It In the Vault: Permastore

If you took a foreign language in high school or college, do you remember the translation of the following words? 
  • Three
  • Mother
  • Water
  • Street
  • House
How many of these words did you remember? Also, how long ago was it that you studied a foreign language? Was it between 0 - 6 years ago, 6 - 25 years ago, or more than 25 years ago? The reason I ask is because it's possible that forgetting occurs at different rates, depending on how long ago the initial learning occurred.


Fifty years of data collection

It's every educator's dream that her students will retain the information from her class for the rest of their lives. The reality is that some amount of forgetting will occur after the class ends. But how much information is retained over an entire lifetime? Has anyone ever attempted to collect data that would reveal how much people forget over the course of their lives?

At least one brave soul has tried. A cognitive scientist, who we shall refer to as Harry, decided to measure the amount of knowledge retained over a lifespan for students who took Spanish in high school and/or college [1]. Before tackling the Herculean task of collecting his data, Harry had to make a methodological decision. One method to track memory over a lifetime would be to measure the knowledge of a large sample of high school or college students, and then test their knowledge every year for fifty years. This is called a longitudinal study. The advantage of a longitudinal design is that Harry would have fairly good control over extraneous factors that are not relevant to his experiment. The obvious disadvantage is that it would take precisely 50 years to collect all of his data. To overcome this disadvantage, Harry decided to conduct a cross-sectional study instead. That is, rather than following the same individuals over time, he recruited a large number of people who studied Spanish between 1 and 50 years ago, and compared current memory of the language for people who learned it at different time points.

What were the results from Harry's study? Do students forget everything after leaving class? If not, how much do they retain 50 years later? Before we dig into the results, let's set the stage by looking at the historical context in which Harry conducted his study.

Whee! Sliding down the forgetting curve

One of the earliest controlled experiments on human memory was conducted by a German psychologist named Hermann Ebbinghaus [2]. To accurately measure the effects of learning and forgetting, he dedicated himself to studying lists of "words" (although they weren't really words). He constructed trigrams, which were composed of a consonant, followed by a vowel, and concluded with a consonant. LEK is an example of a trigram. 

The reason Hermann made lists of trigrams instead of real words is that he was attempting to control for the pre-existing associations that words have in long-term memory. Also, testing memory of made-up words would prevent Harry from using any of the memorization tricks we have talked about in previous posts. Thus, he was attempting to measure rates of learning and forgetting in a "pure" environment. To accurately measure the time course of forgetting, Hermann attempted to recall his lists of words across several time spans. The shortest time span was 20 minutes, and the longest span was one month (i.e., 744 hours).

How did he do? Ebbinghaus's results are plotted in Fig. 1 [3]. You will notice that even after 20 minutes, there was a little bit of forgetting. But after a month, he remembered only about a fifth (21.1%) of the original list of trigrams.


Figure 1: The amount of forgetting over several intervals of time.

How does this stack up to the forgetting curves from the Spanish study? If you squint, they look surprisingly similar! In Fig. 2, you can see that there was an extremely steep drop-off in terms of the correctly recalled vocabulary words between the ending of the class and six years afterwards. But after six years, Harry's participants remembered around 70% of the items that he tested. These items lasted for a very long time. If these vocabulary items made it past six years, then they were highly likely to be recalled 25 years later. The discontinuous nature of forgetting in this graph suggests that there might be a very long-term memory, which Harry called permastore. If memories make it to permastore, they will likely remain with us until the end of our lives.


Figure 2: Amount of forgetting of vocabulary terms over 50 years.

The STEM Connection

Plants grow at the apical meristem and Trypanosoma causes African sleeping sickness are two declarative chunks of information that I learned in 10th grade Biology. That was over 20 years ago. At my current rate of forgetting, I will probably take these facts to my grave. (My Biology teacher would be so proud!) Unfortunately, there are hundreds of other facts from that same class that I will never be able to recall. (My Biology teacher would probably be less than thrilled to hear that.) So should we think of this as good news or bad news? 

I think the concept of permastore should be perceived as good news, and here is why: we are partially in control of how much information makes it into permastore. The line in Fig. 2 glosses over an important variable in the original Spanish study. Harry categorized the individuals in his study into groups who learned the material really, really well and into groups of student who did not learn the material as well. The line shows the forgetting rate averaged over multiple levels of initial learning. It turns out that the lines were different based on how well people learned the material originally. Specifically, those who mastered their Spanish vocabulary demonstrated less forgetting than the students who did not learn their vocabulary words to the same depth (i.e., the lines for each group were completely parallel). Thus, if we study hard, and learn the material really well, then it is likely that more items will enter permastore. That, I think, is very good news indeed.


Share and Enjoy!

Dr. Bob

For More Information

[1] Bahrick, H. P. (1984). Semantic memory content in permastore: fifty years of memory for Spanish learned in school. Journal of Experimental Psychology: General, 113(1), 1.

[2] Ebbinghaus, Hermann (1964/1885) Memory: A contribution to experimental psychologyOxford, England: Dover.

[3] I'm doing some handwaving here. Ebbinghaus's methodology was to memorize a list so that he could reproduce it perfectly. Then he would wait some interval of time. Then he would try to remember it. If he failed, he would study the list again until he could reproduce it without any errors. The amount that he didn't have to study, he called "Savings." For example, when the time interval was short, he didn't have to study as much to get back to 100%. For seriously long intervals (i.e., a month), he had to study a lot to get back to 100%. For the sake of the current discussion, though, we can just talk about this in terms of how much he forgot. 

No comments:

Post a Comment