Langston, Cognitive Psychology, Notes 10 -- Interesting Cognitive Stuff

I.  Goals.
A.  Where we are/themes.
B.  Counting-out rhymes.
C.  Distribution of memories.
II.  Where we are/themes.
A.  Where we are.  We'll look at a diverse mix of research that doesn't fit in the sequence so far.  The criterion for getting in here is that it is interesting and doesn't really go anywhere else.  There isn't really a theme.
As an overview, this will wrap up the "representation and process" part of the class and get us to the "higher cognition" portion of the class.  We will cover these as we can, what we don't get to is out of the class.
III.  Counting-out rhymes.

Demonstration:  Complete this poem:  “Eenie, meenie, ...”  Try this one:  “One potato, two potato...”  Compute frequency for the two modal versions:
Eenie, meenie, miney, mo, One potato, two potato,
Catch a tiger by the toe. Three potato, four,
If he hollers, let him go. Five potato, six potato,
Eenie, meenie, miney, mo. Seven potato, more.
Most native English speakers who grew up in the US will produce the two versions above.  There are some other common variations, mostly related to what gets caught in eenie meenie.

A.  Why are we interested in counting-out rhymes?  There are several reasons.
1.  The rhymes were “designed” to be memorized.  They're used mostly by preliterate children.  These kids can't consult the big book of rhymes to decide who's “it.”  They have to go by oral tradition.  Further, since the rhymes have a ritualistic role to play in maintaining order, they have to be reproduced exactly.  Any variation could constitute cheating, and kids don't tolerate that well.
2.  The rhymes are readily accessible, and almost everyone has learned one or some of them.  This makes them relatively easy to study.
Given these facts about the rhymes, we can get to the truly interesting question:  What would something look like if it had to pass the real-world test of being memorizable?  This is completely different from the laboratory memory experiments we've been considering.  As we'll see, these rhymes have a lot of properties that the laboratory studies have failed to uncover.
B.  Properties of counting-out rhymes.  Is there an identifiable genre that these rhymes all belong to?  In other words, is there some system of rules that constrains these rhymes to all be pretty similar?  If there is, then we want to know what those constraints are.  By knowing the constraints, a person would never really have to “memorize” any of these rhymes.  Instead, the gist of the rhyme plus the rules could allow relatively sure reproduction.  In fact, there are rules.
Rubin (1995) has examined the most popular 24 rhymes used in English, and the least popular 24 rhymes.  Here are the properties:

Mean value Mean value
Property Common rhymes Rare rhymes
Lines per rhyme 4.56 4.04
Words per rhyme 21.42 16.96
Word repetition* .19 .14
Rhyme* .24 .21
Alliteration* .25 .21
Assonance* .25 .26
End rhyme* .79 .63
Overall poetics* .87 .82
Four-beat lines* .83 .60
Rated poetics 4.33 3.72
Poetic words on beats* .90 .79
Nonpoetic words on beats* .61 .68
Meaningful words*  .79 .77
Rated meaning 2.87 2.79
Rated imagery 2.96 3.16
*Represents a proportion.
Assonance is when the stressed vowel of a word repeats.  Alliteration is a repeated first letter.  Overall poetics is the proportion of all the words involved in some form of poetic device.

So, there is evidence of a common genre (or schema) that defines these rhymes.  First, the rare and common rhymes generally don't differ on these measures.  So, when kids were constructing the rare ones, they were made to fit the rules.  The main difference is in poetic words on beats.  The common ones have more poetic words on beats (which makes them easier to detect).
What are the properties of the genre?  Use nonsense words (similar to magic and ritual), use lots of poetic devices (almost 90% of the words are involved in something), have four-line rhymes with four beats per line (this emphasizes the counting and makes cheating harder), and put the poetic words on the beats.  It looks like repeated sound patterns govern the construction of these more than meaning.
What does this do?  It greatly constrains your choices of what to say at any point in the rhyme.  These constraints remove a lot of the burden from memory because the information you need is in the part of the rhyme you've already done.  In other words, if you know the gist, once you get started the rest comes pretty much due to constraints.  Here's an example of constraints.

Demonstration:  Write down a word that rhymes with “most.”  Write down a word that rhymes with red.  Write down a device for calculating numbers.  Write down a device that plays music.  Write down a spiritual or mythological figure.  Write down a spiritual or mythological figure that rhymes with “most.”  Everyone probably wrote “ghost” for the last one.

Ghost rhymes with most and is a spiritual or mythological figure, but until you put the constraints together, it's not very likely.  The sound constraints help a person singing a ballad or an epic poem.  If you finish a line that has the word “host” and you need a spiritual figure at the end of the next line, “ghost” is the only option, whether you remember it or not.  This is how sound patterns can contribute to memory for counting out rhymes.
Let's analyze one line:  “Eenie, meenie, miney, mo.”  First, the vowels follow a front to back progression (e, I, o).  This is also present if “fee, fi, fo” of “fee, fi, fo, fum” and Old MacDonald (ee-i-ee-i-o).  Front to back is a lot more natural than a mixture such as “eenie, miney, meenie, mo.”
There's also a repetition of “eenie” in “meenie.”  You have alliteration in “meenie, miney, mo.”  You have rhyming for “eenie, meenie, miney.”  “Mo” will rhyme with “toe” and “go.”  The /n/ sound is in the same place in three words.  The whole line repeats at the end and “mo” will determine who is “it.”  This puts the selection on a beat, which couldn't happen if “mo” was two syllables, like “monie.”  In other words, changing one sound anywhere will mess up some pattern that's present.  You can change the whole line (as in the Scottish “eena deena dina do”), but not one part.  The sound patterns determine all of the components that could be forgotten.
Meaning constrains the rest, but the meaning is a lot less important.  In fact, if you look at variation, most of it happens in the meaning.  How can we test this?
First, have a group of people write down “eenie, meenie” from memory.  Then, compare to the “standard” version.  What kinds of changes do you get?
Change Percent 
Replace story line 28%
Change tiger 31%
Miscellaneous 38%
Replacing the story line would be substituting something like “if he hollers make him pay fifty dollars every day.”  In the mid 1800’s when the rhyme first started, a n----r was grabbed by the toe.  Since then, this has become unacceptable.  So, something had to substitute.  What did Rubin get?  73 tigers, 10 monkeys, 10 rabbits, 2 fellows, 2 piggies, and one bunny, blackbird, buyer, chicken, doggie, froggie, and wiffer waffer.  The miscellaneous changes were mostly making “the” into “he,” “his,” or “its.”
The sound patterns didn't change.  In fact, the most important change (tiger) actually shows the sound patterns at work.  It had to be a two-syllable thing.  So, if you want a frog, it has to be a froggie.  The two most popular have other poetic constraints.  Tiger alliterates with toe.  Monkey alliterates with the items in line one.  So, the ones that add even more poetic constraint are the most popular.
In fact, looking at samples from people who learned the rhyme in 1953, 1965, 1976, and 1978, the n-word appears 40%, 6%, .5% and 0% of the time.  Tiger appears 56%, 70%, 84%, and 100%.  In other words, social forces outside the rhyme genre made change, and the system settled on the one with the best poetics (better than the original).  Rubin repeats this same analysis for one potato.
One last look:  What historical changes have taken place?  Up until around 1860, the most popular rhyme was:

Onery, twory, tickery, tevin
Alabone, crackabone, ten and eleven.
Pin, pam, musky dam.
Tweedleum, twaddleum, twenty-one.

Over time, it disappeared, and “eenie, meenie” and “one potato” became more popular.  Looking at a sample of former favorites and current favorites, what can we say?  Rhymes are becoming more meaningful.  This provides dual constraints of meaning and sound (the “eenie, meenie” analysis above was all sound).  In a competition, meaning and sound will make recall easier, and increase a rhyme's popularity.  Also, the kids using the rhymes are getting younger, which cuts down on the sophisticated poetics that they can take advantage of.
Upshot:  Rhymes are a genre which uses sound and meaning to constrain what is remembered.  Since you have to remember them accurately, this can tell us something about memory.
Some confirmation of this comes from a study by Rubin, Ciobanu, and Langston (1997).  We collected a sample of Romanian counting-out rhymes from Bucharest school children.  These rhymes were analyzed using the same rules as were developed for English.  The most popular Romanian rhymes:
Rhyme: English version: Rhyme: English version:
Din oceanul pacific From the Pacific ocean Pe o bara On a bar
A iesit un peste mic A little fish came out Se caca o cioara  A crow was pooping
Si pe coada lui scria And on his tail was written Ga, ga, ga Ga, ga, ga
Iesi afara dumneata Out goes you Drept in gura ta Right in your mouth
Last line could be:

Te iubesc, nu ma uita I love you don't forget me

When you look at the variability, you get the same pattern as for English.  There are some differences, but none of them affect the poetics.  When you compare the poetics, you get pretty much the same properties.  So, what we've concluded from the English rhymes (sound and meaning constraints lead to recall) appears to be universal. 
IV.  Distribution of memories.  How are your memories distributed across your lifespan?  When you're older (65-75 or above) what will you remember?  This question about autobiographical memory has produced some very interesting results.
A.  First, how do you ask the question?  The most common technique is to present a person with a list of words and ask that person to write down the first concrete memory that comes to mind when they hear the word.  The memory should be clear and distinct, and the person needs to have been involved in it.  For example, if you hear “store,” you might recall a time you went to the store with your father.  After all the words are presented, the person goes back and dates each memory (at least to the year, maybe more).  Then, you plot the distribution of memories by the age at the time the event happened.
See the results of such a study done by Jansari and Parkin (1996).  The top graph is for people 46-50 at the time of the test, the bottom graph is for people 56-60.  There are three parts to these curves.
1.  The recent part is explained by a power forgetting function.  You forget a lot right away, but less and less as time goes by.  This is similar to the function you get in laboratory research.
2.  The earliest part is explained by a function that has infantile amnesia.  Very few people remember anything before the age of 3, so this part goes to zero around that age.
3.  There's a bump in the middle (especially for older adults).  This bump represents a portion of the curve where you have a lot more memories than would be expected.  It usually goes from around age 10 to around age 30.  We're interested in the bump.
B.  Properties of the bump:
1.  It's very robust.  You get it with word cues, object cues, memory for world events and news stories, memory for music, memory for sporting events, and memory if a person is asked to write events that would go into a story of their life.
2.  It doesn't happen for younger adults.  This may be due to a memory deficit that occurs as you age.  In particular, as you get older you get less efficient at encoding, this means there are fewer recent memories to report, so you go farther back, and that makes a bump.  But, Jansari and Parkin didn't let people recall recent memories, and when you do that, younger people show a bump too, so that might be wrong.
The implication is this:  If you're between the ages of 10-30, you're currently laying down the memories that will be most recallable for the rest of your life.  This might partially explain cohort effects.  For example, I was in prime bump time in the 80s, and now I'm turning to retro stations that play 80s music, I like 80s movies better, I think things like today's music and fashion, etc. aren't as good as when I was going through the bump, etc.  In other words, if you're in the bump, I hope you're having a good time, because at least part of your “self” will always be here in the memory bump with the memories that are “easiest” to recall.
What makes this interesting to us is to ask why.  What produces the bump?  Why are you recalling significantly more memories from this part of your life than any other?  Here's a preview:  Nobody knows.
Rubin and Schulkind (1997) consider seven hypotheses.  They are:
1.  It's demand characteristics from the experiment.  People somehow think these are the memories you're asking for.
2.  People will tolerate longer searches in this period, which helps them get more memories.  A situation would be if you try to remember yesterday, you have this metamemory knowledge that it should be easy, so if it seems the least bit hard, you stop right away.  If you try to remember 10 years ago, you expect it to be hard, so you look longer before quitting.
3.  You have your important memories in the bump.
4.  The bump memories are more likely to be in the central narrative of your life.
5.  The bump memories are more novel and distinctive, so less interference.
6.  The bump memories are more vivid or emotional.
7.  The bump memories are encoded differently, and that makes them easier to retrieve.
How can we test between these?
What do Rubin and Schulkind find?
1.  They had biasing instructions that conveyed to people the desire to get bump memories, and it had no effect.  So, this doesn't seem likely.
2.  No reaction time differences.
3.  People rated all of the memories on importance.  These weren't any more important.  When people write down the five most significant events of their lives, a lot of them are in the bump, but they are all in the 20-30 decade, not the whole bump.
4.  Bump memories were not rated as being more significant.  They were also not rated as being retold more often.
5.  The memories were not rated as being more novel.
6.  The bump memories were not rated as being more emotional or vivid.
7.  It's all that's left.  What could this be?  Maybe a neurological thing.  Maybe this is a critical period as an organism trying to survive in a harsh world, and that means you should pay more attention here.  Maybe, but we have no data. 

Cognitive Psychology Notes 10
Will Langston

 Back to Langston's Cognitive Psychology Page