The March 3 2008 New Yorker had an article, "Numbers Guy", about Stanislas Dehaene, a young neuroscientist based in Paris, who is exploring the co-evolutionary relationship that occurs in each of us between innate functions in our brains, and our "mother tongue", to produce our numeric abilities. The URL (bottom) is still live for me, at least:

"Because Chinese number words are so brief—they take less than a quarter of a second to say, on average, compared with a third of a second for English—the average Chinese speaker has a memory span of nine digits, versus seven digits for English speakers. (Speakers of the marvellously efficient Cantonese dialect, common in Hong Kong, can juggle ten digits in active memory.)"

www.newyorker.com/reporting/2008/03/03/080303fa_fact_holt?currentPage=all

Dave,

Your travel schedule might have caused a momentary confusion - but my recall is that the original work by Miller suggested that people could retain 5 plus or minus 2 items at any one time.

Oops! Please ignore/delete my last comment - I obviously got distracted by the Cup Final (Cardiff sadly losing 1-0 with a couple of minutes to go). But thinking about it experientially, Miller's revision makes sense. Who can remember nine items at any one time unless it's some kind of party trick?

Some of the science base shows that the short term memory is a function of time, so the number of items would vary with language

I wrote this off line last night, before reading the above comments. This is my take.

I think there are a few understandings/definitions of working memory. Dave, you justify using 7 +/- on the basis of time rather than as a number people can handle. I have not read the article from the sciences academy, but I think the 3 items is what can be remembered in a glance – and Klein (Sources of Power) found that the RPD model uses 3 variables to make 6 simulations. So I think Klein and the academy would be in agreement.

As far as I understand working memory is 9-11 seconds. There has been substantial research done by neurological psychologists on education efficacy in Africa. In our curriculum (South Africa) you learn to read until about 8 years and from then on you read to learn. What they have discovered is that due to poor teaching (you don’t want me to go into it) before 8 years of age, our learners are not learning how to read quickly enough to finish sentences in under 11 sentences. This means they get to the end of the sentence and can’t remember the beginning because they have taken longer to complete the sentence than their working memory is – for this (and other reasons) they cannot read to learn and thus we have an education crisis. Therefore it is important to talk about working memory as a time/it is measured in time.

In terms of whether 5 is the right number for the chaos domain – I would put forward the following 2 ideas –

First - I am interested to see if others do the same – if I count (for instance) a lecture hall full of students I will count in three's if the class is small (like less than 30), if it is large I will count in five's (like 200) – however it takes time for me to get used to counting in fives, so it is more efficient to count in 3s for smaller groups, counting in 3s feels natural from the start. If the group was about 100 students I might count in 4s, it will still be unnatural to start, but I will get used to it.

Second – if that article is right and the number 3 is correct – if you work in a group of five to address a chaos issue and each person puts an idea on the table does it mean your working memory will remember your idea plus three others, or three of the ideas put on the table? If it is yours plus 3 then 4 seems like the right number for me, if only 3 then maybe 3 is a better number. However, I see the numbers in the Cynefin framework as guides and as maximums – we trust 15 or less, the 150 is actually 148 etc.