Magic numbers

David Lomas dropped me a note to bring my attention to Donald Clark’s blog, which in turn (acknowledging all your sources can be hard at times) references research from the University of Missouri which purports to show that our working memory struggles to retain more than three to four items. The original work by Miller had the number at seven plus or minus two. Now I have long used five as a working number (the lower Miller limit) along with a trust limit of 15 and Dunbar’s number 150 as representing three natural limits on community size. Details here. I don’t think I am going to change them, although I am happy to acknowledge I find three easier than five, but that may just be old age! In part I don’t think its just about number/letter presentation. If you do experiments on letter recall then there is an absence of context which would not be present in other datasets, or ones with associative links. Looking at crews the number of five rather than three seems a more natural limit (more on this later in the week). Of course three and seven have always been magical numbers in the celtic and other traditions so that makes a case for both