RE: Measureing Memes

William Chambers (
Fri, 4 Jun 1999 10:45:51 -0500

From: "William Chambers" <>
To: <>
Subject: RE: Measureing Memes
Date: Fri, 4 Jun 1999 10:45:51 -0500

On page 196 of the 1989 version of the selfish gene, Dawkins says:

"If Darwin's theory can be subdivided into compnents, such that some people
believe component A but not component B, while others believe B but not A,
then A and B should be regarded as separate memes, If almost everybody who
believes in A also believes in B- if the memes are closely linked to use the
genetic term- then it is convenient to lump them together as one meme."

If this statement is taken as definitive of memes, then there are clear
measurement implications. In terms of mathematics, what Dawkins has
described is akin to a dichotmous factor or abstraction, The meme is like a
true score in classical true score theory. The items (components) are each
parallel tests of the latent true score. As such the items should be
correlated with one another, The result of adding such correlated items to
obtain the total score estimate of the latent true scores, is that we form
an abstraction, in which noncorrelated variances are removed from the total
scores, Errors and multidimensionality cancel out. This type of structure is
measured by factor analysis. This seems to be what Dawkins has in mind for

An alternative to abstraction is construction. Constructs can be the sums
of uncorrelated items, Each item is entailed by the whole construct, though
the separate items do not imply one another.This is the kind of structure
measured by corresponding correlations/regressions. Dawkins seems to be
denying the constructional nature of memes and the possibility of a meme
integrating independent components,

IF Dawkins is correct then memes are highly limited structures, in which
wholes and parts are essentially the same thing. Constructions, on the
other hand, bring together items that are uncorrelated. For example, let
the "immaculate conception" be item x1. Let "No redemption by deeds" be x2.
Some people who believe in God believe both x1 and x2, Some believe x1 or x2
but not both, And some believe neither x1 or x2. A constructional model can
handle this situation by the creation of a whole that is distributed across
a continuum of Christian Fundamentalism.

By requiring the all or none correlation of components, Dawkins denies the
possibility of continuously distributed memes, This, in turn, prevents the
hierachical integration of components in any multidimensional way, Is this
the way memes are defined today?

Bill Chambers

This was distributed via the memetics list associated with the
Journal of Memetics - Evolutionary Models of Information Transmission
For information about the journal and the list (e.g. unsubscribing)