Re: Measuring Memes

William Chambers (
Mon, 7 Jun 1999 15:38:09 -0500

From: "William Chambers" <>
To: <>
Subject: Re: Measuring Memes
Date: Mon, 7 Jun 1999 15:38:09 -0500


Thanks for your response and insider's view of the state of memes. As a
methodologist, I must have some notion of the definition or lack of
definition of memes before I can begin an analysis. Thus far it appears
there are a number of definitions. So far it is unclear to me whether this
is because the discipline can not get beyond mere unilatereral declarations
or if the crucial issues have not yet crystalized or if there really is no
such thing as memes. Perhaps it is simply a specialization in search of a
definition? This can be a formula for hollow expertise.

The idea of correlation has come up in several recent posts and this seems
to be central to Dawkin's notion of meme. The problem with traditional
correlations is that they only reflect associations of events. The
behaviorists built there tower on the British empiricist's notion of
association, with the implicit notion of a tabula rasa (blank slate) mind,
In recent decades, some behaviorists have been willing to entertain
biological predispositions to learn certain kinds of associations but the
field still lacks a sensitivity to associations based on logic instead of
mere contiguity.

If we limit our definition to correlation as mere association, then memes
are simply ideas that are bound together by habit or circumstance. The
"superstitious" behavior of pigeons and people fall into this "meaningless
clump" model of cognition. There is really no logic binding the ideas
together by any operation. If the mind takes on a more proactive and
constructional role, however, then memes are the sum of ideas that are woven
together in a meaningful way, whether they are correlated in the general
semantic space or not. The binding element then becomes how the person
composes the meme using logic rather than mere co-occurence. This goes
against Dawkin's explicit model of memes but seems to support his broader
use of the term,

For example, Dawkins refers to the "belief in God" meme, There are many
components of this believe, which may or may not occur together in any
particular belief. Each and every component integrated by the individual's
belief, however, characterizes that individual's belief in God, Contrast
the following two models of memes.

Correlated components Model (CCM)

x1 x2
1 1
2 2
3 3
4 4

In the correlational model, x1 and x2 are correlated but the values are not
conjugated in every possible combination, The correlation is perfect and
this makes x1 and x2 appear to be the perfect example of what Dawkins
explicitly defined as a meme.

Now consider the The Manifold Model (MM):
x1 x2 y x1 x2 y x1 x2 y x1 x2 y
1 + 1=2 2+1=3 3+1=4 4+1=5
1+ 2=3 2+2=4 3+2=5 4+2=6
1+ 3=4 2+3=5 3+3=6 4+3=7
1+ 4=5 2+4=6 3+4=7 4+4=8

In the Manifold model x1 and x2 are uncorrelated but are combined in every
possible way. (Corresponding correlations can nontheless say that y=x1+x2.)
The thing that binds x1 and x2 to form y is not reflected in the traditional
Pearson correlation. Corresponding correlations does, however, indicate that
some arithmetic operation logically binds x1 and x2 to form y.

These two models are radically different. In the former, there is no room
for variation, In the manifold model, however, there is great variation that
is bound together by some operation into a common construction, We use
addition, subtraction, multiplication or division as the purest examples of
operations in simulations, The point is that nature somehow puts the parts
(x1 and x2) together in a manner that they are mathematically tractable (if,
in fact, memes can be measured).

The advantage of the manifold model is that the composite y can be treated
as a construction such as "belief in God" though the details of that belief
can vary dramatically from individual to individual.

The disadvantage of the correlational model is that the most we can do is
factor analyze the correlation matrix and come up with models of common
variance, completely independent of any inference of operational
combination, Since the operational combination is the mathematical function
for cognitive combination, then the correlational model precludes our
inference of meaningful constructions in memes.

The tautology in the manifold model is not circular, It is asymmetric. y is
tautological with x1 and x2 but neither x1 nor x2 requires y for their

Bill Chambers

-----Original Message-----
From: Gatherer, D. (Derek) <>
To: '' <>
Date: Monday, June 07, 1999 2:27 AM
Subject: RE: Measureing Memes

>If this statement [in previous post] is taken as definitive of memes...
>Unfortunately it isn't, although some may tend in that direction. As far
>memes are concerned, nothing is definitive, because we memeticists can't
>agree on very much.
>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?
>Again the question cannot be answered in any way that would be
>representative of the majority of memeticists. Most of what has been
>published in JoM over the past year or two has been an ongoing argument
>how memes should be defined.
>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)

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)