Re: List of meme definitions

Adrian Kelleher (
Sat, 11 Apr 1998 05:09:27 +0100

Message-Id: <>
Date: Sat, 11 Apr 1998 05:09:27 +0100
From: Adrian Kelleher <>
Subject: Re: List of meme definitions

Aaron, thank you for your prompt and thoughtful reply (full text at bottom)
to my mail.

You write:

>Yet if you do write about memes as replicators, then you need a
>definition that makes sure that ALL of the instances labeled as "memes"
>are indeed replicators.[...]Dawkins does not insist that a particular
>instance of gene succeed in replicating in order to be called a replicator.

These two statements strike me as contradictory....I'd tend to go along
with the second of them and state that if _any_ instance of a mnemon is
replicated then _all_ instances of the same mnemon are memes.

There are some good reasons for taking this approach:

1) Once one communication/replication event has been observed, it
can be stated that all instances of the mnemon concerned are memes
without having to investigate their individual origins. Such investigation is
futile in any case as it can only be known for certain that an individual
_claims_ to have acquired a particular mnemon from someone else.
Often not even the individual concerned may know, especially with
existentially trivial (but perhaps scientifically intereresting or studyable)
mnemons such as buzzwords/phrases or Dawkins' 'reversed baseball
cap' meme.
2) This expansion of the definition doesn't admit any more classes of
mnemons to the set of all memes - only instances of mnemons which
were classified as memes in any case. Uncommunicable mental
information (e.g. recollections of sensory stimuli or emotions) and
information that is not in practise communicated is still excluded.
3) A rider to the definition refining it to only include mnemons that
propagate substantially through communication/imitation as opposed to
invention/realisation can still be added.
For instance, under the existing definition, if I awake from a coma and
ask you what year it is then we are both hosts of the meme 'It is 1998'.
Without the rider, ALL hosts of the mnemon become hosts of the meme,
but with it its not a meme at all.
This excludes a class of mnemons allowed under your definition.
(Paradox/exception: 'It is day'/'It is night' in a population of sightless
persons. Some mnemons will predominantly replicate in certain
populations and be predominantly 'discovered' in others.)
4) This allows us to describe the mnemons A*B in two individuals as
equal in all circumstances.

My own *favourite* would be a definition that captured the nature of
'knowledge packets' (e.g. 'Christianity', Marxism', 'Logical Positivism'), the
constituents of which are rare in isolation but common in unison.
Something like 'A set of memory items that tend to propagate together',
but with mathemathical rigour. Perhaps this can be called something else
- 'Mutually reinforcing mnemon set' or something similar.


>>> Aaron Lynch <> 04/09/98 02:39pm >>>
These are reasonable questions, Adrian.

>(A. Lynch, JoM Vol 2/No. 2), the definition of meme quoted elsewhere in
>the paper results in some odd conclusions, in particular:
>.... A*B formed by the event ~A*B+A*~B->~A*B+A*B is not a meme,
>although it is meme-derived.
>Yet as mentioned earlier, A*B can also propagate as a set by the
>non-parental conversion event A*B+~A*~B->2A*B. Formed this way,
>A*B is a meme. Thus, the host population of A*B is yet another mixture
>memes and meme-derived mnemons.
>[end of extract]
> (under the definition given, classifying only those memory
>abstractions arising from previous instances of an 'identical' memory
>abstraction as memes), A*B is sometimes a meme and sometimes not a
>meme, depending on the circumstances of its creation - this strikes me
>as untenable for the following reasons:

I don't use the term "identical."

> i) A*B=A*B=A*B - equal things should have equal attributes, so
>A*B should surely be a meme either always or never, otherwise the
>notation needs to be changed.

"Sameness" of two instantiations of a memory item does in general not
sameness of causation. So you are not entitled to insist that because
people have "the same" memory item, that the causation must have been
same. Therefore, you cannot, in general, insist that the both have the
attribute of "replicator."

> ii) The host populations of the meme and non-meme varieties
>cannot easily be distinguished.
> For example, in the bee pollen example quoted in the paper, is the
>'bee pollen invigorates' belief in a purchaser a meme or not a meme? We
>need to find the original persuader (the merchant in this example) to find
>out (in order to verify he wasn't being disingenuous). This may
>frequently be impossible.

Someone being swayed by a lie has a heteroderivative mnemon for the
reason that someone being swayed by an honest misstatement does.
The most
useful applications of memetic evolution theory, however, are generally
when the homoderivative instantiations of a mnemon have become
overwhelmingly predominant--as in the case of religions, family mores,

> iii) Under this constraint, a novel idea or discovery which
>propagates through a population is a meme for all hosts except the
>discoverer! (S/he is the only host who's belief doesn't originate from a
>prior instance of the same belief set).

The definition I have chosen to use is the more parsimonious of two
possibilities. But instead of going with just homoderivative mnemons, we
can also define "meme" more broadly to include homogenic mnemons as
That yeilds the following:

MEME: A memory item, or portion of an organism's neurally-stored
information, identified using the abstraction system of the observer,
instantiation depended critically on causation by prior instantiation of
the same memory item in one or more other organisms' nervous systems,
which has caused a subsequent instantiation of the same memory item in
another organism's nervous system. ("Sameness" of memory items is
determined with respect to the above-mentioned abstraction system of

Though wordier, this definition at least includes discoverers. You may
still, however, find it odd that you still do not know if a mnemon is a
meme until it has replicated. You can still also have populations that mix
meme and non-meme instantiations of a mnemon. So if you don't like
circumstances, then you may prefer to write about mnemons rather than

Yet if you do write about memes as replicators, then you need a
that makes sure that ALL of the instances labeled as "memes" are indeed
replicators. If you do not, then people will rightly regard you as
mislabeling all sorts of non-replicated and non-replicating mnemons that
occur as private re-invented thoughts in some but imitated thoughts in
others. Much of the unconscious resistance to memetics is actually
resistance to versions that seem far too glib about labeling ideas as
"replicators." It is not just the people who are able to consciously put
this objection into words.

Taking homoderivative status as sufficient for "replicator" status is
consistent with the usage in Dawkins and elsewhere. Dawkins does not
that a particular instance of gene succeed in replicating in order to be
called a replicator.

Because memes are the NEW replicators, we have a problem that
resembles a
mix of replicating and non-replicating molecules in proto-life. Some
instances of an oligonucleotide, for instance, may have formed from
recombination of smaller pieces, while other instances may have
from rare (due to lack of enzymes) autocatalysis from a prior
of "the same" oligonucleotide. To speak strictly, you cannot therefore
the entire population of molecular isomers and uniformly call them

The problem is not that my position is untenable, but that I do not present
a level of simplicity that can only result from oversimplifying a complex
mixture of causations. In a technical presentation of memetics, it is far
more important to remain accurate than to simplify. Many of the most
applications of memetics, however, are ones where the vast success of
replicated instantiations renders the distinctions between meme and
non-meme instances of a mnemon moot. In these cases, you are not
by any real task of distinguishing host populations of meme and
instnaces of a mnemon. My book Thought Contagion focuses on just
applications (where simplification is permissible), and therefore does not
need to go into the meticulous distinctions found in my paper.


--Aaron Lynch

THOUGHT CONTAGION: How Belief Spreads Through Society The New Science of Memes Basic Books. Info and free sample:

=============================================================== 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) see:

=============================================================== 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) see: