Re: Memes are Interactors

Aaron Lynch (
Mon, 20 Apr 1998 17:06:08 -0500

Message-Id: <>
Date: Mon, 20 Apr 1998 17:06:08 -0500
From: Aaron Lynch <>
Subject: Re: Memes are Interactors
In-Reply-To: <>

>Dear Mr. Lynch,
>you wrote in "Thought Contagion":
>"Another social barrier to thought contagion, the credential system,
provides a method of rejecting important kinds of belief transmission from
the uncredentialed. Those holding a doctorate, professional license, or
clergy post thereby gain more access to minds than those lacking such
distinctions. People do make exceptions, since most are themselves
uncredentialed. Yet highly credentialed individuals may apply credential
systems more vigorously, limiting the acquisition of common thought
contagions among their ranks. Moreover, less credentialed people can
recognize the restrictive effects of credential systems well enough that
they don't even try to impart beliefs to someone with impressive
credentials. All of this helps restrict the circles in which specific
thought contagions can travel."
>I deeply admire your and other people's work on the memetics field, as I
deeply admire any work done by anyone, with or without credentials, who is
attempting to make this World a better place to live. IMHO (thanks to Bruce
Howlett), Credentials are not equal to Competence. Credentials are, as one
of my mentors at the Academy used to say: "the license to forget everything
I have said to you and start from the beginning". That is, Credentials are
a rigid structure granted for a past work, and Competence is a dynamic
structure showing the present "fitness" of a specific system to perform
some imminent future work. Please do not understand me wrong. I surely will
never be "credentialed" enough to attend this discussion on your level, but
I "feel" that my competence in fields other than memetics can help this
list to gain some answers on FAQs that, as I could see, are bothering us all.
>The reason I joined this list was not "looking to make money by pandering
to the masses" (If it was really my intent I would probably start by
currying favour to you as one of the "fathers" of memetics). I simply run
into this list during my meandering path of searching for answers that I
(obviously wrong) believe someone have already found. You still didn't
prove my incompetence for "doing the job" and so I will continue to make
replicas of my thoughts and spreading them in the environment of this list.

I don't mean to say that anyone has joined this list to make money. There
is, however, a phenomenon of physicists "selling out" and using their
credentials to write profitable books expressing ideas that are in fact
unsupported by modern physics--and which the authors themselves may not

>Now, I would like to make an attempt to clear some terms from your
discussion with John Wilkins:
>In sistematics we define ENTROPY as a previous measure of outcome
uncertainty when making an experiment. If we are performing an experiment
for which we do not know for sure which will be the result, but we know
that the set of all possible results consists of i=N elements, and each
result has a probability of happening p(i), than the quantity of
INFORMATION gained at the end (by knowing the outcome) is (according to
Shannon) the N-sum of all the products for the probabilities p(i) and their
logarithms log2p(i). If all N results of the experiment have an equal
probability of happening, then every p(i) is equal 1/N, thus the
uncertainty (entropy) is maximal (log2(N)). If one particular outcome has
the probability p(k)=1 (the event will surely occur) than all the other
results must have a probability of happening that equals 0, consequently
there is no uncertainty about the final outcome (we suppose that the
product 0*log2(0)=0). In all other cases we have a situation where
0<H<log2(N), or the exact (calculated) measure for entropy (H). Moreover,
the received message about the final outcome of the experiment eliminates
the previous uncertainty and caries a quantity of information that is
exactly equal to the previously calculated entropy.
>In my knowledge, the same calculus is used in statistical physics with the
difference that instead of a logarithm with the base of 2 (log2) a
"natural" ln function (with the base of e) is used. I believe that Boltzman
already defined entropy as a "measure of lack of information about the
state of a physical system".

I don't really mean to get into an extensive discussion of thermodynamics
or statistical mechanics. There is, a definition of entropy as S = k ln(w),
where w is a probability of being in a statistical mechanical state, and k
is the Boltzmann constant. A good point to remember, however, is that the
statistical mechanical definition of entropy takes into account all the
state possibilities at the level of atoms and molecules, not just synapses.
This makes it very hard to apply the second law of thermodynamics directly
toward understanding the direction of information changes in the synapses.

>But this is not so important for our discussion. What is important here is
the fact that the "Observer" (the system that is performing the experiment)
MUST have some information BEFORE the experiment (the number of all
possible states of the observed system, the probability for every of the
possible events to happen). Moreover, the message (signal) with the encoded
resultant outcome of the experiment caries information ONLY for the
Observer system that is interested in it and Knows (calculated) the
previous measure of uncertainty.
>In systematics we are dealing with rigid (deterministic) systems. For such
systems the calculation of the previous uncertainty measure is quite easy.
For example, if we know that the system under experiment is in
such-and-such state, we can be pretty sure that if we apply to it
such-and-such signal it will go in such-and-such state.
>Even in genetics we can very exactly foresee the final outcome of various
interaction/replication processes. The offspring will be very similar to
its parents, if we change the environmental conditions we can predict what
repercussion will they have to various species, etc. Genetic individuals
are reactive, they can only RESPOND to environmental stimuli. They are
"programmed gene carrying machines".
>In memetics this is not the case. We are dealing here with dynamic
(self-organized) systems. Such systems are proactive, not reactive. Unlike
genetic individuals which are evolving through selection in the biosphere
environment, meme carrying individuals (hosts) are evolving during their
life time, they can even change their "species" (see Athony Kenny's example
in "Viruses of the Mind").
>A genetic individual (human's brain) is the ONLY environment in which
meme/mnemon selection could be performed. So we can not state that memes
can be stored in structures different from brain neurological structures
like floppy disks, computer memories or other artifacts. They are only
(matter-energy) structures produced by dynamic meme carrying hosts. There
is only one kind of meme, the "active" (interacting) one. There is no
"passive" kind of memes or "potential" memes. A floppy disk is a rigid
structure, like a stone, and it is susceptible to degradation like any
other rigid structure. It is passive, it is completely dependent about the
environment. Even if such rigid structure is purposely "directed"
(transmitted) towards some other particular system able to decode them, the
host that produced the structure have no mean to control which information
(meme), if any, will the receiver host (system) produce upon this
structure. It is all matter of probability. When you are speaking or
writing a letter or a book you are not transmitting memes to other people,
you are only producing replicas, you behave according to your memetic
phemotype. Memes/mnemons will be (probably) PRODUCED and will survive in
some other people brains making them to behave like you (make replicas
similar to yours).
>Any of this hosts will have its particular reasons for behaving according
to the "transmitted" meme. Some will accept it without "understanding" it
only because they like you, some will find in it some portion of
information that fits well enough in some other part of their memetic
structure, or have any other interest why it is better to accept your
thought than reject it.
>Now I would like to put some explicit questions:

>If memes are not interactors, if their replication and instatiation in
other hosts depends only from their "inherent quality" and not because of
"how well do they fit in the present memetic structure of the host", then
why memes of "high quality" does not instatiate in more hosts than "lower
quality" memes?

I don't say that memes are not interactors. They are interactors, but I do
not happen to use this term as a central, organizing term for the
fundamentals of memetics. I discuss a variety of interactions of memes with
hosts, and with the broader environment by way of host-host, host-non-host,
etc. interactions, but do not specifically focus on this word. It somehow
seems redundant to what I am already saying. I'm not really sure that I
could present my arguments more clearly by using the term "interactor,"
either--but the recent discussions certainly have me thinking about it.
(Some in social sciences may regard the term as jargon that makes the prose
more specialized or difficult to follow.) The qualities of memes that get
them replicated are not "inherent," but does need to be consistent over
enough different individuals and across a long enough time span to have the
kinds of population level effects that I talk about in my book and papers.
Interaction with a very pervasive aspect of human nature, such as our
sexual reproduction system, can make a meme's advantage seem "inherent,"
even though it is not. For instance, the replication advantage of
anti-abortion memes depend critically upon being held in a species such as
mammals, for which abortion can reduce the spread of memes to offspring. I
will look forward to reading Hans-Cees Speel's symposium paper on memes as
replicators to learn more about the advantages of explicitly using the term
"interactor" rather than just analyzing the relevant interactions.

Many reasons, of course, exist for low quality memes to spread, as
discussed in my book. One simple reason is that low quality memes can often
be expressed more easily and in shorter time; and can then be learned more
easily too.

>If memes are not interactors, biased to selection in human's brain, what
are the memetic interactors and where do their selection happen?

Selection happens at various levels, including, for instance, biased
receptivity and retention by the brain--in addition to transmissivity

Another term that I use quite sparingly in the "first principles" approach
of my recent paper is "selection" itself. (And the word "fitness" does not
even appear.) Instead, I discuss event forms and event rates that, when
differential, ultimately make up selection.

--Aaron Lynch

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)