Mnemon event diagrams (was Re: On Gatherer's behaviourist

Aaron Lynch (
Sun, 13 Sep 1998 12:18:50 -0500

Message-Id: <>
Date: Sun, 13 Sep 1998 12:18:50 -0500
From: Aaron Lynch <>
Subject: Mnemon event diagrams (was Re: On Gatherer's behaviourist
In-Reply-To: <>

At 04:11 PM 9/11/98 -0400, Bill Benzon wrote:
>Paul Marsden wrote:
>> Yes I agree with all this, but the point I was trying to make that it makes
>> little sense to conceptualise memetics as the study and manipulation of
>> internal thought things inside heads. The chances that our linguistic
>> psychology of thoughts qua serial instructions bear any relation to the
>> instructions operating in our brain are so infinitesimally small that it
>> makes little sense to start doing mathematical operations on them in the
>> hope that they will produce anything more useful than a pair of fetid
>> dingos kidneys.
Bill Benzon wrote:
>This I agree with. Now the question for someone like Aaron is just what
>assumptions is he making about his mnemons. He is certainly trying to
>his assmptions. But he is, it seems to me, assuming that they are discrete
>entities and that these discrete entities can somehow combine. Perhaps he is
>also assuming that when several mnemons combine that they combination can be
>dissolved and the constituent mnemons recovered. And so forth. The
>of discrete entities is problematic.

Bill and Paul,

I should be clear that my system of mnemon event diagrams focusus only
minimally on intra-brain information processing, and does not depend at all
on a general notion of "thoughts qua serial instructions." (I'd suggest
re-reading my paper if you came away with this impression.) The event A -->
~A, (Host of A drops out), for instance, does not address how thoughts are
manipulated inside the head. All it means is that someone who qualifies as
satisfying an observer's definition for hostship of mnemon A makes a
transition to no longer satisfying that definition for hostship of A, as
detected using (generally imperfect) operationalized definitions for
hostship of mnemon A.

In an event such as A*~B + ~A*B --> A*~B + A*B, the combination A*B is
indeed formed anew, possibly for the first time ever. But this is a matter
of inter-individual recombination and not a matter of how thought are
processed and combined inside a single individual. An example would be if
someone who knew how to tie a bow but did not know about laced shoes met
someone who knew how to make laced shoes but not how to tie a bow, with the
result that one of them ended up knowing how make laced shoes tied with
bows. (She might walk away from this encounter with no shoes and no bows,
but only knowledge.)

If I were proposing that there were only one valid system of dividing
knowledge, belief, learned emotional responses, etc. into items such as A
and B, then this would be a problem indeed. However, I specifically say
that there probably is *no* absolute system of memory abstractions--though
I also say that some abstraction systems are more useful than others.
Similarly, physics uses arbitrary "units" of lenght such as meters, etc.,
and rejects the notion of an absolute coordinate system (system of
space-time abstractions) for the universe.

To go back to mnemons A and B above, I do not say that knowledge of how to
lace shoes and tie bows are the only available abstractions for analyzing a
phenomenon, or that they are always the strongest abstractions. It may turn
out that for some population and historical period, a better abstraction is
knowledge of how to lace perforated clothing generally, and knowledge of
how to tie a half bow. We could call these mnemons A' and B', respectively.
They would turn out to be more useful abstractions in studying a society
where A'*~B' + ~A'*B' --> A'*~B' + A'*B' happened first and A'*B'
propagated before A*B did. Even within a given population and historical
period, I do not claim that the mnemons I discuss constitute the only
useful abstractions. Someone might usefully discuss a belief that "bee
pollen is salutary" instead of "bee pollen invigorates," in contemporary
America, for instance.

Similarly, in physics, there might be one phenomenon best analyzed in terms
of a coordinate system with X, Y, and Z axes, and another phenomenon best
analyzed with a rotated and displaced coordinate system of X', Y', and Z'
axes. A physicist can use an X, Y, Z coordinate system to do meaningful
science without asserting that this coordinate system is somehow the One
True coordinate system, or "absolute coordinate system" of the universe.
Indeed, most physicists since Einstein have believed that there is *no*
absolute or privileged coordinate system of the universe.

So my system does not *impose* the kind of *absolutist* discretization of
information of which Bill expresses concern in his post. By contrast in
genetics, there *is* a particularly optimal abstraction system for
discretizing information: the one based on adenosine, guanine, thymine,
cytocine, and (for RNA) uracil. But we do not currently know of such a
universally optimal abstraction system for brain-stored information, and do
not even know if such a system will eventually be discovered. In
particular, we do not yet know if there will be a way of representing
shoe-tying knowledge in detailed neural terms that is *useful* in analyzing
interpersonal knowledge transmission.

My insistance on discussing "neurally stored" information, e.g.,
brain-stored information, does not of itself require knowing the details of
storage or of showing that such details would be useful for analyzing
inter-personal information transfer. There is some similarity to computer
software here, in that the details of storage can vary widely among
computers. Even computers of the same make will store "the same" program in
different disk sectors, with different fragmentations, etc. The differences
become even greater when the hardware varies between machines. Many details
of physical storage are therefore almost useless in discussing the computer
to computer transfer of "software," "data," or (more generally)
"information." This does not, however, imply any Cartesian machine-data
"duality" any more than I have implied a brain-information "duality."
(Caution: this does not imply any perceived isomorphism between brain
architecture and computer architecture.)

I should also be clear that I do not propose to compute on an a priori
basis the retransmission rate of a mnemon. Ultimately, we must measure the
retransmission rates for an operationalized definition of knowledge of how
to tie a shoe. I can still hypothesize that modern parents retransmit such
knowledge to young children in order to save themselves from digging out
terrible knots in their children's shoes, but this does not substitute for
eventually measuring the retransmission rates.

Nothing about my work suggests that responding to surveys is not behavior,
or that observing behavior is somehow diminished as the means of detecting
internally stored information. (A point that goes for computer behavior as
a means of detecting internally stored information, too.) There may turn
out to be useful methods involving neural probes, as Mike mentioned, or PET
scans, etc. as I mentioned before. Yet for the forseeable future, detecting
internal information will depend heavily on observing behaviors. The
reliance on observed behavior to detect internal information does not,
however, excuse us from taking up the population level science of internal
information, as some would prefer.

--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)