Message-Id: <199706060235.WAA23533@global.dca.net>
Date: Thu, 5 Jun 1997 22:37:08 -0500
To: memetics@mmu.ac.uk
From: perpcorn@dca.net (Timothy Perper/Martha Cornog)
Subject: Re: Lamarckism
It's useful to think of a specific device when discussing if a
meme-handling system is or is not Lamarckian. We imagine a robot,
internal nature unspecified, that inputs "memes" from an instruction tape
and emits (a) objects build according to design parameters on the meme
tapes and (b) emits copies of the meme tapes themselves. It might be
programmed sometimes to do only one of these things, but it can, in
principle, do both.
Now we, who stand outside the robot, *alter* one of the objects thus built.
If the robot is Lamarckian, the meme tapes emitted after the alteration
will themselves be different, and reflect the presence of the alteration we
caused. (Note that I said that the meme tapes will be different -- this
means that output meme tapes will differ from the input meme tapes.)
The reason is this. The objects built according to design specifications of
the meme tapes are "phemotype," whereas the meme tapes themselves are input
and output "memotype." If characteristics acquired by the phemotype are
inherited, then the memotype will be changed accordingly. This *defines*
Lamarckian inheritance as the term is used in genetics.
The occurrence of Lamarckian inheritance in the robot obviously depends on
whether or not an information feedback loop exists between output (in this
case, the objects made) and the input (in this case, the meme tapes). In
DNA-genetic systems in living organisms, no such feedback exists, due to
the so-called "central dogma" of biochemical genetics -- in plain English,
no way exists to alter the nucleotide sequence of the DNA according to
information from phenotype. However, two sorts of meme-handling robot do
exist, and they differ in this matter.
One of them -- whose properties give us the word "robot" as a synonym for a
mindless idiot -- will likewise NOT alter its internal tapes even though
something external changes the phemotype. Thus the thing cranks out wheels
that are slightly oblong, and a machinist has to regrind them all manually.
But the robot pays no heed -- it simply cranks 'em out, and they're always
slightly oblong. (You see, when the wheel blanks are mounted on the
machine, they slip slightly out of alignment, but the robot never notices
-- it just grinds 'em all oblong. A "wheel blank" is the piece of metal
from which the wheel is ground.)
By contrast, another set of machines *monitors* its own output, testing the
output sequentially or sampling it. If output deviates by some amount from
a predetermined level, set by quality control considerations, then a
corrective feedback loop inside the robot kicks into action, and (a) resets
the internal operating controls and (b) adjusts, if necessary, the
instruction (or "meme") tapes themselves. This device *is* functioning in
Lamarckian mode, although engineers who have been designing such machinery
for the past half-century do not use such biological jargon. Instead, they
speak of self-corrective feedback. (In this case, the machine adjusts the
position of the wheel blanks, writes the adjustment into the meme tape, and
stores, uses, and outputs the corrected tape.)
I want to stress a point -- the principle of corrective feedback is widely
used in industry. Similar principles are also used in devices which track
a moving object (e.g., targeting computers in the military). Most
electronic circuitry functions in this self-correcting mode. (If anyone
wants some examples, I will be glad to oblige -- for instance, devices
known as "operational amplifiers" are designed specifically to work in the
feedback or self-corrective mode.)
If the preceding is clear -- and I hope it is -- we can now ask if human
beings more closely resemble mindless robots who receive meme tapes and do
*exactly* what is specified on our meme tapes regardless of the output OR
if we receive feedback from what we do and alter our internal (input) meme
tapes accordingly, and, as a result, produce new objects AND different
output meme tapes. Obviously, the answer is sometimes both -- for
repetitive tasks, we do the first. However, human beings are notable for
our ability to do the second as well -- we can "learn from experience" and
tell other people how to do it right.
Not always, but sometimes. When we do not, we are non-Lamarckian. When we
do, we are Lamarckian.
Once again, if one thinks in terms of information flow and feedback, the
ideas can be handled reasonably straightforwardly. I hope this helps.
Tim Perper
===============================================================
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: http://www.cpm.mmu.ac.uk/jom-emit