[asa] PSCF article on immune system

From: Bill Powers <wjp@swcp.com>
Date: Fri Dec 04 2009 - 14:08:45 EST

I have read the recent article by Craig Story on immunology, and I have a
few comments.

1) The main thrust of the piece is to persuade us that "random" processes
are essential in the vital functioning of the creation of antibodies.
There are, he says, some 10^10 different antibody formations possible.
This vast number of possibilities is accomplished with from only about 500
gene segments. It is possible to generate such a large number of
possibilities from so small a number of gene segments because the joining
of these segments is imprecise and influenced by random processes.

In this way Craig wants us to understand what a marvelous chemical machine
is the antibody mechanism, and the essential necessary use of randomness
to protect from a wide host of unseen pathogens.

So here is my question, one Craig devotes very little time to, since it is
not the primary focus of the article.

With 10^10 possible antibodies, it would seem impossible for a random
system to work. We would all die long before the appropriate antibodies
were found in this haystack. This means that, while randomness may very
well be a crucial part of the system, it cannot be the central governing

It must be something like any search program. You will have to get
"close" before a random search becomes useful. You might begin your
search "randomly", but there must be some sort of guiding mechanism that
quickly narrows or directs the search, e.g, gradient search.

I suspect, from what Craig says, that the search is guided by a "memory"
of previously generated antibodies, which may be why mother's milk is so
important for the survival of children, or colostrum for the survival of
baby goats, cows, etc.

Finally, I wonder that such a system has not got the attention of IDers. It
certainly is an inventive approach to a very difficult challenge.

2) Craig makes an attempt at describing what is meant by random. He
suggests two types of randomness, taken from Peacocke. One type is where
we simply lack sufficient knowledge to predict an outcome. So given
probability distributions for the initial conditions, we can derive or
measure a probability distribution for the outcomes, e.g., the flipping of
a coin. The second type is associated with independent causal chains.

He says, "This is the kind of 'accidental' intersection of two (or more)
unrelated causal chains." If a hammer, he says, falls from a building an
hits you on the head, we have the intersection of two independent causal

What I want to investigate is whether this notion of random, one that I
myself have previously considered, is coherent and what exactly is being
gotten at.

In order to make sense of this notion of random, there are a number of issues
that need to be clarified.

First, we must be able to recognize and delineate "causal chains." Moreover,
we need to be able to make sense of such chains being "independent" of another
such chain. All of this must be accomplished without reference to the concept
of "random."

First, some comments.

1) What we observe is temporal sequence of events. We do not observe causal
chains, at least not unambiguously.
2) From this perspective, it is not clear why the striking of a hammer on
someone's head is any different from the hammer falling from a building. Both
are simply sequences of events. Why call one random and another not random?
3) Whatever causal chains look like, they are different from "random" events.
 4) It would seem that a causal chain is not composed of random events.
5) Random events "break" into causal chains, according to the definition
offered. They are the intersection of "independent" causal chains.
6) So it seems that causal chains are sequences of events that are not
independent. They are in some sense "correlated."
7) What does it mean for sequences of events to be "correlated" or not
independent? It seems to entail some kind of coherence. And this appears to
entail some form of communication or contact.
8) Suppose we say, an apple falling from a tree and striking the earth is a
not an independent event. The collision of a neutral gas atom with another
one is an independent event.
9) The apple falls toward the earth, we say, because of the mutual
gravitational attraction of earth and apple. They are in communication with
each other so that the movement of the apple is not independent of location of
the earth and apple. On the other hand, two neutral atoms move in independent
ways and directions prior to their striking each other. Their paths are not
correlated with each other. They are independent just because they are
non-interacting. There is no communication that could correlate their paths.
 In a sense, they "know" nothing of each other until they suddenly bump into
each other.
10) If this makes general sense, it seems that causal chains of events are
those that have a continuing interaction with each other. Whereas,
independent events are non-interacting. The interaction of non-communicating
events is a random event.
11) So the history of events that we associate with the movement of planets
and sun in the solar system are "causal chains" of events. These events are
in continual "contact" with each other so that the events of the various
components are correlated. They are, in a sense, locked into each other,
operating as something like a unit. Events that might go on in the interior
of the sun, causing a violent explosion, might be considered, from the
perspective of the solar system as random. The solar interior events are not
linked, not correlated, but rather independent of the motion of the planets.
12) Independence of sequence of events entails that there is nothing about the
character of the characters associated with the events that will inform us
when, how, and where future events between them will occur.
13) This independence is not about predictability. We can possibly predict
when two neutral atoms will collide, knowing their locations and velocities.
But this does not make the collision correlated. Correlations entail that
there is a kind of necessity associated with the when and where of the
14) If interactions are not "continuous," does that mean they cannot be
"causal"? Once the contact between events is severed, even for a moment, it
seems that the event histories break the causal link, so that subsequent
interactions take on a random character.
15) We can imagine that gravitational fields were turned on and off that a
randomness could be introduced into such interactions. When the solar
gravitational field were turned off, the various planets would begin flying
off in nearly straight lines, but still in "contact" with each other. Were
the solar gravity turned back on shortly thereafter, the exact nature of the
interaction would take on a random character, although it would be still
somewhat correlated. The issue that comes up is one of time scales. If the
time scale for the planetary movements is large compared to the period during
which the solar gravity is off, then the planetary motions appear to retain a
correlated, causal link, and therefore their motions appear correlated and
16) What if all interactions were quantized? Would the discrete nature of
the interactions introduce a randomness into all events? It would seem that
it would have something to do with the degree of discretization. Such
randomness might only be observable with very weak fields, where quantization
becomes more important.

Anyway, that was fun. Does it make any sense at all? What do you think?


To unsubscribe, send a message to majordomo@calvin.edu with
"unsubscribe asa" (no quotes) as the body of the message.
Received on Fri Dec 4 14:09:11 2009

This archive was generated by hypermail 2.1.8 : Fri Dec 04 2009 - 14:09:11 EST