[From Bill Powers (970822.2005 MDT)]
Chris Cherpas (970822.1652 PT)--
I am assuming that e. coli's much-discussed control of the perception
of "concentrations" is a kind of intensity control. Is that right?
Apparently not. The variable that E. coli is sensitive to is not
concentration, but rate of change of concentration (the first time
derivative). Swimming in a uniform concentration, E. coli will tumble at a
baseline rate -- the same rate whatever the concentration. So technically,
E. coli's perception is a transition perception. Obviously, however, it's
not "transition" in the same sense we mean in the neural hierarchy, where
we suppose that transitions are derived from configuration, sensation, OR
intensity signals. E. coli's sensors are just sensitive to the first
derivative of an environmental variable.
It is said that an organism is _born_ with some ability to control
the perception of intensity. Isn't the "organism" also capable of
intensity control at _conception_, since, at a biochemical level,
some form of intensity control is already on-going? Is there
anything more than intensity control at the biochemical level of life?
Intensity means only "intensity of stimulation of a nerve-ending." Other
than that, the word doesn't really have any special physical significance.
It could be that some nerve-endings are placed so that the stimulation they
receive depends on a rate of change of something else in the environment
that we are more likely to notice than the actual energy absorption by the
sensory ending.
At the biochemical level, the "sensors" are probably enzymes or the
substances that immediately affect the state of activation of an enzyme.
However, what is sensed could depend on concentrations of other substances
in complex ways, so what is effectively controlled by a biochemical control
system might be just about anything. Biochemical control systems can be
just as complex as neural ones, at least in comparison with the lower
neural levels. I think.
The basic thing to remember is that the medium that carries a signal is
irrelevant; all it has to do is represent an amount, at any level of
organization. If you have a signal representing "symmetry," the actual
signal indicating degree of symmetry only has to be able to change from
zero (no symmetry) to some maximum amount (perfect symmetry). What matters
is not the signal, but the computing functions by which it is derived from
lower-level information.
There are many levels of biochemical control, starting with the molecular
level inside cells, then at the cellular level, at the level of organs, and
at the level of whole organ systems. As in the neural systems, at any level
the signals themselves are all of basically the same kind, chemical
concentrations that can increase or decrease. But control of a signal at a
high level implies controlling many signals at a lower level, as in the
signals that reach the pituitary, which apparently contains numerous
chemical comparators that produce the error signals for the
hormone-mediated organ control systems. There might be perceptual functions
that make one chemical concentration a function of multiple concentrations
at lower levels, so more than simple "intensity" perceptions are possible.
But since we don't have experiences of such sensation signals, we have few
guidelines as to what they might represent.
This is all guesswork, of course, because biochemists just haven't looked
at the chemical systems as analog computers. There's no reason you can't
have biochemical adders, subtractors, multipliers, dividers, and
integrators; given those components, one can design analog control systems
of all kinds and all degrees of complexity. The biochemical systems are
probably harder to sort out than the neural ones, because instead of
following insulated pathways carrying signals from one place to another,
biochemical signals are all mixed together in the same circulating soup,
and are distinguished by being carried by different molecules. The "target"
cells ignore all the signals except the ones that they are equipped to
detect. So the effect is much like having insulated pathways, except that
the limit to the number of pathways is set by the ability to distinguish
one molecule from another. That probably makes the possibilities much more
limited than they are in a nervous system (which might account for the
development of the nervous system).
I have a feeling that it's going to be a LONG time before biochemistry
starts looking at the chemical _systems_ instead of just fragmentary bits
of chemical reactions.
I haven't looked at it in a while, but I recall Gerald Edelman's
"Neural Darwinism" includes the notion that the epigenetic
development of an organism depends on the behavior of cell
adhesion molecules. Has this been discussed yet from the standpoint
of control theory?
That's really much too vague an idea to be of much help. What behavior, of
what molecules, embedded in what kind of whole system? I don't think anyone
is anywhere near ready to explain processes at that level.
Best,
Bill P.