Dead Wrong, VI Kibbitz

[From Rick Marken (971230.1400)]

Bruce Abbott (971230.1500 EST)

Seriously, that's what distinguishes Life from Nonlife, isn't
it? Control, I mean. You can tell you're dead when you fail to
resist _all_ disturbances.

No. You can tell you're dead (assuming that you are still conscious;-))
when you don't want anything anymore; when you are no longer
controlling any input variables. When there are no longer controlled
variables, there are no longer variables that are "disturbances" to
these variables. When you die, disturbance variables are simply
_causal_ variables. For example, when you are no longer controlling
how close the edge of a scalpel comes to your skin, the location of
the scalpel is no longer a disturbance to a controlled variable
(distance from cutting edge to skin); rather, it is a variable that
causes variations in the state of another variable (depth of scalpel
into the autopsy incision).

There are only "disturbances" when there are controlled variables.
And there are only controlled variables when you are alive. A
person who fails to resist all disturbances to a controlled variable
has either not yet learned to control the variable or is trying
to control a variable that is uncontrollable (because the person
cannot possibly generate the outputs that compensate for _all_
normal disturbances to this variable).

Bruce Abbott (971230.1545 EST) --

The experiment involves testing individual rats on a series of VI
schedules differing in average interval. PCT predicts that, if
the rats are controlling the rate of pellet delivery, then
lengthening the schedule parameter (e.g., from 10-s to 20-s) should
result in an increased rate of responding as the rats attempt to
compensate for the reduction of pellet delivery rate imposed by the
lengthening of average interval.

More important, it predicts that the rate of pellet delivery will
be about the same under both schedules [I presume that you are
measuring the state of the hypothetical controlled variable]. These
predictions are made _only_ if 1) rate of food delivery is, indeed,
the controlled variable (some other variable could be controlled,
like rate of food _consumption_ or probability of food delivery
or ...) and 2) the animal is capable of controlling the hypothetical
controlled variable under the conditions (10-s and 20-s VI)
imposed (these feedback functions may be too stringent to allow
good control; testing for a controlled variable when the feedback
function is extreme is like testing a person's ability maintain
his posture in a hurricaine)

I have had two rats on VI 10-s for over 40 sessions now and
performance has yet to stabilize...One animal in particular is
continuing to produce a slow increase its rate of responding
over sessions;

Is rate of food delivery also increasing? If so, this could
be the result of a slowly changing (increasing) reference for
rate of food delivery. If this is true, then there should be
compensation for changes in the feedback function across sessions.
Why not randomly (and gradually) start varying the VI schedule
over sessions -- 0, 2, 5, 7, 10, 5, 2, 8, 15, 10, 5 etc. If
rate of food delivery is controlled relative to a smoothly
increasing reference, then you should continue to see a smooth
increase in food delivery rate over sessions despite these
variations in feedback function. If food delivery rate is not
controlled, however, and the continuous increase in response
rate is aimed at controlling some other variable, then food
delivery rate may continue its increasing trend, but in a way
that varies (up and down) in proportion to the changing VI

I have been holding this rat's body weight constant by adjusting
supplemental feeding;

Why? I thought you were studying the rat's ability to control
its inputs, not your ability to control one of your inputs
(perception of the rat's weight). Why not cut these babies some

Another part of this study will involve training rats on a multiple
VI-VI schedule...This will enable us to examine the dynamics of
the transition from one schedule value to the other and vice versa.

What will this tell us about the controlling that is being done
by the rats?




Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: