[From Bruce Abbott (971224.1610 EST]
Dag Forssell (971224.1220) --
Dag, you can call "mush" anything you wish to. I don't care. But I think
that you have confused the observations with theories that may or may not
have been proposed to explain them. None of the phenomena on my list is
either a theory or an explanation, and none of them were offered as "laws."
To use your astronomical analogy, you are saying that the observations on
which both the Ptolemeic and Copernican explanations rested -- the observed
motions of the planets against the backdrop of stars as viewed from Earth --
are mush. Water, when chilled sufficiently, freezes. Mush. Thunder
follows lightning with a delay that depends on distance. Mush. People die
if deprived of oxygen. Bunch of nonsense. Mush, mush, all mush. Right, Dag?
When you created a computer "model" of reinforcement theory, you took
liberties by representing dnut and probabilities as physical entities. You
did not require that the quantities you defined and made operate through
logical manipulation be physically feasible. Physical feasibility appears
to not be a requirement in your personal concept of psychological "science."
This absence of stringent criteria allows you to imagine that anything that
appears to be observed is for real, and that to describe it is to make a
major contribution to mankind and the progress of science.
It was not a computer "model" of reinforcement theory. You keep forgetting,
Dag, that this exercise was not intended as an attempt to actually explain
e-coli behavior using reinforcement principles. In fact, I explicitly
denied that this was to be the case, noting that real e-coli do not learn to
behave as they do. The challenge was to demonstrate that current
reinforcement principles could be used to produce the requisite behavior
under the test conditions specified by the Marken/Powers paper. It was
stated categorically that these principles _could not produce_ the requisite
behavior, but I proved that they could. Of course, I had to make some
fanciful assumptions to do so, as the real e-coli do not learn, and
reinforcement theory is above all a learning theory.
By the way, dnut is the change in nutrient level owing to e-coli's motion
across the nutrient gradient. The same variable appears in the control
model, where it fails to evoke your scorn. The probabilities represented
changes in behavior mediated by physical mechanisms not described in the
model itself. Certainly the physical mechanisms themselves would be greatly
preferred over this representation -- on that I think we can agree.
Now, can we get back to talking about PCT?
Regards,
Bruce