[From Bruce Abbott (960814.0900)]
Bill Powers (960813.1430 MDT) --
Nor did he modify his experimental approach. In the first excerpt above,
the pigeon is actually operating in a basically analog situation,
continuously varying the position of the hoist to keep the target
approaching. But in the Pelican control system design, Skinner went back
to what he was comfortable with -- pigeons pecking. He did this despite
having already found a far more elegant design, much better suited to
the task.
In reading that article, I had the impression that the first approach was
abandoned because the pigeon failed to take corrective action for small
errors, waiting until they became larger and then making large control
actions. This would have been appropriate for a pigeon homing in for a
landing on a window ledge, but the Pelican was too fast and too slow to turn
for this action to work. Perhaps the pigeons could have been trained to
take action sooner, but I suspect that Skinner was forced to abandon the
more natural scheme because of this problem. As to placing little "pots" on
their heads to read out head position, I doubt that this would have worked;
pigeons do a lot of head-turning even while flying about, they don't
necessarily keep the beak fixated on the direction in which they intend to fly.
Of course the military might not
have liked the idea that all it takes to fly a weapon of war is the
brain of a pigeon.
It takes much LESS than the brain of a pigeon! How many "smart bombs" not
only can steer to target, but can find their own fuel, take evasive action
when attacked, perform routine maintenance on themselves, attract a mate,
build a nest, and raise little smart bombs to maturity? (:->
I think it was in the 1960s that I wrote to Skinner telling him about a
control theory model of behavior. His reply was terse -- he didn't
believe that control theory had anything to offer to the experimental
analysis of behavior.
Around 1990 (not long before Skinner's death) a colleague of mine wrote to
Skinner asking him to consider opening up his view to include the
physiological substrates of behavior; Skinner's reply was equally terse and
equally negative. Too bad on both counts.
In mitigation, I think it's clear that Skinner ran afoul (or is that
afowl?) of the "modern control engineers" of his time.
I wonder to what extend plain old bitterness, well-justified, might
account for Skinner's abandonment of the concepts of control theory.
The Pelican account includes a number of references to instances in which
engineers -- and a panel of physicists -- repeatedly rejected the concept on
grounds that it couldn't possibly work; the fact that Skinner had a working
prototype seems to have had little effect on these views. It's an ironic
twist on the problem you've been having, Bill: Skinner couldn't get
physicists and engineers to take seriously the ideas of a psychologist.
Apparently people are people, subject to the same faults, no matter what
their training.
Regards,
Bruce