[From Chris Cherpas (951109.1502 PT)]
[re: Bruce Abbott (951109.1655 EST)]
By the way, is is quite easy to resolve this question through computer
simulation, without needing any equipment other than the computer. In fact,
didn't we do this last year? I recall writing a program that implemented a
VI schedule and giving it a random response generator for testing purposes,
but I can't recall whether we tried this with concurrent VI-VI.
If I get around to it, or if you do a simulation of Conc VIs, I
recommend linear VIs (a la Vaughan). That way the obtained overall
reinforcement rate is even surer to not vary from the programmed rate,
since the VIs don't wait for the food pick-up to start timing their
next intervals (they just store them and wait for however many
CRF responses pick 'em up if necessary).
Not that it makes a huge difference, but
if somebody gets that idea that the controlled variable is session/overall
rate of reinforcement, then they might think that matching is explained
by this. With linear VIs, "random responding" should result in the
same overall rft rate as when matching occurs. Also, what I've seen
of "VIs" around this list is really RIs; not that it matters much,
but the Catania/Reynolds or Flescher/Hoffman series are certainly
available.
Also, by the way, and, yes, this is fairly irrelevant to a simulation,
I've gotten pretty good results by using a three-key chamber (pigeons)
in which each side key is a schedule key with a distinctive color
projected on it, and the center key as a changeover key which is lit
with a distinctive color, depending on which side key is lit. When
the CO is pecked, the old side key goes off and the changed-to side
key goes on, AND the CO goes off until the first peck on the new
side key. This seems to result in pretty good separation/discriminability
between alternatives without a COD, per se.
Finally, Bruce, do you have any doubt that random responding on
Concurrent VIs would produce anything other than severe undermatching
(flat or even slight "anti-matching" -- negative slope?).
Regardo,
cc