Cyclic ratio-random thoughts

[Samuel Saunders (950803:1430 EDT)]

I will be out of town and out of touch for the next week, and probably
another week to catch up when I get back, so I will delay comments that may
lead to lots of discussion. I did want to make a couple of comments on the
state of the ratio schedule discussion.

The history:

Some effort was being made to understand the Motherall data, and develop a
PCT model for the ratio schedule situation. There was considerable
discussion of the bitonicity in response rate - reinforcement rate
function, and some question about why this had not raised a flag in the EAB
community. I quoted the party line on interval schedules, and suggested
that monotonicity in interval schedules gave EAB an out. There was then
consideration of bitonicity in interval schedules, and why it was not seen.
I then jumped the gun on Bruce, and posted some data from Campbell and
Dougan which showed bitonicity in interval schedules. The Campbell and
Dougan data, and some data from McSweeney et al. which Bruce posted,
suggested that there were within session changes which looked like changes
in deprivation level (or reference). This suggested that much of the
response rate versus schedule data might have an important confound, since
the change within a session would likely be greater with a richer schedule.
Bruce then suggested the Ettinger and Staddon cyclic ratio data as a way
around the problem, since all schedules would occur within a session with
roughly equivalent "pre-loading." Bruce then made the big discovery about
the constancy of response rate in cyclic ratio, suggesting that much of the
schedule of reinforcement literature may be irrelevant to the study of
control.

Given this path, it seems wise to look at traditional blocked ratio studies
to see if they fit with the analysis coming out of the cyclic ratio data.
I posted some concerns previously about the cyclic ratio schedule. Bill
Powers promised comments when modelling was more complete, but Bruce has
preempted that line of work for the moment. Briefly, my concern was this:

There are two ways an animal in a typical ratio schedule study could adjust
response rate to control changes in reinforcement rate with changes in
schedule. Within a run, as time since the last pellet increases, the error
in "reinforcement rate" would increase, and response rate could be
increased. This suggests that responding within a run should show
acceleration. There is very little evidence for this in cumulative records
from fixed ratio responding. The alternative is that over successive runs,
the animal increases response rate, and reinforcement rate increases as
well. It is easy to see how this come about with blocked ratios, since the
feedback function is the same from run to run, and changes in response rate
are directly reflected in changes in reinforcement rate. With cyclic
ratio, the feedback function is constantly changing. The simple
relationship between rate of responding and rate of reinforcement is
disguised by changes in the feedback function, and it is more difficult for
the animal to "come into contact with the contingencies" (as EAB would have
it). Perhaps what is needed is a comparison for blocked ratio conditions,
at the same number of reinforcers into a session, for different ratios.
Now all we have to do is find someone who has the data in a format for that
comparison :slight_smile:

On the question of minimal collection time- there have been studies with
monkeys using fruit juice squirted directly into the mouth as reinforcer.
That should produce minimal collection time, if anything does. I will try
to dig up some references when I get back next week. Another place to look
might be the reinforcing electrical brain stimulation literature, although
that has the potential of raising a whole new set of problems.

Among EAB workers with a "behavioral economics" orientation, there has been
some work on differences between "open economies" (the usual situation,
where the "commodity" is available both for work in the chamber, and in a
different location at other times (usually the home cage of course)) and
"closed economies" (the entire supply of the "commodity" must be earned
in the measured situation). This literature may also have some information
of interest to the current discussion, and I will look into it when I get
back as well.

//----------------------------------------------------------------------------
//Samuel Spence Saunders,Ph.D.