Let's get experimental: Humans on schedules

[From Samuel Saunders (951208:16:22:10 EST)]

Rick Marken (951208.0830)

Samuel Saunders (951207:21:59:36) --
I ran into something which may pose serious problems. I was aware that
human FI performance tends to be different from that of most non-human
subjects. Apparently that is true of most schedules.

Does this mean that reinforcement doesn't apply to humans? Or that it can't
be tested with humans? What's going on here?

WARNING-CONTROL USED IN EAB RATHER THAN PCT SENSE IN WHAT FOLLOWS

Rick, I thought the object was to design an experiment that would be
accepted by EAB devotees. The proposal was to use normal adult human
subjects. Although it may seem odd to you, EAB data on normal adult humans
are relatively rare. Most EAB work, particularly basic work such as that
on basic schedules, is done on non-human animals. Much of the work on
humans has been done with subjects with various handicaps, in association
with therapeutic or assistive programs. I, like most people with EAB
background, am not very familiar with work with normal adult humans. I had
a vague recollection that there were problems with schedule performance of
adult humans, but that was all. I thought that a quick literature review
would help to insure that our proposed experiment could avoid what to the
segment of the EAB community familiar with work using adult human subjects
would appear as obvious errors. I posted information on the problem I
found.

I am still looking into the literature. As a quick summary, the EAB view
is that there are two kinds of control important for adult humans:
contingency based control and rule based control. In non-human animals on
basic schedules, contingency based control alone is assumed to be found.
The notion is that, if rule based control can be avoided in adult humans,
behavior typical of contingency based control would be found. There are
some results that suggest that, if the contingencies are carefully
disguised from the subjects, it may be possible to obtain contingency based
control in adult humans. The consensus appears to be that it is very
difficult to arrange an experiment to avoid rule-governed behavior. I
suspect that it would be nearly impossible to devise an experiment in which
we served as subjects that would be expected by an EAB audience to work,
although it may be that we could devise one using our friends.

You have been pushing for the reinforcement model to go with the
experiment. If you read my last posting carefully, you will see that I
mentioned that models which have anything approaching the specificity
required to go head to head with the PCT model are designed for use with
non-human subjects, which to EAB 'theory' implies contingency-based
behavior. Just as Bill Powers has noted that using a CONC schedule would
require PCT to add new models to account for choice, producing a
reinforcement model for use with adult human subjects on basic schedules
may require adding to existing models to include rule-governed behavior.
Making such an addition, without posing a straw man, may prove difficult.

As to instructions, that is one of the variables that have been cited as
influencing the development of rule-governed behavior by adult human
subjects. For an EAB audience, the detailed instructions, and details of
how and when they are presented, will likely be considered of substantial
importance.

I would also like to continue the discussion of the wisdom of posing our
studies as a direct contest between PCT and 'reinforcement theory', but to
avoid diluting the intent of this post, I will defer to a later posting.

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