"Stimulus Control"

[from Samuel Saunders (950307:1310 EST)]

[From Rick Marken (950305.2000)]

I think control of a logical relationship looks like "stimulus control" to
behaviorists because the actions that preserve the relationship are "part of"
the relationship. The bird, for example, must peck after the red light and
circle after green to maintain the relationship; pecking and circling are
part of the relationship. Suppose, however, that the bird could control this
relationship without its own actions being part of it; for exmaple, the bird
might be able to do something like what the subjects could do in the program
control task; the bird could peck (to restore the relationship) when the
relationship was NOT occurring and stop pecking when it was. If, for example,
the bird saw a red light but no subsequent peck (by a robot bird?), it would
peck a key; if it saw a green light and no subsequent circling (by the
robot bird) it would also peck. Of course, if the bird maintained the
relationship (by pecking ONLY at the appropriate times -- when the
relationship was not happening) it would get its "reinforcement", otherwise
not. (Rather than using robot birds to do the pecking and circling part of
the relationship, it would be logically equivalent to require that a
particular picture occur after the red light and another picture after the

This sounds like a conditional discrimination procedure. A well-studied
example is the "symbolic matching to sample" procedure: Present a vertical
or horizontal line. Follow with a choice of red or green keys. If the
line was vertical, choice of red is reinforced; choice of green is not. If
the line was horizontal, choice of green is reinforce; choice of red is
not. Birds readily learn this task to high accuracy. There are many
variants, but a detailed discussion is probably not warranted at this
point. One variant on this procedure, not widely used, is to follow the
line by a single key illuminated by a hue, and a second white key. Each
peck on the hue key changes the hue, and when the "associated" hue is
present, a "report" on the white key produces the food; an inappropriate
"report" terminates the trial. These procedures are usually discrete
trials, but there are non-discrete versions as well. This is, I think,
closer to what Rick is describing.

By the way, even the red-green discrimination has an element of conditional
discrimination to it, since the mapping is valid in the context of the
operant chamber, etc. Premack once successfully trained birds with red Sd
and Green Sdelta in the morning, and Green Sd and Red Sdelta in the
afternoon (or similar; I don't recall the exact mappings).

This demonstration would at least show that the bird can control a logical
relationship even when it's own reponses are not part of the relationship.
Of course, if the bird cannot control the relationship in this way if
would be evidence that it is not controlling this relationship in the
"stimulus control" situation. That is, it would suggest that the
discriminative stimuli (red and green lights) really are causing the
pecking and circling responses.

A minor point, but few if any EAB types would say the stimuli are causing
the responses, but rather that the stimuli "set the occasion" for
reinforcement of the respective responses.



Bruce Abbott (950306.1505 EST) --

A pigeon has been trained to peck while a vertical line is projected on the
key and not to peck when a horizontal line is projected there. What happens
at the logic level when a line at 45 degrees is projected?

What the pigeon actually DOES (at least initially) is peck at an
intermediate rate.

While this is the majority view in EAB, there is a sizeable minority that
would argue that intermediate stimuli produce a weighted combination of
the behaviors appropriate to the end-points, and averaging over a stimulus
presentation gives the appearance of intermediate responding. Research on
this issue has been inconclusive.


Bill Powers (950306.0930 MST)

As it stands, the SD is not totally necessary; I'm picking up the switch as
much from tracking error as from the color change. I'm still not very good
at it, but it may be worth taking some data and seeing what it looks like.

Is it usually the case that an SD is redundant with some other indication
that the situation has changed?

This is one of the reasons that testing in extinction or with probes is
done when "stimulus control" is an issue. By the way, most EAB types would
consider "noticing the tracking error", the occurrence of a payoff in a
schedule, etc., to be stimuli, and modulation of behavior based on that to
be "stimulus control" as well as behavior modulated by stimuli directly
manipulated by the experimenter.


I liked Rick's idea about reversal with correlated stimuli as a tracking
analog. A variant would be to change the sensitivity of the control handle
(since this can be done for a number of values, while reversal cannot).
Why not just change the background from red to green, rather than changing
the cursor or target? That would be a lot less subtle, I would think.


//Samuel Spence Saunders,Ph.D.