Timberlake and Killeen

[From Bruce Abbott (951130.1010 EST)]

Chris Cherpas (951129.1921 PT) --

However, I suggest reading some of Timberlake's work
in JEAB:

"Tiny Tim'" (I think he's about 6'6") is no reinforcement theory
apologist. If you're looking for someone who highlights those
Kuhnian chinks, Timberlake is a good source; plus, he proposes
control-theory-esque theories of behavior regulation (mixes in
economics and ethology too); in fact, I don't know of anybody
else who refers directly to PCT/BP within JEAB.

Chris, we've discussed some of Timberlake's work on CSG-L. It has two major
problems: (1) Timberlake's theory is about control of _behavior_, not
perception, and (2) it is functional, not mechanistic. It has some of the
trappings of PCT but not the substance. Behavior is not regulated.

Killeen's article (the one you promised to send Rick) presents Killeen's
argument that theories in EAB should be mechanistic, which seems like a step
in the right direction. However, the mechanisms he proposes are not
physical. For example, one assumtion of his theory is that

A = aR, where A is "arousal level," a is the "specific activation" (defined
         later in the paper), and R is the "rate of incitement."

Neither a nor R is defined in terms of any physical mechanism; rather they
are themselves inferred from behavior. And here is what Killeen has to say
about mechanism:

    As in the case of physics, in behavior analysis the term _mechanics_ is
    something of an atavism; but in both cases, it may be interpreted as an
    emphasis on the analysis of complex resultants into their constituent
    forces, as a focus on causal rather than statistical explanations, and on
    mathematical rather than mechanical linkages between cause and effect. It
    is in those senses, ones common to the behavior-analytic tradition, that
    it is used here. . . .
        Killeen (1995, p. 407)

So when Killeen speaks of mechanism, he means a system of mathematical
relationships delineating how variables influence other variables. Such a
system could be open-loop or closed-loop (good). However, in Killeen's
system the equations (such as the one given above) are disembodied entities
having no identifiable physical substrate (bad).

PCT also offers a series of system equations. For example, perception might
be related to input through the equation below:

p = ki * i

This equation might represent transduction at the sensory interface. One
would look for -- and expect to find -- a physical mechanism that
accomplishes this transduction, converting, say, muscle length, into a
neural current proportional to the length. The same may be said for all the
other equations that define a given control system.

Killeen's "mechanics" specifies a number of ghostly functions whose
spiritual homes in the body remain completely unidentified. Rick is going
to have a field day with this one, if I haven't already spoiled the fun by
giving away the punch line! (:->