[Martin Taylor 970306 15:05]
Tracy Harms, apparently Thu, 6 Mar 1997 11:58:24 -0700
Notes on JJ Gibson's approach to
the haptic system,based upon:
The Senses Considered as Perceptual Systems, Chapts VI -
VII, Houghton Mifflin Co., Boston (1966).
Observations on Active Touch, Psychological Review, 69,
477-491 (1962)Emphasis on pickup of information by *perceptual* systems rather
than by isolated coding of sensory data. Molar vs. molecular
approach. The identification of invariants. Argues that sensation is
not a prerequisite of perception ". . . sense impressions are not the
"raw data" of perception". He almost seems to deny that neural
transduction processes from receptor to brain are important (p. 48)
(see Handwerker (1984) for evidence of opposite view, and also
Iggo (1982)). Stresses that perception is an *active* process of
seeking out information about the environment.
***********************************************************************This stress on active process sounds right, eh folks?
At least, so far as it goes.Hoping to elicit reaction to Gibson from somebody who has studied his work
with PCT in mind. Any takers?
We didn't do this with PCT in mind, but if you check out our paper, you
might think we had, since we use a three-level control hierarchy to account
for the perception of texture:
Taylor, Lederman, and Gibson "Tactual perception of texture", in "Handbook
of Perception. III. Biology of Perceptual Systems" (Eds) Carterette and
Friedman, Academic Press, 1974.
Our three levels were called "Behaviour Control," "Movement Control," and
"Motor Control." If I may quote a couple of paragraphs:
------------------start quote-----------
At least three basic behavioural feedback loops are probably important
(Fig 2). The major one is an overall control loop (i) whose function is
to carry out the policy decision to look for a certain feature of the
texture, such as roughness or elasticity. The original intention to
touch something, to investigate an aspect of texture such as roughness,
is taken to be a command to the major loop control element, labelled
Behaviour Control. The function of this module is to select a touching
strategy adequate for the job...
The secondary loop (ii) has the function of executing the individual
motions required to implement the desired touching mode. Its commands
are produced by the element labeled Movement Control. This element
breaks down the general command from the Behaviour Control into a
sequence of specific motion commands, which go to the control element of
the innermost loop (iii) labeled Motor Control. This is the familiar
kinesthetic control loop, which breaks down the individual movement
commands into muscle commands and monitors the effects to ensure that
the movement command is properly executed.
...
All the sensory inputs from the transducer...are available to the large and
largely unknown module called Texture Analyzer. This module also has as
input the information from the Behaviour Control module concernint the
intent of the motion....
At its output, the Texture analyzer provides a "texture evaluation" (for
lack of a better term). If this evaluation satisfies the requirements of
the commadn to the Behaviour Control module, then the touching process
has been completed. But if the information so far gained is insufficient
to satisfy the command, then the movement strategy must be continued or
modified. Hence the informational link between the Texture Analyzer and
the Behaviour Control element completes the major strategic feedback loop.
---------end of quote----------
What is lacking from here is the _explicit_ recognition that the control
is _perceptual_ control, though it is implicit in a couple of places. This
was written before BCP was published, and perhaps if I had seen BCP when
it was being written, it would have been a better chapter. Anyway, at that
time it seems as if I was almost onto PCT, but later I must have veered
away into other things. Pity it took me another 2 decades to come back.
Incidentally, our work was based not so much on Gibson, though he definitely
had an influence, and more on David Katz (1925) Der Aufbau der Tastwelt.
(Notice the reference to "Building" in the title). What Gibson pointed
out in his Psych Review paper was that the same sequence of tactile
sensations produced a quite different result if the hand was being
touched ("I feel I am being touched") as compared to if the hand was
doing the touching ("I feel an object out there"). What we showed was
related. For example, we tried to compare open-loop studies with closed-
loop studies of the same thing. We got people to judge the roughness of
a scribed aluminum plate when they had control over how hard to press and
when we put the plate on a balance so that we could be sure they were
pressing with N grams of force. I forget the results, but I suppose I
could look them up if pressed to do so.
Martin