[From Rick Marken (2003.12.03.1030)]
"Williams, William D." wrote:
In a seminar scheduled for this Friday one of the texts is Susan Haack's 1993 _Evidence and Inquiry_. Haack is a contemporary pragmatist, and basicly I like what she says. I'm not an especially critical sort. But, she does mention in support of her position the work of the psychologist J.J. Gibson. And, she says, "Gibson's theory of perception is a thoroughly evolutionary conception of the aptness of organisms' perceptual systems for coping with their enviornmental nitche, ... " I seem to recall that some CSG people have thought highly of Gibson. I'd be interested in hearing characterizations regarding the merits, or not, of his work.
I think what is (or was) attractive about Gibson's theory of perception (to those of us who are also attracted to PCT) is that is takes into account the fact that what an organism _does_ influences what it perceives. In other words, Gibson recognized what in PCT we call the feedback function connecting output to input. Of course, Gibson never understood that perceptions are not only influenced but are controlled relative to internal specifications. But some of his students -- like David Lee -- did research that looks a lot like the study of control of perception. Lee was able to knock toddlers over without touching them by simply moving
the walls of the room in which they were standing suddenly forward or backward. This works, of course, because the toddlers are controlling for visual alignment with the edges of the walls.
Gibson's theory of perception is based on the idea that the sensory systems were "tuned" to detect _invariants_ in the changing sensory array. Invariants are those aspects of sensory input that remain the same despite transformations that result from movements of the perceiving system itself. For example, when an animal moves toward an object there is an optical flow out from the point to which the organism is moving. This flow is invariant (as long as the animal is moving toward a fixed point) although the sensory components of the flow are changing. So Gibson concluded that organisms navigate to objects by "detecting" an invariant --
the direction of the optic flow.
It all sounds pretty good until you start trying to build models based on Gibson's ideas. When you do, you find that Gibson was an S-R theorist with stimuli (S) being "invariants" that guide the actions (R) of the organism. Things get even worse when you get to the idea of _affordances_, which are perceptual properties of objects (super-invariants, I suppose) that tell the organism what to do with the object. For example, chairs have perceptual invariants that "afford" sitting. The idea of an affordance reveals clearly that Gibson's theory of perception is based on naive realism -- the assumption that we know what's really "out there"
(chairs, tables, people, words) and that the job of a perceptual theory is to explain how what's out there gets mapped into what's "in here" (perceptions of chairs, tables, people, words). This approach to perception is somewhat different than that taken in PCT, which doesn't assume that perception is necessarily a map of what's really out there. PCT assumes that perceptions of chairs, tables, people, words are a _construction_, not necessarily a map (a concept illustrated by Bill's "multi-system reorganization" model where hundreds of control systems successfully control different, randomly selected aspects of the same environmental
reality). In PCT, perception depends on environmental variables (reality); it just don't "map" that reality into mental correlates.
Like Skinner, Gibson got very close to the concept of control of perception, but no cigar (he was a chain smoker of cigarettes, anyway).
Best regards
Rick
ยทยทยท
---
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org