Fw: Re: direct/indirect perception (Al Bregman )


Subject: Fw: Re: direct/indirect perception
From:    Al Bregman  <bregman(at)HEBB.PSYCH.MCGILL.CA>
Date:    Thu, 1 Jul 2004 00:18:15 -0400

Hi Peter, Let me respond to some of your comments (with a copy to the list): You wrote: "As I understand it, Gibson's formulation of direct perception actually came out of his desire to circumvent the 'ghost in the machine' - is that about right? Nevertheless, his idea that scientific reductionism can actually result in impoverished experimental circumstances that exclude the very elements of perception requiring examination, has been useful. The 'embedded-ness' of perception in 'proper' environments might not be best represented in some laboratory environments. Doesn't some of your own work come from a similarly ecological (no capitalised 'E') motivation?" My work came about by accident, but early on, I realized that I was looking at a process that had a function in a natural environment. As time passed, I guided my research increasingly by this idea. I agree with Gibson's idea that it is important to look at the stimulation offered by the stimulus input before concluding that memory and inference are playing a role. I take this merely as good advice without going along with his more radical ideas. You also wrote: "That said, I agree with the synthetic component of perception,though not necessarily going as far as R.L Gregory, who claims up to 90% synthetic." Such numbers are meaningless. My approach to perception through synthesis is this: The world is populated by distinct entities and causes, each of which is going about its business. However, because our senses receive information about all of them at once, our brain develops bets about which entities may be present in the scene, and sees whether a composition of such entities will account for the incoming data. Furthermore the brain "knows about" common transformations in the world, such as rotation, occlusion, deletion of parts, color of light source, reflections, shadowing, changes in the perceiver's viewpoint, and so on. So it tests whether the application of a set of proposed transformations to a set of proposed schemas would yield the data that reaches the sense organs. If it accounts for part of the data well but leaves over some unexplained sense data, some additional hypotheses may be introduced that fit the "shape" of the unexplained data. A process similar to this was called assimilation by Jean Piaget. This is the same strategy that a scientist uses. The data are interpreted as having been shaped by a number of causes in tandem. The data set is viewed as a composite, shaped by many causal factors. The more you know about the other factors that have shaped the data, the more easy it is to see the effects of the particular one you are interested in. Piaget had a name for the human mind's ability to represent the simultaneous influence of two causal factors (each represented by a schema). He called it "one schema assimilating another". I believe that all understanding, whether perceptual or cognitive, works by this process of composing hypotheses about underlying factors to account for the data. That's why I can't be a Gibsonian. The analysis by synthesis approach doesn't deny that simple bottom-up processes are important. Without them, how would one know which underlying schemas should be entered into the composition? Gibsonians (such as Michael Turvey and his colleagues) have had their greatest successes in managing without representation by accounting for sensorimotor loops where the human is controlling some external process. No representation is needed. But I would argue that this is because no "understanding" is involved. You also wrote: "In respect of Bruno's point as to whether anything would be altered by the assertion that perception is direct , the point is that experimental circumstances could never *completely* represent real ones; but surely we know that already?" In reply to Bruno, I would say that if perception were indeed direct, i.e., occurred without representations, then the process of "imagination", which obviously requires a representation, would have nothing to do with perception. Indeed, it would be a complete mystery as to where the representation used by imagination had come from, if not from sensory perception. From extra-sensory perception? Apparently not so, because the parts of the brain that light up in a brain image when somebody is asked to imagine something are intimately related to those that light up when the subject is perceiving the thing itself (sorry; I don't have the reference). I think the Gibsonian view can only be maintained by ignoring the functions of the mind that take place in reference to an object, but not in the immediate presence of the object itself (e.g., imagination, dreaming, judgement, and so on, which we can call "offline phenomena"), Another theoretical position that rejected representation, the Behaviorism of J.B. Watson and B.F. Skinner, was ultimately discarded because of its banishing of the offline phenomena from its account of behavior. Similarly, an account of perception that does not connect it with the offline phenomena may have to be discarded, no matter how successful it is in dealing with a narrow range of phenomena. Best, Al ------------------------------------------------ Albert S. Bregman, Emeritus Professor Dept of Psychology, McGill University 1205 Docteur Penfield Avenue Montreal, QC, Canada H3A 1B1 Office: Phone: +1 (514) 398-6103 Fax: +1 (514) 398-4896 Home phone & Fax: +1 (514) 484-2592 Email: al.bregman(at)mcgill.ca ------------------------------------------------- | |


This message came from the mail archive
http://www.auditory.org/postings/2004/
maintained by:
DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University