Subject: Re: direct/indirect perception From: Al Bregman <bregman(at)HEBB.PSYCH.MCGILL.CA> Date: Sun, 27 Jun 2004 20:27:10 -0400Dear Simon and List, Simon Makin wrote: "> But what would a Gibsonian have to say about phenomena like phonemic restoration? " As I said, I don't agree with Gibson's approach. There are a number of reasons, including the one you gave concerning phoneme restoration. My guess is that Gibson would have to say that phoneme restoration wasn't really a case of raw perception, but rather something else -- perhaps "interpretation". The issue here is whether you can draw a boundary around a set of perceptual phenomena that might not need representation. You wrote: "But what would a Gibsonian have to say about phenomena like ... the McGurk effect?" Gibson rejected all illusions as valid data about perception, sparing himself the necessity of dealing with these problems. Without the concept of "representation", it is difficult to have a concept of "illusion". Without a representation, what is an illusion a mismatch between? The world and some behavior? But you don't have to react overtly to experience an illusion. Furthermore, Gibson ignored the role of synthesis (or composition) in perception. I don't have the space to go into the many issues concerning cmposition here. You can read about it in Bregman, A.S. (1977) Perception and behavior as compositions of ideals. Cognitive Psychology, 9, 250-292. Just to give the flavor, consider the presentation of a picture of a printed capital Z. Suppose it were drawn so that it would be a good capital N if it were rotated by 90 degrees. Now let us present it rotated by 45 degrees clockwise. Is it a Z or an N? Your decision about this is coupled with your decision about the rotation. If it is a Z, then it has been rotated clockwise, but if it's an N it has been rotated counter-clockwise. Conversely, if it has been rotated counterclockwise it is an N. Therefore the act of perception is a composition of two underlying aspects of visual reality(we can call them schemas if we agree that schemas are composable). Another example is a rotated capital Q. What are its physical features? Should it be perceived as "rotated"? Only if it's a Q! Some foreign alphabet could have the form that we call a "rotated Q" as an unrotated letter. The fact is that many of the apparently raw features of visual inputs depend on their interpretation. Take the case of an A overlaid on top of a B. Consider the visible fragments of the B. They are only interpretable as a single coherent form if they are understood as being created by the occusion introduced by an overlaid A. This can only be addressed by a compositional theory of perception (what used to be called analysis by synthesis). By the way, here's an another interesting detail of Gibson's theory: Gibson argued that in addition to the raw qualities of the perceived objects, our brains also extract from the environment some qualities called "affordances". These are aspects of the environment that are linked directly to action. An example is that a small object affords picking it up. For a fly, a vertical surface affords walking on it. For a person, the wall does not yield this affordance; so affordances are species-specific. Gibson does not view these affordances as being the result of inference or memory, but as given directly in the environment as soon as we perceive it. In other words the links to action are direct, not mediated by any other mental process. - Al Bregman ----- Original Message ----- From: "Simon Makin" <S.Makin(at)dcs.shef.ac.uk> To: "Al Bregman" <al.bregman(at)mcgill.ca> Sent: Sunday, June 27, 2004 4:10 AM Subject: Re: direct/indirect perception ... > But what would a Gibsonian have to say about phenomena like > phonemic restoration? Or the McGurk effect? Or the fact that light > levels reflected onto our eyes from a shiny white leather sofa in > a dimly lit room are lower than from a crow seen in bright daylight, > but we still see the sofa as white and the crow as black? > > Regarding Brian's comments - I don't know if "impoverished" is the > best word to use for "information processing" accounts of stimuli. > There is a tremendous amount of information "out there" which our > sensory systems don't even pick up - and a lot of redundancy in the > signals that they do. I think "sometimes necessarily ambiguous" might > be a better term. Is this fair comment? > > Cheers, > > Simon. > > > Al Bregman wrote: > > > > Dear Bruno, Brian, and list: > > > > As I understand Gibson's concept of direct perception, it is more > > than just the claim that the necessary information to form the > > percept is in the stimulus. It also claims that there is no > > mental representation at all of the stimulus. > > > > The argument goes like this: suppose that the way we know the > > world is by forming a representation of it. How would that help? > > How would we know what was in the representation? Would we have > > to form a representation of the representation? This leads to an > > infinite regress. An advocate of representations would argue > > that the brain, having formed the representation, could then read > > features off it as required. Gibson rejects the need for a > > representation at all. He would argue that the world is its own > > best representation. Why would you need another one? The br ain > > can read off features, as it needs them, directly from the world, > > not from a representation. Hence the name "direct perception". > > Furthermore, all the forms of interaction that a representation > > theorist might envision between the rest of the brain and the > > representation, e.g., feedback loops for the control of action, > > can be done directly with the world. We are so directly coupled > > with the world that we can consider the world and our actions > > upon it as a single system for purposes of modeling. So we don't > > need a representation. Of course we know we have representations > > for purposes of thought and imagination, but these processes are > > different from simple perception, which does not use > > representations. > > > > I don't agree with this, but it's the argument for direct > > perception. > > > > Al Bregman > > > > ----- Original Message ----- > > From: "Bruno L. Giordano" <bruno.giordano(at)UNIPD.IT> > > To: <AUDITORY(at)LISTS.MCGILL.CA> > > Sent: Friday, June 25, 2004 6:39 AM > > Subject: Re: direct/indirect perception > > > > > Dear Brian, and list, > > > > > > the ecological approach has the merit of directing attention of > > research to the > > > information in the environment, and to how the adaptive animal > > tends to > > > structure the incoming information in terms of properties of > > the environment. > > > These issues are important to me. > > > > > > However, what would we gain the day we will be able to state > > clearly: [1] > > > perception is direct vs. [2] perception is not direct? > > > Which will be the advantages of such knowledge? > > > Bruno > > > > > > Quoting Brian Gygi <bgygi(at)EBIRE.ORG>: > > > > > > > Julien, > > > > > > > > I'm not the authority on this, but I always thought that > > "direct > > > > perception" is direct in the sense that all the information > > for > > > > perception > > > > is available in the environment, as opposed to more > > > > information-processing-oriented theories of perception which > > posit that > > > > the > > > > stimulus is impoverished and the job of the sensory system is > > improve it > > > > through the use of prior knowledge and inferences. So direct > > perception > > > > does not really require intermediate representations or > > memory models, > > > > although I believe only the most hardcore Gibsonians would > > insist on no > > > > role for memory. > > > > > > > > Brian Gygi > > > > East Bay Institute for Research and Education > > > > Martinez, CA > > > > > > > > > > > > > > > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > > > Bruno L. Giordano - Ph. D. student > > > Dipartimento di Psicologia Generale > > > Via Venezia 8 - 35131 Padova, Italy > > > > > > currently hosted by: > > > > > > Equipe Perception et Cognition Musicales > > > Ircam-CNRS (UMR 9912) > > > 1 place Igor-Stravinsky > > > F-75004 Paris, France > > > > > > ------------------------------------------------- > > > This mail sent through IMP: webmail.unipd.it > > > >