Wednesday, February 23, 2005

Picture Perception Framework

Hi CS24 folks. I thought that it would be useful to provide a framework with which to better understand the theories of picture perception we have been discussing in class. At this point important to realize that there is not one synthesized view. Instead, there are 4 main schools of thought when it comes to thinking about how we perceive 2D visual stimuli.

1. The Symbolic View, which was spearheaded by Goodman. This view purports that pictures are arbitrary symbols of depicted scene--a code that must be decoded by the viewer. The idea is that arbitrary pictorial conventions must be learned, just like learning a language or learning how to read. While this is good for explaining images filled with metaphors, it looses ground with experiments that show the universal interpretations of pictures (like the anthropological studies of African tribes that have never seen pictures before, but can easily understand pictorial representation with familiar objects).

2. The Constructivist View, which is a very popular view lead by academic figures such as Gombrich, and the writer of Visual Intelligence, Donald Hoffman. This view suggests that a picture is an ambiguous index of a depicted scene and that pictorial cues or clues that are correlated but not reliable (for example, breaking into a sweat is an index of fever, but it does not necessarily mean that we have a fever). We as viewers must infer or construct an interpretation of the scene, based upon prior knowledge. The idea is that we go “beyond the information given” which is consistent with the “Inference Perspective” (proposed by Rock) and current Bayesian models of vision.

3. The Ecological View, is the viewpoint proposed by Gibson. This view suggests that pictures are icons of depicted scene--that is, 2D picture are not “similar” to 3D scene, but Light from the picture “shares characteristics” with light from real scene. Essentially, we are able to recognize the contents of pictures because they capture aspects of information about true 3D scene, which are not arbitrary or ambiguous (e.g. linear perspective was not chosen arbitrarily but actually captures the effects of viewing depth in real 3D space). This viewpoint is consistent with the “Stimulus Perspective” where the viewer detects information within pictures, without prior knowledge or inference, it is based on how we evolved as humans.

4. The Gestalt View, best represented by the work of Arnheim. This view suggests that picture elements are organized into 2D or 3D figures according to innate principles such as simplicity, good form, Pragnänz, Figure/ground, grouping, continuation (the ideas we discussed in class). The problem of this view is that there is no definition of “simpler” and many of the boundaries between the innate principles are hard to parse apart. I should mention though that the research of Gestalt theorists have huge implications for composition in abstract art, and many artists, such as Klee were inspired by the ideas of Gestalt Theorists.

Hope this explanation is useful. If you have questions I will do my best to answer.


Blogger ksl said...

Thanks, Tiffany. What a comprehensive & concise reference. I'm sure to use it when sorting out my CG ideas.

February 23, 2005 at 2:42 PM  

Post a Comment

<< Home