Wednesday, 5 May 2010

Reading Group - Gibson (1979) Chapter 5

Chapter 5: The ambient optic array

This chapter is long and dense, so I’m going to go through it in two posts. Essentially, Gibson is going to make the case that light can be structured in a way that specifies things about the environment. The important thing to keep in mind is that the structure arises from relations between things ( in this case, relations between solid visual angles). Traditional optics talks about points of light falling on an object. These points of light change all the time and this creates a real mess for any perceptual systems that perceives points of light. Ecological optics talks about the relationship between solid visual angles. Although the particular conditions of light or perspective might change, the nature of these relationships is preserved. To refresh your memory see Chapter 1, 2, 3, and 4. Now, on to details...

Array means that something has an arrangement. Ambient means that a position is surrounded in an environment that could be occupied by an observer. So, an ambient array, is a structured array that surrounds a point completely. From a potential point of observation (which may or may not contain an observer) the ambient array is available from every direction.

How is ambient light structured? Preliminary considerations

It’s best not to think too much about figure vs. ground. This construal encourages us to think about objects in terms of outlines seen against an empty environment. And, outlines are a pretty useless way to perceive something if your point of observation is always moving (as is ours). The environment is actually quite cluttered and all the myriad surfaces, events, people, and artefacts in it influence light – create structure in light.

A better way to think about the environment (than figure vs. ground) is as a divided spherical field. The upper part consists of the sky and the lower part consists of the ground. This is already informative because the upper part of the environment will tend to be brighter than the lower part. Furthermore, the lower part will tend to be more cluttered than the upper part. Within this lower ground area are all the nested components of the environment described in Chapter 1 (e.g., leaves within trees within forests upon mountains within ranges, etc.). How are these components of the environment related to the components of the optic array? Things in the environment interact with light, creating structure as visual angles in the optic array. In Gibson’s words, “the optic array...[is]... a nested hierarchy of solid angles all having a common apex instead of as a set of rays intersecting at a point” (p. 68). The benefit of angles is that they are unique,whereas rays must be identified by coordinates.

The laws of natural perspective: The intercept angle

The Greeks understood that each object can be conceived as the location of the base of an angle that converges at the eye. Considering the changes in this angle across the extent of the object yields a solid angle that is something like an envelope. This is an object’s outline. If an observer approaches the object, the solid angle grows. If an object is tilted laterally away from an observer, the solid angle is squeezed. These are the laws of natural perspective. That’s fine, as far as it goes, but this provides an oversimplified view of the environment. For instance, Gibson emphasises the environment consisting of illuminated surfaces, rather than only free-standing objects, as the Greeks did. He also emphasises nesting solid angles, rather than groups of them. Most importantly, Gibson emphasises the role of motion in perception, which the Greeks completely neglected.

Optical structure with a moving point of observation

Motion is the normal state of observation. Imagine how little you’d know about your environment if, for your entire life, you observed the world from the same spot. When an observer moves, the optic array changes  - the solid angles within the array change. However, the rigid layout of environmental surfaces does not change. The changes in the optic array tell us about locomotion. The persistence in the environment tells us about the environment, itself.

Perspective structure and invariant structure

Perspective structure refers to changes in the optic array with every bit of motion. Little movements lead to small changes in perspective structure and large movements lead to big changes in perspective structure. So, change in perspective structure is gradual. But, essentially, perspective structure is unique to a particular point of observation. In contrast, invariant structure is common to all points of observation in a given location. It is easiest to detect invariant structure by moving. That is, the invariant structure emerges from the fluctuation in perspective structure. To illustrate, Gibson asks how we can tell that a table top is rectangular since will appear trapezoidal from every angle except one (directly above, and, if you’re being thorough, from directly below as well).  From the majority of perspectives, the  projected shape will be trapezoidal - just take a moment to consider how bizarrely shaped the table pictured above is. But, “the unchanging relations among the four angles and the invariant proportions over the set...uniquely specify a rectangular object” (p. 74).From a static image we might not be confident that a given table top is rectangular (although past experience might tell us that it's a good bet). But, if we were able to walk around the table from the picture, we would know it's shape with certainty.

The significance of changing perspective in the ambient array

Another way to think about locomotion vs. fixed positions is in terms of the meaning of the optical array. Flowing perspective structures mean locomotion and stationary perspective structures mean a fixed position. The differences in the optic array inform us about what’s happening. More specifically, a particular flow in the optic array specifies a particular path of motion – both in terms of the distance covered and in terms of the particular route taken. Importantly, the changing perspective structure implies the invariant structure. Thus, the optic array is specifying “two things at the same time” (p. 76).


There are several crucial points here. 1) The optic array consists of nested solid visual angles 2) These angles change as we move and this tells us about how we’re moving 3) But, the relations between these angles are invariant and this tells us about the surfaces and layout in the environment. Now we can understand a bit
more concretely why the white-out example from Chapter 4 is a problem for perception. The uniform whiteness is the snow means that there are no visual angles to structure the optic array. Thus, we have no visual information about how we’re moving or what the environment is like. I’ll post the rest of the chapter soon. In it, Gibson goes into detail about how various characteristics of the environment are specified in the optic array (e.g. how can we tell that an object continues if it is partially hidden from view?).

No comments:

Post a Comment