In the previous chapter, Gibson laid the foundations for an ecological optics by distinguishing between stimulation and stimulus information. In this chapter, he lays out the details of how to analyse the source of visual stimulus information, as an ambient optic array that contains both invariant and perspective (variant) structure.
Sabrina blogged this chapter here, here, and here; there is a lot going on in this chapter!
Gibson first defines what he means by an ambient optic array. An array has structure; it is not homogeneous. Optic implies it is made of light. Ambient means completely surround a point of observation (which may be occupied or not). A point of observation is really a place in an environment, not an abstract point. When occupied, the point of observation includes information about the environment, but also the perceiver.
So what sort of structure is in light? That depends on how we describe the environment, which we did in Chapters 1, 2 and 3. An environment is not simply objects in space; it is 'places, surfaces, layouts, motions, events, animals, people, and artifacts' (pg 59). Environments consist of a layout these things nested in amongst each other, and so should the optic array.
The best way to describe the optical structure is then as a nested hierarchy of solid visual angles. The base of the angles have a shape that depends on the surface that creates them, and their common apex is at the point of observation (geometrically, the centre of a sphere). The solid visual angles overlap, and sit on top of each other, and completely and densely fill the space. There are angles from the faces of objects, filled with smaller angles from the facets (texture) of the faces; the exact boundary is, as usual, negotiable.
This, however, is only part of the optical structure, because it describes the array from a stationary point of observation. Locomotion is, in part, the act of going from one point to another, different point along a trajectory of other points. As you locomote, the array changes; but not all of it changes, and it does not change in a jump (as from image to image). Some of it changes entirely with any change in point of observation (the perspective structure; every point of observation is unique) and some of it remains the same because it is about the persisting surface layout (the invariant structure). The details of the perspective flow specifies the details of your locomotion. As you locomote, then, the ambient optic array flows in time from one view to another and another, along your path. This flow is reversible, and it immediately reveals what is variant and what is invariant. For a specific perceptual question (e.g. how do I perceive a table is a rectangle from any viewpoint?) the scientist must examine the flow and find what persists because of the surface layout (in this case, the relations between the corner angles). Finally, the perspective structure and the invariant structure co-occur; they are both always present and defining each other.
The Significance of Changing Perspective - The Case of Occlusion
The general case for examining what the optic array can specify is when the perspective can change; if the observer can locomote (change in the environment, events, are the next chapter). Locomotion produces flow, which reveals variant and invariant structure as the ambient optic array smoothly transforms. The former is about the locomotion, the latter is about the layout of surfaces in the environment.
Environments are cluttered with opaque surfaces - from a given point of observation, some things are in front of other things and those rear surfaces are not projected into the optic array at that point of observation. They are occluded (see these posts, and these demos). And yet we perceive that the rear surface continues to exist, and that it has merely gone out of view. You cannot see something that is unseen; so there must be information about the persistence of the rear surface in the optic array.
Occlusion (moving in and out of view as the point of observation changes) is different from disappearing because the object ceased to exist, or if the light goes out, or if it gets too far away. Occlusion by a surface can be of itself (in which case it is revealed by perspective, continuous changes as it moves; the near side comes into/out of view) or of something else (in which case it is revealed by a discontinuity, with solid angles from the rear surface going from being projected to unprojected to the array at a point of observation). That discontinuity is a kinetic disruption, a violation of the usual continuity. This discontinuity sweeps across the solid angles from the rear surface; at the occluding edge those angles go from projected to unprojected, and at the rear edge they do the reverse. To the extent the angles being occluded and revealed have the same structure, a continuous rear surface is specified.
This case is one of the key demonstrations that visual experience is based on information, not stimulation. The changes to the ambient optic array specify what is happening, and people perceive what is specified.
How is Ambient Light Structured?
The final section reviews some of the ways in which the ambient optic array comes to include variant and invariant structure. Note the array will always include both; they are, like many things in ecological psychology, a mutually defining pair.
Invariant structure comes from the persisting layout of surfaces; they persist, therefore the structure they cause persists. This simple point is a straight up refutation of the notion of sensations. Variant structure comes from motion; locomotion of the observer, or motion of the illumination source. Again, though, these changes are specific to the motion causing the change.
Gibson then discusses some of the consequences of a moving source of illumination. Surfaces directly in the light are brightly lit; but light reverberates through the environment and so surfaces out of the direct line to the source of illumination can also be illuminated; just less brightly. As the illumination source moves (e.g. as the sun passes overhead) there are regular changes in how surfaces are illuminated, and this regularity is all part of the perception of an environment. Within that change there are invariants due to layout, and regularities in the relative reflectance we experience as the persisting colour of surfaces (colour is a relational property across surfaces, not an absolute property of a given surface). Finally, the special case of water; the surface layout ebbs and flows quickly and so does the pattern of who is directly vs indirectly illuminated; but this is what identifies the surface as water!
Summary
Gibson has introduced all the key elements of ecological optics. First, the basis of vision is not sensations, but the ambient optic array; an indefinitely nested hierarchy of solid visual angles specific to the surface layout and flow variables specific to the motion of the observer or the source of illumination. This is what it means to say that vision begins in the environment, before the light hits the eye: optical structure is revealed, not computed. We perceive what the information specifies (e.g. occlusion, going out of view and not ceasing to exist).
Posts
No comments:
Post a Comment