In Chapter 1 we introduced the example of the dog guiding a blind human across the road. The guide dog must perceive its environment geocentrically in order to behave in the manner it does; it must respond to the edge of the pavement, the width of the road, and to the approaching vehicles in much the same way as a sighted human would. Both guide dogs and sighted humans respond to objects in terms of their visual locations and dimensions in three-dimensional space, rather than the projective aspects of these (the locations and dimensions on the retina). That is, objects are normally seen as having constant dimensions despite changes in the retinal projections of their sizes and shapes. This is called perceptual constancy, and it is the prerequisite of more complex perception like recognition.
The subject of perceptual constancies is treated in virtually every book on visual perception, and in most psychology textbooks. However, it is not generally considered as a precursor to object recognition, as we do here. The frequency with which constancy is discussed is unfortunately not correlated with the clarity or accuracy of the discussion. Several of the sources cited for earlier chapters contain good general discussions of constancy, for example, Gregory (1997), Kaufman (1974), and Rock (1984, 1997). Gibson's argument that the problem of constancy is easily understood when perception occurs in normal environments is expressed in most of his writings and those of his followers (see Gibson, 1966; Michaels & Carello, 1981). Also, many of the references for Chapter 4 are appropriate here, especially Epstein (1977).
In the preceding chapters we have described how geocentric perception of the environment is obtained from the patterns of stimulation reaching the eyes. An important feature of this approach has been the emphasis on a moving, active observer. By moving our eyes and our heads we gain information about new aspects of the world that would otherwise not be available. This information has to be extracted from the complex changes in stimulation that are caused by our own movements. As a result we obtain a geocentric representation of the direction, distance, and movement of objects with respect to the environment. Without this, we, and other animals, would be unable to carry out the co-ordinated activities necessary for survival. In