Task dependency and heuristics

From A conversation about the brain
Jump to: navigation, search

The idea of task dependency is very common in vision research and does not need rehearsing here in detail. A clear example is the observation that people are very accurate at matching the depth relief of two similar-sized objects (same height) and yet they make substantial errors in a depth-to-height judgement for the same surfaces. Taken together, these data rule out the possibility that observers are using the same internal 3D model of the scene to carry out both tasks: performance is task-dependent[1]. The depth matching can be done accurately only because the visual system is able to use a trick, or 'heuristic', that is unavailable for the depth-to-height judgement.

Gibson [2]and his followers have emphasised the use of heuristics by the visual system but this does not avoid the need for representation altogether, as Gibson sometimes seemed to imply. It is still necessary to navigate, using head and eye movements, to a relevant part of the optic array (or a distant optic array) to find the relevant piece of visual information to control the task at hand. That cannot be done without some store of instructions, i.e. a representation. For sure, this does not need to be 3D, but something needs to be stored. Hence the idea of a 'base representation' that has enough information in it to get the eye from the current image to the image containing the next relevant piece of visual information. The pages in this section sketch out some of the requirements for such a representation.

References

  1. Glennerster, A., Rogers, B. J., & Bradshaw, M. F. (1996). Stereoscopic depth constancy depends on the subject's task. Vision Research, 36(21), 3441-3456.
  2. Gibson, J.J. (1979) The ecological approach to visual perception. Boston: Houghton Mifflin