Tarr, M. J. & Black, M. J. (1994). A computational and evolutionary perspective on the role of representation in vision. CVGIP: Image Understanding 60(1) 65--73.

@Article{TarrBlack1994,
  author = 	 {Tarr, Michael J. and 
                  Black, Michael J.},
  title = 	 {A computational and evolutionary perspective on the                                                         
role of representation in vision},
  journal = 	 {CVGIP: Image Understanding},
  year = 	 {1994},
  key = 	 {},
  volume = 	 {60},
  number = 	 {1},
  pages = 	 {65--73},
  month = 	 {},
  note = 	 {},
  annote = 	 {}
}

• Human vision cannot be a grab bag of task-specific competencies.

Author of the summary: Jim Davies, 2006, jim@jimdavies.org

Summary:

The purposive approach claims that vision is a set of task-specific competencies. In defense of this approach, they note that computer vision has had difficulty replicating human competence and in doing real tasks, and that it's more in line with an evolutionary perspective. [65]

Without flexible representation structures, knowledge remains compartmentalized and not accessible to other processes. It can't be generalized to other contexts. [66] This is what distinguishes human vision from many animals.

The "recovery paradigm" seeks to build functional descriptions of the visual world in terms of visual and spatial aspects, and then to build symbolic descriptions of them that can be widely used in cognition. In other words, to make a model of the world.

The purposive approach should not be confused with active vision, which is the seeking of information based on higher-level cognition. [67]

Purposive vision is the study of "tasks that organisms possessing viosion can accomplish" (cited from someone else). They are represented in terms of their purposes and roles (similar to Gibsonian affordance theory).

It is an alternate goal for computer vision, one that abandons the dreams of a general-purpose intelligent visual system.

Summary author's notes:

• none