Published online by Cambridge University Press: 20 May 2010
Recognition Based on Functionality
Minsky (1991) is one of several well-known researchers who have argued for the necessity of representing knowledge about functionality:
… it is not enough to classify items of information simply in terms of the features or structures of those items themselves. This is because we rarely use a representation in an intentional vacuum, but we always have goals – and two objects may seem similar for one purpose but different for another purpose. Consequently, we must also take into account the functional aspects of what we know, and therefore we must classify things (and ideas) according to what they can be used for, or which goals they can help us achieve. Two armchairs of identical shape may seem equally comfortable as objects for sitting in, but those same chairs may seem very different for other purposes, for example, if they differ much in weight, fragility, cost, or appearance. … In each functional context we need to represent particularly well the heuristic connections between each object's internal features and relationships, and the possible functions of those objects.
The early part of this quote contrasts the approach of representing (only) features or structure of objects with the approach of representing knowledge about how an object functions to achieve a goal. Particularly in computer vision, objects have traditionally been represented by their shape or their appearance. Object recognition based on reasoning about functionality stands in contrast to these more traditional approaches, with the aim of achieving recognition at a more generic level.