"Where" and "what" in vision

Abstract

The mixture of a few horizontal and vertical line segments embedded in an aggregate of diagonal line segments can be rapidly counted and their positions rapidly determined by a parallel (preattentive) process. However, the discrimination between horizontal and vertical orientation (that is, discrimination of a single conspicuous feature) requires serial search by focal attention. Under recent theories of attention, focal attention has been assumed to be required for the recognition of different combinations of features. According to the findings of this experiment, knowing "what" even a single feature is requires time-consuming search by focal attention. Only knowing "where" a target it is mediated by a parallel process.