Abstract: | Feature-integration theory is outlined to clarify its claims and assumptions. The concepts of separated vs. integrated and of latent vs. manifest features are then examined. It is argued that quadratic polynomial search functions would be predicted only if the feature and the location were both latent—an unlikely assumption, against which there is considerable evidence from the efficiency of spatial cues to attention. An experiment on iconic memory suggests that attention can address features through their locations both for a physically present stimulus and for briefly stored sensory information. Finally, a comparison of D. Navon's (see record 1990-27540-001) preferred model to feature-integration theory seems to reveal no substantive differences. (PsycINFO Database Record (c) 2010 APA, all rights reserved) |