A theory of voluntary, top-down control of visual spatial attention is
presented that explains how linguistic cues like ''above,'' ''below,'
' ''left,'' and ''right'' are used to direct attention from one object
to another. The theory distinguishes between perceptual and conceptua
l representations of space and views attention as a set of mechanisms
that establish correspondences between the representations. Spatial re
ference frames play an important part in this analysis. The theory int
erprets reference frames as mechanisms of attention, similar to spatia
l indices but with more computational power. The theory was tested in
11 experiments that assessed the importance of linguistic distinctions
between classes of spatial relations (basic, deictic, and intrinsic)
and examined the flexibility with which subjects manipulated spatial r
eference frames. (C) 1995 Academic Press, Inc.