Three experiments examine whether spatial attention and visual word recogni
tion processes operate independently or interactively in a spatially cued l
exical-decision task. Participants responded to target strings that had bee
n preceded first by a prime word at fixation and then by an abrupt onset cu
e either above or below fixation. Targets appeared either in the cued (i.e.
, valid) or uncued (i.e., invalid) location. The proportion of validly cued
trials and the proportion of semantically related prime-target pairs were
manipulated independently. It is concluded that spatial attention and visua
l word recognition processes are best, seen as interactive. Spatial attenti
on affects word recognition in 2 distinct ways: (a) it affects the uptake o
f orthographic information, possibly acting as "glue" to hold Letters in th
eir proper places in words, and (b) it (partly) determines whether or nor a
ctivation from the semantic level feeds down to the lexical level during wo
rd recognition.