According to the interactive activation framework proposed by McClelland an
d Rumelhart (1981), activation spreads both forward and backward between so
me levels of representation during visual word recognition. An important bo
undary condition, however, is that the spread of activation from lower to h
igher levels can be prevented (e.g., explicit letter processing during prim
e processing eliminates the well-documented semantic priming effect). Can t
he spread of activation from higher to lower levels also be prevented? This
question was addressed with a choice task procedure in which subjects read
a prime word and then responded to a target, performing either lexical dec
ision or letter search depending on the color of the target. A semantic con
text effect was observed in lexical decision, providing evidence of semanti
c-level activation. In contrast, there was no semantic context effect in th
e letter search task, despite evidence of lexical involvement: Words were s
earched faster than nonwords. Further evidence of lexical involvement in th
e letter search task appeared in Experiment 2 in the form of greater identi
ty griming for words than for nonwords. The results of these experiments ar
e consistent with the conclusion that feedback from the semantic level to t
he lexical level can be blocked. Hence, between-level activation blocks can
be instantiated in both bottom-up and top-down directions.