Figure 2.
RT to reject a word in the conjunction task as a function priority give to lexical or sentence semantics. Only data from “no” responses are presented. The cross-over interaction indicates that the instruction, and training, to give priority to a given element was effective in changing performance.