Matching bias in conditional reasoning consists of a tendency to selec
t as relevant cases whose lexical content marches thai referred to in
the conditional statement, regardless of the presence of negatives. Ev
ans (1983) demonstrated that use of explicit rather than implicit nega
tive eases markedly reduced the matching bias effect on the conditiona
l truth table task. In apparent contrast, recent studies of explicit n
egation on the Wason selection task have failed to iind evidence of lo
gical facilitation. Experiment I of the present study strongly replica
ted the Evans (1983) findings and extended them to three forms of cond
itional statement. Experiments 2 and 3 showed further that the use of
explicit negatives removed completely the matching bias effect on the
Wason selection task. However, consistent with other recent studies, t
his elimination of bias did not lead to facilitation of correct respon
ding. The findings are interpreted as providing evidence chat matching
bias reflects a linguistically cued relevance effect. (C) 1996 Academ
ic Press, Inc.