Comparison questions in physiological detection of deception were stud
ied with 60 ''guilty'' and 60 ''innocent'' participants in a mock crim
e experiment. Different types of comparison questions were used in fou
r conditions: relevant-irrelevant (R-I) participants answered only rel
evant and neutral questions; trivial directed lie participants were in
structed to lie to three of the six neutral questions; personal direct
ed lie participants were instructed to lie to personally relevant ques
tions; and probable lie participants received traditional probable lie
comparison questions. Respiration, cardiovascular, vasomotor, and ele
ctrodermal activity were recorded. Manipulation of the comparison ques
tions produced different patterns of physiological responses for innoc
ent but not for guilty participants. The R-I test produced an unaccept
able rate of false positive decisions.