This paper considers certain experimental tasks used by cognitive psyc
hologists, performance on which has been thought to show that intellig
ent and able human subjects are 'irrational'. It is argued that the re
sponses judged to be 'correct' (which the subjects usually fail to giv
e) are so only in a pedantic sense. They are the responses that would
follow if thinking was in accord with some simple abstract model and i
f it could legitimately be claimed that following the model constitute
d rationality. The models chosen are shown to be inappropriate for the
purpose. Furthermore, no account is taken of the subjects' desire gen
uinely to participate in the experiment-to make sense of the task give
n and respond according to that sense. Subjects do not without specifi
c instruction try to dissect out an abstract model and respond accordi
ng to that. It is concluded that the responses actually given are prim
a facie evidence of rationality in the subjects.