We use experimental data to investigate what subjects have learned after pa
rticipating in an economic experiment for many periods. At the end of our e
xperiments subjects are given another experiment to perform which functions
like a surprise quiz since they had not been informed about its existence
before. The: results of this surprise-quiz round, along with the time-serie
s of their responses before the quiz, allow us to judge what they have lear
ned. We find that depending on the manner in which subjects are paid, they
attempt to learn different aspects of the experimental task and perform dif
ferently. (C) 1999 Academic Press.