Expectations formation: Rational or adaptive or...?

Abstract This paper reports on an experimental investigation in which subjects are appropriately motivated to reveal their expectations of two time series. Two hypotheses are investigated: that expectations are formed rationally; that they are formed adaptively. The paper concludes that the ‘truth’ is somewhat in between: subjects are trying to behave rationally, but frequently in a way that appears adaptively. A more general hypothesis is tentatively proposed.