Abstract
Two experiments were carried out to investigate human prediction behavior of future states of graphically displayed discrete time series. In Experiment 1, 14 subjects showed the tendency to predict the next point of a time series on the basis of the autoregressive information despite differences in periodicity of the time series, while they showed another tendency to predict the more distant future point of the time series on the basis of the periodicity information, if the time series was periodic. In Experiment 2, 10 subjects revealed that the effects of displaying the future states on human predictions depended on the periodicity of the time series. These results were discussed in terms of autoregressive model of human prediction behavior.
