The present study is largely concerned with the relation between reaction time and the temporal uncertainty of the signal. It is suggested that the appropriate measure of signal uncertainty is the conditional probability that the signal would be presented given that it has not been presented before. This conditional probability is called the expectancy function. It is assumed that the expectancy function influences the subject's state of readiness and this in turn influences reaction time; and ways of expressing these relations mathematically are discussed. It is shown that, with a suitable representation of subjective errors in time and probability estimation, a theory can be stated which accounts for the qualitative features of a variety of experimental data.The relevance of this theory to studies concerned with central intermittency is then discussed. It is argued that temporal uncertainty alone is insufficient to account for the increase in reaction time to the second of two closely spaced signals, both of which require a response. On the other hand, a model for the interaction of two responses is stated in which refractoriness in the central processes is not inevitable.