Conrad's (1951) data on the performance of a human operator who has to respond to a complex temporal sequence of stimuli (obtained by a parallel arrangement of two, three or four signal sources and each delivering periodic signals) are compared with the properties of three kinds of signal processing systems, each constituting special cases of a general concept. In model I a simple dead time is assumed, while in models II and III the dead time after each response is a linear function of the preceding inter‐response interval duration.It is argued that the mean number of errors per unit of time, the criterion used so far, is a somewhat crude measure and that more decisive conclusions can be drawn when a more detailed description of the timing of the responses is available; for instance, in the form of the probability density function, or the joint probability density function of the response interval durations. The models may also find application in neuronal spike train analysis, and for problems in the field of operations researc