Maximum-likelihood estimation of the accuracy of auditory discrimination in the absence of an overt response to every stimulus

In research on human performance in which techniques for estimating hits and false-alarm rates are used in conjunction with other measurements, such as evoked potentials, it is necessary to devise special data-collection methods. This paper describes such data-collection methods, together with a statistical method for summarizing the data. The statistical method consists of setting up a probability model whose parameters describe the process and then estimating the parameters using maximum likelihood.