Operator Noise in a Discrete Signal Detection Task

Man's ability to detect visual signals in noise is the concern of this paper. An operator is presented a computer generated two dimensional binary—or ``dot''—display and is asked to indicate the presence or absence of a ``signal.'' Previously developed signal detectability theory is expanded. A model of the operator as a threshold detector hampered by a Gaussian noise source is developed. The noise source is defined by two parameters—first and second moment operator factors, ε and ϒ. The most important parameter, ϒ, is investigated experimentally and found to be essentially independent of signal to noise ratio, SNR. This is interpreted to mean that the noise source decrements the actual SNR by 2 to 3 dB and that the operator sets a near optimum decision threshold as a function of SNR.