Although we have talked about distributions of signal and noise nothing has been said

about the shapes they might assume. The sensitivity measures described in Chapter 1 were

non-parametric and made no assumptions about the underlying distributions. There is, of

course, no a priori reason that these distributions should be of any particular kind. Their

shapes may vary capriciously from experiment to experiment. However if it did turn out that

sensory events were distributed in the same way, this would allow very efficient sensitivi-

ty measures to be used, as will soon be seen. Most psychological variables seem to be at

least approximately normally distributed and the first question that can be asked about the

signal and noise distributions is, do they conform to a normal, or Gaussian, distribution?