ABSTRACT
Consider the general reaction-time paradigm involving n potential stimuli with a probability distribution P = (p7,..., p n) and with each stimulus requiring a specific response. That is, Stimulus i occurs with probability p t on any given trial, with i = l,...,w and/?i+...+pn = and the subject responds as quickly as possible. In the case of error-free performance, or nearly so, the classic relationship known as HickHyman's law, after Hick (1952) and Hyman (1953), is given by
where RT is the overall mean reaction time for all stimuli, a and b are parameters to be empirically determined, and I(P) is the mean information content, or uncertainty, of one stimulus event as measured by the information measure due to Shannon (1948). When the stimuli are all equally likely (i.e.,/?! = ... =pn = 1 /n), Equation (1) reduces to
RT= a + b\og2n-(2)
In the case of a single stimulus in = 1), if follows from both equations that RT = a so that a is the so-called simple reaction time.
