The general-gamma distribution and reaction times

Detection of auditory signals presented at random times: III

“…First, given any false-alarm rate, and any S, mean reaction time decreases with increasing signal intensity. This trend is consistent with several previous studies (McGill, 1961;McGill & Gibbon, 1965;Kohfeld, 1968). However, the pattern as a function of false-alarm rate is not simple.…”

On the processing of simple visual and auditory stimuli at distinct levels

“…However, the more complex-looking distributions produced by noise and tone catch trials depart from symmetry and high peaks and are more similar to those obtained in choice RT tasks of Snodgrass et al (1967) and of Hohle (1967). These types of low-peaked skewed distributions are fitted variously by the displaced gamma distribution (McGill & Gibbon, 1965;Snodgrass etal, 1967), normal plus exponential (Hohle, 1967), and by a negative binomial (LaBerge, 1962).…”

The ex-Wald distribution as a descriptive model of response times

“…However, for over 35 years, it has proved difficult to reach an agreement as to which of the two components represents which processing stage. For example, McGill and Gibbon (1965) interpreted the exponential component as a residual motor latency, in direct conflict with the original interpretation of Hohle, who took it to represent the perception and decision latency. As a consequence of these stage identification problems, part of the more recent research has abandoned substantive interpretations of the ex-Gaussian model altogether.…”

“…Given that empirical RT distributions are not symmetrical, the normal component is convolved with an independent ex-ponential component M, which generates the required skew. The assumption of an additive exponential RT component is supported by independent evidence from (1) approximations to the tail of the log RT survivor function or its derivative the hazard function (McGill & Gibbon, 1965;Ueno, 1992), (2) RT deconvolutionapproaches (Burbeck & Luce, 1982;Luce, 1986), and (3) statistical procedures specifically designed to test for additive exponential RT components (Ashby, 1982;Ashby & Townsend, 1980). Initially attempts were made to interpret D and M as the durations of two serially organized stages and to identify them with specific component processes such as "perception and decision" versus "organization and execution of the motor response" (Hohle, 1965, p. 384).…”