A parsimonious alternative to the pacemaker/accumulator process in animal timing

Models of animal timing often assume the existence of a pacemaker whose var iance is produced by a Poisson process. Later stages in these models typica lly include an accumulator, additional sources of variance, and a process i n which the accumulated value is compared to reference memory. Together, th ese stages produce a constant coefficient of variation (i.e. the 'scalar pr operty,' often observed in experiments) even though at least one of the sou rces of variance is not scalar. We propose a more parsimonious source of sc alar variance. We define a linear failure rate function and demonstrate tha t its resulting distribution function, called the Rayleigh distribution, al ways produces the scalar property without requiring: (a) additional sources of variance, or (b) that comparisons with reference memory always be based on ratios. Continuous and discrete versions of this model are derived. Thi s model may serve as a more parsimonious pacemaker than others currently pr oposed and may replace the pacemaker/accumulator process in timing models. (C) 1998 Elsevier Science B.V. All rights reserved.