THE COMPARATIVE PSYCHOLOGY OF FIXED-INTERVAL RESPONDING - SOME QUANTITATIVE-ANALYSES

Data on variation in the average relative frequency of responding with elapsed time since reinforcement during various fixed-interval schedules were available from cats, rats, woodmice, pigeons, turtle doves, a fish species Tilapia, and fresh-water turtles. These data were analyzed in a uniform manner by fitting Gaussian curves to the response frequency versus time functions, with the curve peak forced to be at the fixed-interval value. This analysis yielded a curve coefficient of variation (curve standard deviation/peak location), a measure of the precision of within-interval temporal control. Overall, Gaussian curves fitted data well, and two general trends could be noted. First, curve coefficient of variation tended to increase with increases in the fixed-interval value (although for most species used a range of interval values could be noted over which the coefficient of variation remained approximately constant). Second, different species differed markedly with respect to the absolute value of coefficient of variation obtained, by implication the quality of temporal control manifested. Lowest values (i.e., best temporal control) were obtained in data from cats, rats, and mice (as well as data from monkeys taken from another study). Pigeons produced higher values, then fish, then turtle doves (whose temporal control was markedly worse than that exhibited by pigeons), and finally turtles. A simple model deriving responding during fixed intervals from a mixture of timing and nontiming processes predicted that curve coefficient of variation would increase with interval value, even if the sensitivity of an underlying timing mechanism were constant. The model thus reconciled an underlying scalar timing process with obtained behavior, which at first sight violated scalar timing. The model further suggested that different species probably did differ in their underlying timing capacity, as the amount of response generation resulting from nontiming processes would need to be implausibly large to reconcile all the obtained coefficients of variation with an underlying timing process of uniform senstivity. © 1991.