The Behavioral Effects of Some Temporally Defined Schedules of Reinforcement

The differences in behavioral effects between interval and ratio reinforcement have been pointed out by several investigators (1, 6, 9, 15), and reveal themselves most clearly in comparisons of over-all response rates, temporal patterns of cumulative-response curves, and subsequent extinction responding.In view of the different procedures followed by the experimenter and these contrasting behavioral effects, most researchers have regarded the two categories as basically distinct.Skinner (15) has given a plausible explanation of why differences in behavior emerge from the two types of schedule.On interval schedules, long inter-response times (IRT's) are more likely to be reinforced than short IRT's, and this fact may account for the relatively low rates characteristic of interval reinforcement.On ratio schedules, however, rapid rates of responding will be more frequently rein- forced than low rates, so that short IRT's have a greater probability of reinforcement than they would on a comparable interval schedule.Thus, the higher response rates obtained with ratio reinforcement may result from this contingency.Several studies have shown that IRT distributions do, in fact, change, depending on the relative reinforcement of different IRT's.(Actually, of course, it is the response which terminates the IRT that is "reinforced.")Wilson and Keller (18), for example, found that response rates decreased as successively longer IRT's were selectively reinforced, while Sidman (14), in studying the effects of certain drugs on timing behavior, obtained IRT distributions which displayed peaks in the region of the minimum IRT reinforced.Anger (1) has recently offered a further analysis of the effects of several different schedules on IRT distributions.He found that the probability of occurrence of different IRT's on a variable-interval (VI) schedule was closely related to the number of reinforcements per hour given each IRT class, rather than to the number 1 This study is based on a dissertation submitted to the Faculty of Pure Science of Columbia University in partial fulfillment of the requirements for the Ph.D. degree,