A COMPUTER ANALYSIS OF SERIAL INTERACTIONS IN SPACED RESPONDING¹

Abstract

Serial dependencies in interresponse times were studied by means of a digital computer. In monkeys exposed to a DRL 20-sec schedule of reinforcement, serial interactions appeared at all stages of training. Early in training the serial effects consisted of trains of relatively long interresponse times interspersed among trains of relatively short ones. Later on, the serial effects appeared to be characterized by a tendency to drift up and down in long wavelength periods around the minimum interval required for reinforcement. After training to a point at which most interresponse times produced reinforcement, serial effects of a still more subtle nature appeared. These effects were made apparent by autocorrelation and power spectrum methods and consisted of both long-term and extremely short-term fluctuations in interresponse times.When exposed to a schedule of reinforcement that reinforces only those interresponse times (IRTs) exceeding a specified durationthe DRL schedule-organisms adjust to such requirements by lengthening the interval between responses.Various aspects of how organisms space responses in time on the DRL schedule of reinforcement have been studied but little is known about the apparent serial interaction in IRT duration noted by a number of authors. One instance of this interaction is a tendency for reinforced IRTs to occur in clumps (e.g., Kelleher, Fry, and Cook, 1959;Farmer and Schoenfeld, 1964). Another instance is the tendency for closely-spaced responses to be emitted in bursts, especially after IRTs just short of the duration required for reinforcement (e.g., Sidman, 1956). Recently, Ferraro, Schoenfeld, and Snapper (1965) positively correlated. In order to be able to offer a more detailed analysis of these rather subtle effects than has been possible in the past, the present study employed digital computers to acquire and process data. METHODThe subjects were three female M. speciosa monkeys, weighing about 3 kg each, which were maintained in primate restraining chairs (Foringer) enclosed in large chambers. The operanda were microswitch levers (Lehigh Valley monkey model) mounted on the chairs. About 30 g of force was needed to operate a lever. The levers were connected to the external sense lines of a LINC computer (Clark and Molnar, 1965) through Schmitt triggers and flip-flops. The LINC programmed the schedule and recorded all IRTs with a resolution of 40 msec. Each depression and release of a lever was recorded as a response and produced a 20 msec, 2500 cps tone in the chamber from a Sonalert (Electropac, Inc.), immediately after release. The release contingency was introduced to avoid prolonged holding of the levers. A white noise generator helped mask extraneous sounds.Reinforcement occurred each time a response terminated an interresponse interval of 20 sec or more (DRL 20). The reinforcing stimulus consisted of 0.3 ml of a fruit drink delivered via a solenoid valve. A short metal nipple extending from the outlet port of the valve enabled the monkey to ingest the fruit 619 VOLUME...

Response—reinforcer Contingency and Spatially Defined Operants: Testing an Invariance Property of Phi

“…Increasing m will decrease the effect of any first or higher order sequentials. The latter are seldom observed at substantial levels in operant behavior, even when explicitly reinforced (see Shimp, 1973;Weiss, Laties, Siegel, & Goldstein, 1966). Hence, m must be large enough to attenuate the effects of sequential dependencies but small enough to reflect the current response distribution.…”

Serial properties of behavior and their chemical modification

“…A time series with slow, long-wavelength fluctuations will then show most of its power at low frequencies, and a rapidly varying time series will show a preponderance of power at high frequencies. Such an analysis was undertaken by Weiss, Laties, Siegel, and Goldstein (1966), to quantify some of the serial aspects of spaced-responding performance. It demonstrated that, at least under the specific conditions of the experiment, spaced-responding performance displays long-wavelength oscillations in IRT duration.…”

Experimental Strategies for Research on Multiple Chemical Sensitivity

“…They also make no assumptions about the statistical structure of a time series and are fairly simple to conduct. They are also fairly simple to interpret because, as Edgington (11) (14) and of interresponse times in schedule-controlled operant behavior (15).…”