Within- and between-subjects partial reinforcement effects with an autoshaped response using Japanese quail (Coturnix coturnix japonica)

Abstract

Animals exposed to schedules of partial reinforcement are typically more resistant to extinction than are animals trained with continuous reinforcement. This is the partial reinforcement effect (PRE). Animals experienced with both partial and continuous schedules are often more persistent on the continuous schedule, yielding a reversed PRE. Both conventional and reversed PREs have been elusive with classical conditioning paradigms. The present experiment attempted to demonstrate between-and within-subject PREs using 50% and 100% autoshaping schedules. Presence or absence of a PRE depended on the behavioral measures used. Marked terminal group differences in acquisition produced a between-subjects PRE with absolute response levels but not with rate-of-change measures. Within subjects, only choice trial comparisons were sensitive enough to differentiate the two schedules. Acquisition data were inconsistent with most of the classical conditioning PRE literature, but consistent with results reported in the autoshaping literature. These discrepancies may reflect the operant-classical interaction in autoshaping.Brown and Jenkins (1968) signaled upcoming noncontingent food presentations with a keylight and found that their pigeons pecked reliably at the key, even though responding had no scheduled effect. This training procedure, known as autoshaping, was devised originally as a method of training pigeons to keypeck, a response widely regarded (at the time) as an operant. Subsequent research has caused the autoshaped response to be regarded as largely the result of classical conditioning, with operant conditioning playing a less important role in both acquisition and maintenance (e

Acquisition with partial and continuous reinforcement in pigeon autoshaping

“…If one replots their data by reinforcer, there is some evidence of superiority in the partially reinforced conditions; however, such replotting necessarily consists of a single data point with no statistical evaluation. Crawford, Steirn, and Pavlik (1985), using Japanese quail, compared 100% and 50% reinforcement, plotting their results by trial, and not by reinforcer. If one replots their data, there is no obvious difference in initial acquisition and an asymptotic difference in favor of a partially reinforced group in both rate and percent CR measures.…”

Effects of trial duration on overall and momentary rates of maintained autoshaped keypecking: Choice and single-stimulus trials

“…However, a more traditional measure of strength of keypeck responding-rate of maintained responding-shows less sensitivity to these variables (Balsam & Payne, 1979;Crawford, Steirn, & Pavlik, 1985;Gonzalez, 1974;Kay, Hemmes, & Brown, 1984;Lucas, Deich, & Wasserman, 1981; Newlin & LoLordo, 1976;O'Connell & Rashotte, 1982;Perkins et al, 1975).…”

Hope, exploration, and equilibrated action schemes

“…Since the work of Amsel and Roussel (1952), psychologists have found that Pavlovian responses are magnified when cues are unreliable predictors of reward. In Pavlovian autoshaping, signtracking often comes to reach a higher asymptotic level when a CS is unsystematically followed by a UCS than when a CS is always followed by a UCS (Amsel et al 1964;Boakes 1977;Collins et al 1983;Crawford et al 1985;Gibbon et al 1980;Gottlieb 2004;Robinson et al 2014;Swan & Pearce 1987;Torres et al 2016). The effect is easy to replicate, although some studies failed to obtain it (e.g., Papini & Overmier 1984;1985;Rescorla 1999).…”