Reinforcement expectancy and trial spacing effects in delayed matching-to-sample by pigeons

Abstract

Four experiments assessed the role of reinforcement expectancies in the trial spacing effect obtained in delayed matching-to-sample by pigeons. In Experiment 1, a differential outcome (DO) group received reinforcement with a probability of 1.0 for correct comparison responses following one sample stimulus and a probability of 0.2 for correct comparison responses following the other sample stimulus. The nondifferential outcome (NDO) group received reinforcement with a probability of 0.6 for correct responses to either stimulus. While matching accuracy was higher for the DO group than for the NDO group, both groups showed an equivalent decline in accuracy as the intertrial interval (ITI) duration was decreased. However, within the DO group, ITI duration affected performance on low-probability-of-reinforcement trials but not on high-probabilityof-reinforcement trials. In Experiment 2, delay interval (DI) duration was 5, 10, or 15 sec and accuracy was higher for the DO group than for the NDO group at all DI durations. In addition, accuracy decreased similarly on high-and low-probability-of-reinforcement trials for the DO group as DI was increased. In Experiment 3, all birds were studied under DO conditions and ITI duration was manipulated along with DI duration. At the short DI duration, decreasing IT! duration had a detrimental effect on low-probability-of-reinforcement trials but no effect on high-probabilityof-reinforcement trials. At the long DI duration, decreasing ITI duration had detrimental effects on both types of trials. In Experiment 4, unsignaled IT! reinforcers disrupted accuracy when the DI was long and when the ITI was short. The applicability of scalar expectancy theory to these data is discussed.Numerous studies have shown that choice delayed matching-to-sample (DMTS) performance by pigeons is degraded by decreases in the temporal separation of trials (Grant, 1975;Hogan, Edwards, & Zentall, 1981;Maki, Moe, & Bierley, 1977). The poorer performance obtained with short intertrial intervals (ITI) was initially viewed as being due to competing memories from previous trials (Grant, 1975;Maki et al., 1977). However, this explanation has been called into question by several recent findings. Roberts (1980) found that short ITls disrupted performance on homogeneous DMTS trials as well as on standard DMTS trials. Homogeneous DMTS trials present the same sample stimulus on each trial, eliminating the possibility of competing sample memories. Although there is evidence that memories on trial n -1 do carry over the ITI and affect performance on trial n, both Roberts (1980) and Roitblat and Scopatz (1983) have shown that these are memories of the choice made on the previous trial rather than of the sample stimulus presented on the trial n -1. More importantly, neither study found any evidence that the carryover of memories between trials interacts with ITI duration, thus weakening explanations of the trial spacing effect based on proactive inhibition (PI). Finally,

Differential outcome expectancies and directed forgetting effects in pigeons

“…In addition, these data reveal that accuracy declined similarly on high-and low-probabilityof-reinforcement trials as the delay interval was lengthened. Santi and Roberts (1985b) reported a similar result. It appears that the use of differential reinforcement probabilities does not result in a strong bias to select the comparison associated with either the high or the low reinforcement probability as the relevant trial information is lost.…”

“…This study helps to extend the range of conditions under which DO effects can be observed (DeLong & Wasserman, 1981;Santi & Roberts, 1985a, 1985b. The failure of Santi and Savich (1985) to obtain this effect cannot be attributed to their use of a directed forgetting procedure; they probably did not provide sufficient training with 00 conditions.…”

“…However, during test sessions, performance was evaluated as a function of the various cue conditions by infrequently presenting comparison stimuli on F-cue and no-cue trials. Probability of reinforcement was the DO (DeLong & Wasserman, 1981;Santi & Roberts, 1985a, 1985b. For the 00 group, the probabilities of reinforcement were 1.0 and 0.2, whereas for the NDO group, the probabilities were both 0.6.…”

“…A number of studies have demonstrated enhanced acquisition of discrimination and superior memory performance in pigeons when sample stimui are associated with different trial outcomes than when outcomes are nondifferential. The DO effect has been demonstrated with qualitatively different reinforcers (Brodigan & Peterson, 1976;Edwards, Jagielo, Zentall, & Hogan, 1982;Honig, Matheson, & Dodd, 1984;Peterson, Wheeler, & Armstrong, 1978), reward versus no reward (Peterson, 1984;Peterson, Wheeler, & Trapold, 1980), biologically neutral outcomes (Fedorchak & Bolles, 1986), and differential probability of reward (DeLong & Wasserman, 1981;Santi & Roberts, 1985a, 1985b. Several alternative explanations of the 00 effect have been proposed and tested.…”

ROCs in rats? Response to Wixted and Squire: Figure 1.

“…In every study, including all three that are cited by Wixted and Squire (Trapold 1970;Santi and Roberts 1985;DeMarse and Urcuioli 1993), the differential outcomes (rewards) occur following a correct response to a stimulus or stimulus pairing, and learning of a set of stimulus-response-reward combinations is measured by the increase in performance accuracy (always measured as percent correct) over the course of many repetitions of the same combinations. In other words, this effect involves learning to associate specific stimuli and responses with different rewards, and the effect emerges only after hundreds of repetitions of these specific stimulus-response-reward combinations.…”