Relative validity effects with either one or two more valid cues in Pavlovian and instrumental conditioning.

Abstract

Two experiments investigated the relative validity effect with either 1 or 2 continuously reinforced cues in Wistar rats using appetitive Pavlovian and instrumental preparations. Discrimination training involved 3 compound cues containing a common element (1AX: 1BX: 2CX). In the first true-discrimination group (TD-1), CX was followed by food, but AX and BX were not. In the second true-discrimination group (TD-2), AX and BX but not CX were followed by food. In the third, pseudodiscrimination group (PD), food followed 50% of each compound. Compared with the PD group, there were lower levels of responding to X in Groups TD-1 and TD-2, which did not differ. That is, both TD treatments showed equivalent relative validity effects. There was evidence for a relative validity effect on the context. The Rescorla-Wagner model incorrectly predicts a smaller relative validity effect after the TD-2 than the TD-1 treatment. Comparator theory predicts these results.

A comparative approach to cue competition with one and two strong predictors

“…Murphy et al (2001a) reported simulations from the Rescorla-Wagner model for differences between the learning rate parameters for reinforcement that were similar to those we report here in Figure 1 for the standard relative validity preparation. The model predicted that TD-1 and TD-2 treatments would have markedly different effects on the associative strength controlled by X.…”

“…Clearly the theory predicts that the difference between TD-1 and TD-2 must be at least of the same magnitude as the difference between TD-2 and PD, and in most cases it is larger. Murphy et al (2001a) reported two experiments using rats to test this prediction. The first experiment used a classical conditioning paradigm in which brief (10-sec) audiovisual stimuli predicted presentations of food pellets.…”

“…That is to say, the associative strength of X, following a TD treatment, is independent of the difference between learning rate parameters for reinforcement and nonreinforcement, whereas it is not following PD. Murphy, Baker, and Fouquet (2001a) have described how changing the relative proportion of reinforced and nonreinforced trials has a similar influence on the relative validity effect.…”

“…One can attenuate the effectiveness of a blocking stimulus (here A) by including a second cue that is also always followed by the US. Murphy et al (2001a) made a simple modification to the basic relative validity design that allowed them to compare the effects of one and two strong predictors. This modification also generated clear, and somewhat counterintuitive, predictions from the Rescorla-Wagner model.…”

A test of Rescorla and Wagner’s (1972) prediction of nonlinear effects in contingency learning

“…When β 1 and β 2 have the same value, the model predicts that X should have the same associative strength in both the correlated and the uncorrelated groups. On the other hand, if β 1 is higher than β 2 , it predicts correctly that X in the uncorrelated group should have more associative strength than in the correlated group (see Murphy, Baker, & Fouquet, 2001, for an extensive discussion). In their original work, Rescorla and Wagner used values of 0.2 for β 1 and 0.1 for β 2 , which would lead to a correct prediction.…”