Pavlovian Conditioning Under Partial Reinforcement: The Effects of Nonreinforced Trials Versus Cumulative Conditioned Stimulus Duration

摘要

A core feature of associative models, such as those proposed by Allan Wagner (Rescorla & Wagner, 1972; Wagner, 1981), is that conditioning proceeds in a trial-by-trial fashion. with increments and decrements in associative strength occurring on each occasion that the conditioned stimulus (conditional stimulus, or CS) is present either with or without the unconditioned stimulus (US). A very different approach has been taken by theories that assume animals continuously accumulate information about the total length of time spent waiting for the US both during the CS and in the absence of the CS (e.g., Gallistel & Gibbon. 2000). Here we describe 3 experiments using within-subject designs that tested trial-based and time-accumulation accounts of the acquisition of conditioned responding using magazine approach conditioning in rats. We found that responding was affected by the total (cumulative) duration of exposure to the CS without the US rather than the number of trials on which the CS occurred without the US. We also found that exposure to the CS without the US had the same effect on conditioning whether that exposure occurred shortly (60 s) before each CS-US pairing or whether it occurred long (240 s) before each pairing. These findings are more consistent with time-accumulation models of conditioning than trial-based models like the Rescorla-Wagner model and Wagner's (1981) sometimes opponent process model. We discuss these findings in relation to other evidence that favors trial-based models rather than time-accumulation models.