Study 5: Associations in Second Order Conditioning

February 10, 1997

Synopsis

In the first stage of this study, each rat will receive a continuation of last week's second order training. Then, each rat will be placed into one of two grlups. THe first group will receive extinction training to the first order CS. The second will receive continued pairings with the US. The effects of these two manipulations will be used to infer the nature of the association between the second and first order CSs.

Background

In this study, we will attempt a manipulation that historically played an important role in the analysis of associative processes. The issue is actually an old one: What do animals learn in a Pavlovian conditioning procedure? Some theories of learning assume that what is learned is a response process; others assume that some stimulus process is associated. In general, the debate relied more on polemics than it did on real data.

In 1958, Rozeboom suggested a direct empirical test of the two alternatives. Basically, Rozeboom proposed that the issue should be rephrased into the question: Does the CS resemble the response made during conditioning, or does it resemble the response elicited by the US? Unfortunately, no one at the time followed up Rozeboom's suggestion (although Konorski and his students published a little-noticed paper using the same procedure).

Later, Rizely and REscorla (1972) tried out a procedure in second order conditioning using Rozeboom's test. Their procedure was a simple one: after conditioning a second-order stimulus to elicit a CS, extinguish the first order stimulus. If the second-order CS was affected by this extinction, then one would have evidence that second-order conditioning was mediated by first-order associations. Rizley and Rescorla found that the second-order response was completely unaffected by extinction of the first-order reinforcer; this implied to them that second-order conditioning was a form of learning involving stimulus-response associations.

Rizley and Rescorla's procedure has been applied to a number of different conditioning situations, including activity conditioning with food (Holland and Rescorla, 1975a). In our study for this week, we will attempt a partial replication of Holland and Rescorla's experiment, using behavioural categories rather than activity counts. Please re-read the Holland and Rescorla paper. Also included in the references is a companion paper (Holland and REscorla, 1975b) that shows how this paradigm can be applied to analyze first order conditioning.

Methodology