Pavlovian conditioning: How excitation and inhibition determine ideomotion

Associative theories assume a simple ordinal mapping between the strength of an association between a conditioned stimulus (CS) and an unconditioned stimulus (US) and conditioned behaviour in an experimental preparation. Recent studies that have taken multiple measures of conditioned behaviour challenge this assumption. The pur- pose of this thesis is a better understanding of the nature of these individual differences in Pavlovian conditioning, combining empirical evidence with theoretical development. It has been observed that simple auto-shaping procedures result in marked individual differences: some rats show learning by interacting with the sign (sign-trackers, STs), others by interacting with the food-well or the goal (goal-trackers, GTs). In Chapter 2, I examined the sensitivity of these two behaviours (sign-tracking and goal-tracking) to changing contingencies. In both STs and GTs, US-oriented behaviour was more sensitive to contingency changes than CS-oriented behaviour. Most attempts to ex- plain this dissociation have appealed to a dual-mechanisms approach. In Chapter 3, I present a new theoretical model, HeiDI, which integrates learning and performance from a single-process perspective. In Chapter 4, I examine two of these predictions. The first prediction relates to how the US-value affects the distribution of conditioned behaviour. According to HeiDI, a higher US-value will result in higher levels of goal-tracking in contrast with lower US value. Experiment 3 suggested that that a higher US-value results in more CS-oriented behaviour, however this was not replicated in Experiment 4. The second prediction concerns an analysis of the feature positive effect, where the discrimination emerges more readily for a feature positive design. HeiDI predicts that a feature positive effect should be more evident in CS-oriented behaviour. Experiments 4 and 5 addressed this prediction, however the animals did not show learning in the feature negative design. The implications of the new findings and the new model are discussed in Chapter 5