| Summary: | Considerable evidence has accumulated demonstrating that instrumental actions in rats can be controlled by two dissociable associative structures. During the early stages of training, responding is guided by action-outcome (A-O) associations that require both a representation of the outcome and knowledge of the instrumental contingency. However after more extended training, behaviour comes to be controlled by stimulus-response (S-R) habits that are no longer goal-directed. Despite the behavioural evidence supporting this dissociation, the psychological and neurochemical mechanisms underpinning this transition are poorly understood. To address this first issue, we compared the sensitivity to outcome devaluation of instrumental responses that were moderately or extensively trained on either interval or ratio schedules. It was found that S-R habits developed as animals achieved stable rates of responding. This was further supported by demonstrations that when well-established performance is disrupted by changes in the schedule of reinforcement or reward magnitude, or where there is no consistent relationship between behaviour and reward delivery animals appear sensitive to goal-value despite extended training. It is suggested that S-R habits develop as the molar correlation between behaviour and reward becomes well-predicted. Moreover, the work presented here also sought to elucidate further the neurochemical processes involved in the transition from action to habit. These experiments demonstrated that pre-training exposure to amphetamine leads to the early and excessive dominance of S-R processes. This provides the first direct evidence that dopamine transmission is critical to the development of S-R habits. Further experiments explored the neuropharmacological specificity of this effect and found a dissociation at the level of the receptor subtype: amphetamine enhancement of S-R learning is reversed by D, but enhanced by D2, receptor antagonists. Finally, the results are discussed in terms of optimality and certainty-based models of instrumental performance with reference to both phasic and tonic dopamine activity.
|
|---|
