The discovery of novel actions

The aim of this research was to develop a behavioural paradigm capable of quantifying action acquisition. It takes the form of a series of experiments in which human participants learn to produce new actions with a joystick. Research questions were focussed on the behavioural implications of Redgrave and Gurney’s (2006) theory that dopamine neurons in the ventral midbrain play a pivotal role in the reinforcement and reselection of motor output that is essential to action learning. The first study looked at the effect of delayed audio and visual reinforcement on the ability to learn stable hand positions. Delays of 100 ms were found to impair acquisition in both modalities. This degree of temporal sensitivity supports the idea that dopamine neurons fire at low latencies to reduce the reinforcement of non-contiguous motor output. The second study investigated the effect of delay on the learning of hand movements. The movements produced during the delay period were analysed to address the question of whether the quantity of non-contingent output would impact on learning over and above the mismatch in temporal alignment. The results revealed that this was not the case, thus suggesting that timing is of primary importance to learning. The final study utilised a task requiring more complex movements, in an attempt to reduce the contribution of high-level, conscious, learning in favour of low-level non-declarative learning. Performance was compared across conditions, which differed in the quantity of spatial information provided. No evidence was found that the type of movements produced during learning impacted on later performance, thus indicating a tendency to use high-level spatial guidance of movements. All findings are discussed in terms of the value of the current paradigm and the extent to which they support the theory that action learning is mediated by a time-stamping mechanism in the midbrain.