Reinforcement and the serial reaction time task

Abstract

The aim of this thesis was to assess learning in the absence of verbal mediation in human participants. The Serial Reaction Time task (or SRT task) has been used in the cognitive literature to assess implicit learning in human populations. A modified version of this task was generated where points based reinforcement was available for responses that met a speed-based criterion. In Experiment 1, points delivery was programmed to occur based on the four simple schedules of reinforcement. Patterns of responding were analysed and compared to non-human animal responding on each of the four schedules. Some aspects of responding paralleled patterns previously described in the non-human animal literature, but others did not. In Experiment 2, the influence of continuous reinforcement on performance was assessed. Evidence of schedule sensitivity was obtained, indicating that reinforcement facilitates performance on both implicit and explicit aspects of this task. Experiment 3, contained two mini studies where sequence length and the addition of accurate instructions were manipulated in order to assess whether null results from Experiment 1 were due to floor or ceiling effects. It was concluded that the standard sequence used in Experiments 1 and 2 did allow for the expression of learning in this task. In Experiment 4, reinforcer magnitude was manipulated. It was found that performance was facilitated by money-based reinforcement to a greater extent than it had been when points were obtained for fast responding. The results from this set of studies allow several conclusions to be made about learning in the absence of verbal mediation. They support suggestions made by previous researchers that learning in the absence of verbal mediation is best assessed using a covert response and or a cover story about the true nature of the task. The finding that verbally mediated and non-verbally mediated aspects of responding were influenced by reinforcement and reinforcer magnitude suggests that the influence of other behavioural phenomena (for example, punishment and negative reinforcement) could also be assessed in the absence of verbally mediation using this task. This finding also has implications for the cognitive literature where implicit and explicit learning are regarded as two separate learning systems. The modified SRT task developed in this set of studies allows for the demonstration of contingency sensitive behaviour. It also addresses some of the criticisms levelled at previous tasks used in the operant literature such as inadequate awareness measures. It was concluded that SRT task could be modified to be a useful tool is assessing learning in the absence of verbal mediation in humans.