Human task switching and the role of inhibitory processes : a computational modelling and empirical approach

Task switching is a behavioural paradigm within cognitive psychology that has been claimed to reflect the activity of high-level cognitive control processes. However, classic behavioural markers such as the (n-1) switch cost have also been shown to reflect a multitude of other cognitive processes. The n-2 repetition paradigm has proven more successful, with a behavioural measure (the n-2 repetition cost) agreed to be reflective of a cognitive inhibition mechanism (‘backward inhibition’). The present thesis develops computational models of task switching, including a backward inhibition model. The models are developed within the interactive-activation and competition (IAC) framework, as a development of an existing task switching model. Modelling is constrained by the general computational principles of the IAC framework and default parameter settings where these are shared with earlier models. The effect of specific novel parameter settings on behaviour is explored systematically. The backward inhibition model predicts a range of empirically observed behavioural phenomena including both n-1 switch and n-2 repetition costs, and the modulation of the n-2 repetition cost under certain circumstances, including the manipulation of intertrial intervals. A specific prediction of the model, the modulation of n-2 repetition costs according to switch direction when tasks are of different difficulties, is tested empirically, with results confirming and providing validation of the model. Finally, consideration is given to how such a backward inhibition model could be adapted to maximise performance benefits in different task switching contexts, via a process of parameter tuning.