How to train a module: testing assumptions of modularity hypothesis in reasoning
Dual process and modular theories of reasoning have traditionally opposed each other. Recent development of hybrid dual models on the one hand, and a more flexible conception of modularity on the other have arrived at the same point. Both views now focus on the role of background knowledge (mindware, virtual modules) and the interaction between automatic processing mechanisms in producing normatively correct responses on reasoning tasks. One of the key next steps for both theories is understanding how the properties of modules (mindware) change through practice and how that affects reasoning. This thesis set out to develop a method for studying module acquisition, by training participants to use diagrams for representing word problems, structurally similar to the ‘conflict’ problems commonly used in reasoning research. The key properties of modularity (automaticity, rapidity) were evaluated by comparing pre-test and post-test solution accuracy and reaction time under cognitive load on a series of diagram recognition questions. The relevance of the diagramming skill to solving ‘conflict’ problems was assessed by comparing changes in solution accuracy and reaction time under load on a series of ‘conflict’ problems. The results suggest that overall the training was ineffective in creating a virtual module (acquiring new mindware) and influencing performance on ‘conflict’ problems.