The role of chunking and schemas in learning and drawing

Learning by drawing raises questions related to the organization and internal processing involved during graphical production. This thesis explores how and to what extent, spatial and semantic information influences learning through drawings. It investigates the roles of chunking and schemas in learning through drawings by manipulating the spatial and semantic content of the presented stimuli, which participants reproduced using different methods over repeated sessions. Over three experiments with adult participants, multiple measures were used, including: pause durations between drawn elements, numbers of reproduced objects, error rates, sequences of element production, and transitions among chunk patterns. The first exploratory study investigated the effects of chunking in the drawing of a complex abstract diagram. Five participants reproduced a single stimulus in four types of tasks, which involved delayed recall, tracing, copying and immediate recall across 10 sessions. It was found that participants learned the diagram surprisingly quickly. They used chunking in order to aid the learning processes. This effect was most obvious in the delayed recall task and least so in the tracing. The analysis of the participants' sequence of chunk production revealed that they used a spatial schema to organise the chunks. This appears to explain their rapid learning. The second study investigated the effects of semantic and spatial schemas in learning. Twelve participants drew four types of stimuli (i.e. no-structure, semantic, spatial and spatial-semantic) across six sessions. Learning was easiest in the presence of both spatial and semantic coding, followed by semantic coding alone. By contrast, it was most difficult when the stimuli had neither semantic nor spatial information. Contrary to the predictions, the spatial stimulus was far worse to learn than the semantic. The third study manipulated the strength of the spatial and semantic information in the stimulus to investigate the effects on learning of the weak and strong organisation of information in the two types of schemas. Twelve participants performed four drawings (i.e. strong-semantic, weak-semantic, strong-spatial, weak-spatial) in four sessions. In line with the hypothesis, the findings revealed that the strong semantic stimulus is a better type of stimulus for learning than the weak semantic one. The opposite applies, however, to the strong and weak spatial stimuli. A detailed analysis of the performance of these two stimuli showed that the weak stimulus had evoked a stronger schema than the strong stimulus, which reveals that spatial properties may contribute to the strength of a schema. The concluding results of these studies proposed that even purely diagrammatic stimuli are likely to be encoded semantically, as well as spatially. Furthermore, learning based on spatial coding alone may be difficult to achieve, in contrast to learning based on semantic coding alone. The combined spatial and semantic coding, however, facilitates learning better than either coding alone. These findings suggest key features that need to be considered for diagrammatic presentations used for learning in scientific and technical domains.