The Use of Parenthetical and Spatial Cues to Facilitate Learning in a Statistical Learning Task Michelle Ellefson Southern Illinois University at Carbondale ellefson@siu.edu   During the past few decades attempts have been made to integrate cognitive and education theory in order to create more effective educational curricula. Across the educational curriculum students have difficulty learning hierarchically structured information. The difficulty in learning this type of information may have more to do with the underlying statistical structure rather than the complexity of the content itself. The goal of this project is to gain a better understanding of the relative difficulty of extracting the statistical properties of recursive structure from sequentially presented information, with and without cues. Recursive structures are created using self-referential procedures. Each pass through a self-referential procedure allows information to be added; thus increasing the hierarchical complexity by embedding information to the previous structure. Information can be embedded into various locations, depending on the specific self-referential procedures used to create the structure. There were two types of recursive structure used in this experiment:  right-branching and center-embedded. Right-branching structures produce extensions of a sequence at its right end. Center-embedded structures embed information from the inside out. In the experiment, which was developed using a combination of artificial grammar learning and serial reaction time techniques, college students learned the underlying statistical properties of right-branching and center-embedded recursive structures with either spatial, parenthetical, or no cues. Overall, the results indicated that the presence of cues facilitated learning. Generally, center-embedded structures were more difficult to learn than right-branching recursive structures, although there were some differences in performance between the center-embedded and right-branching structures depending on the presence or absence of cues. Items containing more levels of embeddedness were more difficult to learn than items with less embeddedness. A better understanding of why certain statistical structures are more difficult to learn and the variables that enhance learning may begin to provide a better understanding of how to facilitate learning in the classroom. This work is supervised by Morten H. Christiansen, Cornell University Michael E. Young, Southern Illinois University at Carbondale   Back to main page