Research
Thesis Research
Status: Complete
Links: Thesis (PDF), Poster (PDF), Defense Presentation Slides (PDF)
Summary
Lorem ipsum dolor sit amet, consectetur adipisicing elit. Voluptatum deleniti consectetur suscipit minima, nam, sint labore vitae perspiciatis voluptatibus repudiandae quisquam. Harum libero fuga iure officiis ad et, deleniti est?
Abstract
Geological taxonomies traditionally divide rocks into broad superordinate categories (Igneous, Sedimentary, and Metamorphic) each comprising many subtypes (e.g. Granite, Slate, or Sandstone). We investigated the interaction of category structure and attended category level in learning these categories. In a supervised learning experiment, subjects learned to classify photographs of rocks by either superordinate category or subtype. In a compact condition, similarities between subtypes within each superordinate category were high, whereas in a dispersed condition they were low. Hypothesizing an interaction between the compactness of categories and learned level, an advantage was found for learning of superordinate categories when categories were compact if subjects learned categories at the superordinate level. However when categories were dispersed, learning at the subtype level was found to be advantageous.
