Sidhe O’Naughton, “A Meta-Analysis of Prototype vs. Exemplar-Based Category Learning”
Mentor: Caitlin Bowman, Psychology, Letters & Science (College of)
Poster #158
Category learning a basic cognitive process that enables us to organize and classify new information, which aides in learning and memory. Decades of research has led to the development of multiple quantitative models of the cognitive processes underlying categorization. The exemplar model posits that every category member is stored individually in memory, and each new example is compared to this library of examples (Devraj, 2021). In contrast, the prototype model theorizes that we average across individual exemplars to create an abstracted ‘platonic ideal’ of a given group, containing all the essential features of the category, which then serves as the basis of categorization. Our lab has conducted multiple similarly structured studies to test predictions from these models. Here, we aim to create a lab-specific meta-analysis that combines these various studies to test how factors such as training set size, training set coherence, participant age, etc. may affect category learning performance and relative prototype vs. exemplar model fit. In these studies, participants were trained to group cartoon animals and then tested on the ability to generalize categories to new stimuli. A total of 935 participants were collected from the UWM student pool, the Milwaukee area, and Eugene, OR, with 817 young adults (between the ages of 18-30) and 118 older adults (60+). Our goal, following analysis, is to develop a web-based interface for the public to access and interact with our accumulated data and to continuously update the meta-analysis with new studies.