Presenter(s): Wenjia Cao
Faculty Mentor(s): Dasa Zeithamova-Demircan & Caitlin Bowman
Poster 132
Session: Social Sciences & Humanities
Concept learning involves linking pieces of information to a shared category label, like learning that furry creatures that live with humans and bark are called dogs. What factors affect how well people learn new categories? Prior research suggests that people can learn categories either by memorizing individual category members or by averaging across category members to form an abstract representation of the perfect category member, known as the prototype. We reasoned that if people learn categories through memorization, then they should learn better from small training sets, but set size should not affect prototype learning. We also reasoned that if people learn categories through memorization, then they should learn better from training sets where items are distinct from one another (i.e., atypical), but that prototype learning would be facilitated by training on typical items. To test our hypotheses, separate groups of participants underwent category training that varied in the number and typicality of category members. During training, participants saw category members one at a time, guessed which category they belonged to, and were told whether they were right or wrong. Following training, participants were tested on their ability to categorize the training items as well as new examples that they had never seen. We found that people who trained with more typical items learned more quickly and were better at categorizing new examples. Training size did not have a significant influence on learning rate or categorization accuracy. Therefore, our results support the idea that the typicality of training items greatly influences category learning, which is likely because it promotes formation of abstract, prototype category representations.