Schedule of Events | Search Abstracts | Invited Symposia | Symposia | Poster Sessions | Data Blitz
Mnemonic effect of category exception learning as a function of prediction error type
Poster Session B - Sunday, March 8, 2026, 8:00 – 10:00 am PDT, Fairview/Kitsilano Ballroom
Yongzhen Xie1 (), Linda He1, Michael L. Mack1; 1Department of Psychology, University of Toronto
We extract patterns from our experiences to form category knowledge, a process sometimes violated by new conflicting information known as exceptions. Current literature suggests that prediction error induced by exceptions can shift our memory into a flexible state that promotes the encoding of exception information. However, can such encoding enhancement generalize to category-irrelevant information in the learning context? We tested the impact of exception learning on memory by presenting participants (n=71) with random animal facts during a category learning task in which two exceptions were introduced late in learning. One exception was confusable with the competing category and thus violated the conceptual prediction of category membership, whereas the other exception was a visual oddball that violated the perceptual prediction of stimulus features presented in the learning task. We found that, over exception learning, participants exhibited unique memory enhancement for facts presented on learning trials related to the conceptual prediction error, but not for learning trials related to the perceptual prediction error. These findings indicate that memory for irrelevant information in the learning context can become strengthened with category exceptions, but the specific type of prediction error induced by the exception matters. Particularly, the confusable exception, by violating the conceptual prediction based on prior knowledge, can elicit more extensive memory updating and encoding compared to the oddball exception. This account was supported by measuring representation updates in category learning model simulations. In conclusion, we demonstrate the broad mnemonic consequences of exception learning and highlight the nuances of prediction error-driven memory enhancements.
Topic Area: LONG-TERM MEMORY: Episodic
CNS Account Login
March 7 – 10, 2026