Year: 2023
Type: dissertation
Abstract: This dissertation explores the assumptions of the Generalized Context Model (GCM) concerning human categorization, emphasizing the impact of category frequency during the learning stage. Through category learning tasks, we assessed three proposed variants of GCM compared with Nosofsky’s original model, focusing on examining the computational mechanisms influencing classification decisions. We employed extensive sets of stimuli, derived from bivariate normal distributions, and manipulated category frequency during training across three independent studies employing different category structures. Participants classified novel stimuli in the transfer phases. The results showed a notable frequency effect, where participants displayed a preference for categorizing novel items based on the more frequently encountered category during training. Model-based analyses indicated that the variant of GCM implementing a reinforcement learning rule, consistently outperformed others in model fits along with successfully reproducing the observed classification pattern across all studies. This suggests a recency-weighted representation accumulation as a plausible mechanism underlying categorization. Our results signify a crucial link between category frequency and categorization performance across both unverbalizable, implicit, and verbalizable, rule-based structures. The finding extends the theoretical understanding of the GCM, offering a refined model variant that accurately reflects the observed frequency effects in human categorization, and posits a foundation for further explorations into the nuanced mechanisms of categorization.
Author: Yang, Dong-Yu
Institution: Texas A&M University
Language: English
Subfield: Experimental and Cognitive Psychology
Field: Psychology
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7047245481
