The probabilistic and dynamic nature of perception in human generalization behavior

Abstract

Generalization theories traditionally overlook how our mental representations dynamically change in the process of transferring learned knowledge to new contexts. We integrated perceptual and generalization theories into a computational model using data from 80 participants who underwent Pavlovian fear conditioning experiments. The model analyzed continuous measures of perception and fear generalization to understand their relationship. Our findings revealed large individual variations in perceptual processes that directly influence generalization patterns. By examining how perceptual and generalization mechanisms work together, we uncovered their combined role in producing generalization behavior. This research illuminates the probabilistic perceptual foundations underlying individual differences in generalization, emphasizing the crucial integration between perceptual and generalization processes. Understanding this relationship enhances our knowledge of generalization behavior and has potential implications for various cognitive domains including categorization, motor learning, language processing, and face recognition-all of which rely on generalization as a fundamental cognitive process.