Post-training flexibility in category learning.

Area of Science:

• Cognitive Psychology
• Computational Neuroscience

Background:

• Exemplar models of categorization assume all stored exemplars inform decisions.
• Individual differences in categorization suggest selective memory use.

Purpose of the Study:

• To test the hypothesis that people selectively emphasize subsets of exemplars for categorization.
• To investigate how selective exemplar accessibility influences categorization patterns.

Main Methods:

• Utilized a partial-XOR category structure to elicit Proximity and XOR response patterns.
• Experiment 2 involved training participants on category labels and selective exemplar memorization (Category A, B, or both).
• Bayesian inference and the Soft-max Decision Category Model (SDGCM) were used to model categorization data.

Main Results:

• Participants exhibited XOR and Proximity patterns based on targeted exemplar recognition.
• SDGCM analysis revealed differential exemplar accessibility between categories.
• Exemplar accessibility of Category A was higher for XOR patterns, while Category B was higher for Proximity patterns.

Conclusions:

• Selective emphasis on subsets of exemplars drives individual differences in categorization.
• Memory accessibility, not just content, is crucial for understanding categorization processes.
• Findings support a nuanced view of exemplar models, incorporating selective memory retrieval.