Criterion learning in rule-based categorization: simulation of neural mechanism and new data

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Modeling

Background:

• Perceptual categorization involves rule selection and criterion learning.
• The neuroscience of criterion learning is poorly understood, with limited biological mechanisms in existing models.

Purpose of the Study:

• Introduce a novel computational model, Heterosynaptic Inhibitory Criterion Learning (HICL), for criterion learning.
• Provide a biologically-based explanation for criterion learning within a rule-learning framework.
• Test key aspects of the HICL model using new human category-learning data.

Main Methods:

• Developed the Heterosynaptic Inhibitory Criterion Learning (HICL) model.
• HICL utilizes prefrontal cortex cells and heterosynaptic error-driven Hebbian learning at inhibitory synapses.
• Conducted two human experiments to gather categorization data.

Main Results:

• HICL provides a biological mechanism for criterion learning.
• Experiment 1 showed changing rule criteria is easier when irrelevant dimensions also change.
• Experiment 2 demonstrated that changing the relevant rule dimension and learning a new criterion is harder but also facilitated by irrelevant dimension changes.

Conclusions:

• The HICL model accounts for new human categorization data.
• Findings suggest that changes in irrelevant dimensions can facilitate criterion learning and rule changes.
• The study has implications for future research on neural mechanisms of rule learning.