The Structure of Perceptual Categories

Area of Science:

• Cognitive Science
• Computational Neuroscience
• Psychology

Background:

• Human observers exhibit remarkable generalization capabilities, forming broad categories from limited data.
• One-shot categorization, generalizing from a single instance, highlights sophisticated inference mechanisms.

Purpose of the Study:

• To formalize the principle guiding human generalization, termed the Genericity Constraint.
• To propose a theory where categorical hypotheses are generative models interpreted via transformations.
• To investigate the role of a structured lattice of models in categorization.

Main Methods:

• Developed a theory of categorization based on generative models and a lattice structure.
• Formalized the Genericity Constraint: choosing the simplest, most generic model for an object.
• Conducted experiments where subjects generalized from simple figures to test the theory.

Main Results:

• Experimental results support the proposed theory of categorization.
• The Genericity Constraint and the lattice structure accurately predict human generalization behavior.
• Subjects favored interpretations where observed objects were generic examples of a category.

Conclusions:

• The Genericity Constraint provides a powerful principle for understanding human categorization.
• Generative models organized in a lattice effectively explain how humans generalize from limited data.
• This framework elucidates the cognitive mechanisms underlying one-shot categorization and broader generalization.