Category Structure and Categorical Perception Jointly Explained by Similarity-Based Information Theory

Area of Science:

• Cognitive Science
• Neuroscience
• Information Theory

Background:

• Categorization is a fundamental brain process for efficient information processing.
• Categories have internal structures, including 'goodness' (item belongingness).
• Categorical perception and perceptual magnet effects demonstrate altered perception due to categorization.

Purpose of the Study:

• To derive category internal structure and categorical perception from predefined categories.
• To utilize an information theoretical framework involving probabilities and item similarities.
• To provide a unified understanding of categorization and perception.

Main Methods:

• Employing an information theoretical framework.
• Tuning pairwise similarities between items to best represent predefined categories.
• Analyzing constraints on pairwise similarities to explain perceptual phenomena.

Main Results:

• The framework naturally derives category internal structure and categorical perception.
• Constraints on similarities lead to approximate category matching, explaining 'goodness' and perceptual warping.
• Demonstrated a unified principled understanding of categorization and perception.

Conclusions:

• Similarity-based information theory offers a unified framework for understanding categorization.
• This approach explains both the internal structure of categories and perceptual alterations.
• Provides a principled explanation for categorical perception and related phenomena.