Testing Principles of Choice in Categorization.

Area of Science:

• Cognitive Science
• Decision Science
• Psychology

Background:

• Perceptual categorization is often viewed as a decision-making process.
• Decision-making research highlights choice phenomena influenced by the size and composition of choice sets.
• Rational choice principles like IIA and regularity are fundamental in understanding decision-making.

Purpose of the Study:

• To investigate potential violations of independence of irrelevant alternatives (IIA) and regularity in perceptual categorization.
• To explore how variations in learned categories and test choice sets affect adherence to rational choice principles.
• To assess the predictive power of the Luce Choice Rule and the Generalized Context Model (GCM) under these conditions.

Main Methods:

• Utilized an empirical framework adapted from the attraction effect paradigm in decision-making research.
• Tested two variations: one where learned categories differed across participants (Experiments 1a, 1b), and another where test choice sets varied within participants (Experiments 2a, 2b).
• Examined adherence to IIA and regularity principles based on the experimental manipulations.

Main Results:

• Violations of both IIA and regularity were observed when the set of learned categories varied across participants.
• These rational choice principles were not violated when test choice sets varied within participants for a fixed set of learned categories.
• The Generalized Context Model (GCM) could explain these violations, but required condition-specific response bias parameters.

Conclusions:

• Perceptual categorization can exhibit emergent phenomena, deviating from standard rational choice predictions.
• The structure of the learned category set significantly influences whether rational choice principles are upheld.
• Findings suggest that decision-making models need to account for context-dependent response biases in categorization tasks.