Premise typicality as feature inference decision-making in perceptual categories

Area of Science:

• Cognitive psychology
• Computational neuroscience
• Artificial intelligence

Background:

• Property inferences for category instances are crucial but studied separately in categorical induction and perceptual categorization.
• Categorical induction uses real-world categories, but its representational basis is unclear.
• Perceptual categorization uses controlled stimuli, allowing for well-specified representation models.

Purpose of the Study:

• To clarify the category representation underlying feature inference by embedding categorical induction effects within perceptual categories.
• To investigate the distinctness of premise typicality and premise-conclusion similarity effects.
• To determine if premise typicality effects persist when instance similarity is rigorously controlled.

Main Methods:

• Embedded analogues of categorical induction effects (premise typicality, premise-conclusion similarity) into perceptual categories.
• Controlled similarity between instances using overlap of a small number of constrained features.
• Participants made property inferences for test cases based on displayed category instances.

Main Results:

• Demonstrated typicality effects and premise-conclusion similarity effects.
• Found no evidence of premise typicality effects when similarity was controlled, with Bayesian support for the null hypothesis.
• Observed that premise typicality effects were absent despite their prevalence in the perceptual categorization paradigm.

Conclusions:

• Premise typicality effects may not be distinct from instance similarity effects in attribute inference.
• The absence of premise typicality effects under controlled similarity conditions warrants further investigation.
• Findings challenge the separation of categorical induction and perceptual categorization research by highlighting the role of controlled similarity.