Disentangling Perceptual and Process-Related Sources of Behavioral Variability in Categorization.

Area of Science:

• Cognitive psychology
• Computational neuroscience
• Human perception

Background:

• Human categorization exhibits intraindividual variability, making it challenging to distinguish from bias.
• Variability can stem from different stages of the categorization process, including perception and cognition.
• Existing methods struggle to isolate the precise source of this behavioral variability.

Purpose of the Study:

• To investigate and differentiate the sources of behavioral variability in human categorization.
• To explore the impact of noise in perceptual and cognitive processes on categorization accuracy.
• To develop a method for identifying the specific step in categorization where variability originates.

Main Methods:

• A similarity-based categorization model was simulated to analyze behavioral variability.
• The study examined how noise at different processing stages (perception, attention, computation) affects model predictions.
• Continuous model predictions were compared against discrete category responses to identify noise sources.

Main Results:

• Noise in perceptual or cognitive processes led to significant misestimations of response determinism.
• Discrete category responses were insufficient to identify the source of variability, as different noise types produced similar patterns.
• Continuous model predictions successfully distinguished noise sources: noisy feature perception, attention, and similarity computation yielded distinct prediction patterns.

Conclusions:

• Continuous measurement of category beliefs, like probability judgments, is crucial for disentangling sources of behavioral variability.
• Identifying the specific source of noise (perception, attention, or computation) can inform targeted interventions to improve human categorization skills.
• This research provides a framework for understanding and addressing variability in human decision-making processes.