Characterizing the impact of category uncertainty on human auditory categorization behavior

Area of Science:

• Cognitive Psychology
• Auditory Perception
• Decision Making

Background:

• Categorization is a key cognitive function, crucial for interpreting sensory information.
• While visual categorization is well-researched, auditory categorization, especially with overlapping categories, remains less explored.
• Understanding how humans adapt to changing category statistics is vital for auditory perception research.

Purpose of the Study:

• To investigate if and how humans learn and utilize category distributions and prior probabilities in auditory categorization tasks.
• To determine if human decision-making in auditory categorization aligns with optimal strategies or exhibits sub-optimal behavior like probability matching.
• To explore the influence of sensory uncertainty and noise on auditory categorization.

Main Methods:

• Subjects classified tone burst frequencies into two overlapping uniform categories.
• Prior probabilities of category occurrences were systematically varied.
• Frequency-discrimination thresholds (sensory uncertainty) were measured.
• Behavior was modeled using Bayesian approaches, comparing optimal strategies with probability matching.

Main Results:

• Most subjects rapidly learned and adapted to changes in prior probabilities.
• Probability matching, rather than a strictly optimal strategy, better predicted and fit individual subject behavior.
• Models confirmed subjects' ability to learn category priors and approximate category distributions.

Conclusions:

• Human auditory categorization learning adapts to changing prior probabilities.
• Decision-making in this auditory task often follows probability matching, suggesting a deviation from purely optimal strategies.
• Optimal decision-making may emerge from strategies using non-stationary category information and short-term stimulus history, reconciling optimal and observed behavior.