Perception of categories: from coding efficiency to reaction times.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Perception

Background:

• Reaction times in perceptual tasks are crucial for understanding neural decision-making.
• Most studies focus on discrete choice tasks, not category perception.
• Distinguishing category identification from continuous parameter estimation is key.

Purpose of the Study:

• To link optimal Bayesian decoding with coding efficiency in category perception.
• To characterize the best estimator for category likelihood from neural activity.
• To analytically relate decision-making processes to measurable quantities.

Main Methods:

• Utilized optimal Bayesian decoding and mutual information for coding efficiency.
• Characterized estimators using large populations of stimulus-specific cells.
• Employed the diffusion-to-bound approach to model decision processes.

Main Results:

• Established a link between optimal Bayesian decoding and coding efficiency.
• Related diffusion process bias and variance to measurable quantities.
• Quantitatively linked reaction times and discrimination accuracy.

Conclusions:

• Developed an analytical expression for mean reaction times in identification tasks.
• The model explains empirical data on phoneme identification.
• Demonstrated that reaction times depend on stimulus category boundaries.