Correlated cortical populations can enhance sound localization performance

Area of Science:

• Neuroscience
• Auditory Perception
• Computational Neuroscience

Background:

• Cortical neurons often exhibit correlated activity.
• Correlation is typically viewed as hindering accurate sensory decoding.
• Previous models assumed a vector-averaging observer, where correlated neural responses degrade performance.

Purpose of the Study:

• To investigate the impact of neural response correlation on sensory decoding.
• To determine if correlated neural activity can enhance, rather than impair, performance.
• To analyze sound localization accuracy using different decoding strategies.

Main Methods:

• Analytical demonstration using Fisher information.
• Monte Carlo simulations based on empirical neural recordings.
• Utilized data from single neurons in the cat primary auditory cortex.

Main Results:

• Correlated neural responses can improve sound localization performance.
• This improvement is demonstrated for a Maximum Likelihood estimator.
• Analytical and simulation results support the enhancement of decoding performance.

Conclusions:

• Neural population correlation is not universally detrimental to decoding.
• Maximum Likelihood estimation benefits from correlated neural activity in auditory localization.
• Findings challenge traditional views and offer new insights into sensory processing in the auditory cortex.