Optimal dynamic coding by mixed-dimensionality neurons in the head-direction system of bats

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Neural representations often involve neurons with pure tuning (one dimension) and conjunctive tuning (multiple dimensions).
• This mixed-dimensionality coding appears redundant but may offer functional advantages.

Purpose of the Study:

• To theoretically analyze and experimentally validate the efficiency of mixed-dimensionality neural codes.
• To understand how stimulus dynamics influence the optimal representation strategy.

Main Methods:

• Theoretical modeling of neural population coding under varying stimulus dynamics.
• Experimental recordings in the bat head-direction system to observe neural tuning.
• Analysis of neural responses in relation to angular velocity.

Main Results:

• Theoretical analysis showed mixed-dimensionality codes are efficient across different behavioral timescales.
• Conjunctive coding is robust to rapid stimulus changes; pure coding is efficient for slow changes.
• Experimental data confirmed that head-direction cells dynamically switch between pure and conjunctive tuning based on angular velocity.

Conclusions:

• Optimal neural dimensionality depends on population size and decoding time.
• Mixed-dimensionality codes provide a flexible and efficient strategy for representing complex stimuli.
• This principle likely underlies mixed representations in diverse brain systems across species.