Encoding of interdependent features of head direction and angular head velocity in navigation.

Area of Science:

• Neuroscience
• Computational Neuroscience

Background:

• The brain must encode temporal interdependencies between features for navigation.
• The head direction (HD) system requires continuous updates based on angular head velocity (AHV).
• Neural representations of interdependent features are not well understood.

Purpose of the Study:

• Investigate the population coding strategy for HD and AHV.
• Characterize the neural representation of dynamically coupled features.
• Propose a novel hybrid coding scheme.

Main Methods:

• Recurrent neural network modeling for hypothesis generation.
• Analysis of neural responses to predict HD from AHV.
• Neurophysiological recordings in the mouse HD system.
• Neural geometric analysis.

Main Results:

• Identified two distinct neuronal subpopulations: single-peaked (SP) encoding HD and multi-peaked (MP) encoding AHV.
• Both SP and MP units exhibit mixed selectivity but with functional divergence.
• MP neurons expand representational dimensionality for higher-resolution AHV encoding.
• Empirical validation confirmed differential encoding of HD and AHV by SP and MP neurons.

Conclusions:

• The mouse HD system employs a hybrid coding scheme for interdependent features.
• This scheme balances specificity (sparsity) and interdependency (mixed selectivity).
• The findings suggest this scheme may be applicable across various sensorimotor and cognitive domains.