Temporal structure of mouse courtship vocalizations facilitates syllable labeling

Area of Science:

• Animal Behavior
• Bioacoustics
• Computational Neuroscience

Background:

• Mice produce ultrasonic vocalizations (USVs) during social interactions, particularly courtship.
• The rules governing the temporal sequencing of USVs and their classification into discrete syllables remain poorly understood.
• Existing USV labeling algorithms lack consensus and can impact the analysis of higher-order temporal structures.

Purpose of the Study:

• To investigate the temporal structure of mouse ultrasonic vocalizations (USVs) during courtship.
• To evaluate and compare different USV labeling algorithms.
• To develop a novel method for improving USV classification based on sequence statistics.

Main Methods:

• Recording USVs during male-female mouse courtship.
• Applying three popular algorithms for USV labeling and analyzing label consistency.
• Developing the Syntax Information Score (SIS) based on information theory to rank labeling algorithms.
• Deriving the Syntax Information Maximization (SIM) algorithm to improve USV clustering using sequence statistics.

Main Results:

• A significant temporal structure was identified in mouse courtship USVs.
• No one-to-one relationship was found between labels assigned by different popular USV classification algorithms.
• The Syntax Information Score effectively ranked labeling algorithms based on predictive accuracy.
• The novel Syntax Information Maximization algorithm demonstrated improved USV clustering aligned with sequence structure.

Conclusions:

• USV sequences exhibit meaningful syntax that can be analyzed using information theory.
• Improved USV classification through methods like Syntax Information Maximization is crucial for understanding the neural control of vocalization.
• USV syntax analysis offers valuable insights into the complex communication systems of rodents.