Probabilistic modeling reveals coordinated social interaction states and their multisensory bases

Area of Science:

• Ethology
• Neuroscience
• Computational Biology

Background:

• Social behavior in animals is complex, involving sensory input and internal states.
• Understanding social interactions is key to ethology and behavioral neuroscience.
• Zebrafish (Danio rerio) are a highly social model organism for studying collective behaviors.

Purpose of the Study:

• Investigate how zebrafish use multiple sensory modalities to guide social behavior.
• Uncover the complex features of pairwise social interactions in early development.
• Develop computational models to classify states of coordinated interaction.

Main Methods:

• Utilized a novel hidden Markov model with constrained linear-model emissions.
• Automatically classified states of coordinated interaction using predictive movement analysis.
• Observed spontaneous pairwise interactions in zebrafish over time.

Main Results:

• Identified two distinct social interaction states in zebrafish, alternating within sessions.
• Discovered long-range interactions (shoaling) depend on vision.
• Found mechanosensation underlies rapid synchronized movements and parallel swimming (schooling precursors).

Conclusions:

• Two fundamental modes of pairwise social interaction exist in zebrafish, differentiated by movement and timescale.
• Vision and mechanosensation play distinct roles in mediating different types of social behaviors.
• The developed hidden Markov model approach is broadly applicable to analyzing pairwise interactions in various species.