Comprehensive ethological analysis of fear expression in rats using DeepLabCut and SimBA machine learning model

Area of Science:

• Neuroscience
• Ethology
• Behavioral Science

Background:

• Defensive behaviors are crucial for survival and are studied using fear conditioning paradigms.
• Traditional methods like freezing and operant suppression don't capture the full spectrum of defensive responses.
• Rats exhibit a range of context-dependent defensive behaviors influenced by threat immediacy.

Purpose of the Study:

• To systematically quantify diverse defensive behaviors in rats triggered by threat cues.
• To evaluate the accuracy of DeepLabCut and SimBA for automated ethological analysis of these behaviors.
• To assess the anxiolytic effects of diazepam on a broad range of fear-related behaviors.

Main Methods:

• Auditory fear conditioning paradigm in rats.
• Manual behavioral scoring using Ethovision XT.
• Development and validation of a DeepLabCut + SimBA machine learning workflow.
• Administration of varying doses of diazepam to assess anxiolytic effects.

Main Results:

• Rats display risk assessment behaviors (sniffing, rearing, whisking, head scanning) in addition to freezing.
• Diazepam dose-dependently reduced these risk assessment behaviors in both sexes.
• The DeepLabCut + SimBA workflow achieved high accuracy in classifying well-represented behaviors with sufficient training data.
• Maintaining consistent experimental conditions between training and evaluation datasets is recommended for optimal accuracy.

Conclusions:

• Ethological analysis, enhanced by DeepLabCut and SimBA, offers a robust method for assessing fear learning.
• This automated approach provides an alternative to manual scoring for decoding defensive behaviors.
• The findings support further investigation into the neurobiological underpinnings of fear and anxiety in both male and female rats.