Nonparametric causal inference for optogenetics: sequential excursion effects for dynamic regimes.

Area of Science:

• Neuroscience
• Causal Inference
• Behavioral Science

Background:

• Optogenetics is crucial for understanding neural circuits and behavior.
• Standard analysis methods for optogenetics discard valuable data and limit causal questions.
• Existing techniques struggle with complex, dynamic experimental designs.

Purpose of the Study:

• To develop advanced causal inference frameworks for analyzing optogenetics experiments.
• To extend methods for handling dynamic treatment regimes and positivity violations.
• To enable a richer set of causal questions about neural circuit manipulation and behavior.

Main Methods:

• Connecting optogenetics analysis to sequentially randomized experiments.
• Proposing non-parametric frameworks for open-loop and closed-loop optogenetics.
• Extending history-restricted marginal structural models and excursion effect methods.
• Developing estimators, hypothesis testing, and scalable implementations.

Main Results:

• A novel taxonomy of identifiable causal effects for optogenetics.
• Methods for analyzing dynamic treatment regimes with positivity violations.
• Demonstrated application to real neuroscience data, revealing obscured causal insights.
• Statistical guarantees for proposed estimators.

Conclusions:

• The proposed framework significantly expands the scope of causal questions addressable in optogenetics research.
• New methods provide deeper insights into the causal effects of neural manipulation on behavior.
• This work bridges causal inference and neuroscience for more powerful behavioral analysis.