Nonparametric Causal Inference for Optogenetics: Sequential Excursion Effects for Dynamic Regimes.

Area of Science:

• Neuroscience
• Causal Inference
• Behavioral Science

Background:

• Optogenetics is a key neuroscience tool for linking neural circuits to behavior.
• Standard optogenetics analysis methods discard valuable data and limit causal questions.
• Existing methods struggle with complex, dynamic neural circuit manipulations.

Purpose of the Study:

• To develop novel causal inference methods for analyzing optogenetics experiments.
• To enable richer causal questions about neural circuit manipulation and behavior.
• To address limitations in standard optogenetics data analysis.

Main Methods:

• Connecting optogenetics analysis to sequentially randomized experiments in causal inference.
• Proposing nonparametric estimators for open-loop (static) and closed-loop (dynamic) optogenetics designs.
• Extending marginal structural models and adapting excursion effect methods for positivity violations.

Main Results:

• Developed a taxonomy of identifiable causal effects for optogenetics.
• Proposed and provided statistical guarantees for inverse probability-weighted and multiply-robust estimators.
• Demonstrated enhanced behavioral insights from real optogenetics data compared to standard analyses.

Conclusions:

• The proposed causal inference framework significantly expands the scope of questions addressable with optogenetics data.
• New methods provide statistically sound and computationally scalable tools for analyzing complex optogenetics experiments.
• This approach uncovers causal effects of neural circuit manipulation on behavior previously obscured by standard analyses.