How to test for phasic modulation of neural and behavioural responses

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• The influence of neural oscillation phase on cognitive functions like perception is a growing research area.
• Optimal statistical methods for detecting these phase-dependent effects remain unclear.

Purpose of the Study:

• To compare the efficacy of various statistical methods in detecting phase effects on neural and behavioral responses.
• To identify the most suitable analytical approaches for experimental designs investigating neural oscillation phase.

Main Methods:

• A simulation approach was employed to test different statistical methods.
• Simulations involved manipulating parameters related to phase effects and experimental design.
• The performance of methods was evaluated on both dichotomous (e.g., hit/miss) and continuous (e.g., scalp potentials) response variables.

Main Results:

• Methods utilizing regression of single-trial responses on circular predictors (sine and cosine) demonstrated superior performance.
• This optimal method was effective across all simulated parameters and for both dichotomous and continuous response variables.
• The study identified specific optimal methods for various parameter combinations.

Conclusions:

• Regression on circular predictors is the recommended method for analyzing phase effects in neural and behavioral data.
• This finding provides a foundation for optimizing experimental designs and statistical analyses in future research.
• The study offers practical tools by providing MATLAB code for the tested statistical methods.