Detecting switching and intermittent causalities in time series

Area of Science:

• Neuroscience
• Complex Network Analysis
• Cognitive Neuroscience

Background:

• Complex network representations are vital for understanding brain region interactions in various states and conditions.
• The transient, intermittent nature of neural cross-talk has been largely overlooked due to limitations in traditional causality metrics requiring long time series.
• Existing methods often provide a coarse-grained view of dynamic brain interactions.

Purpose of the Study:

• To develop a novel methodology for analyzing intermittent causal coupling in neural activity.
• To capture the time-varying properties of brain interactions with high temporal resolution.
• To investigate the utility of transient network properties in discriminating between control and patient groups.

Main Methods:

• A new methodology was developed to identify non-overlapping time windows with the strongest causal coupling.
• This approach allows for a less coarse-grained assessment of time-varying brain interactions.
• The method was applied to analyze brain activity data from control subjects and alcoholic patients during an image recognition task.

Main Results:

• Short-lived, intermittent, local-scale causality metrics were found to be more effective in discriminating between control and alcoholic patient groups.
• The proposed methodology enables the creation of high temporal resolution time-varying brain networks.
• Global network metrics were less effective in distinguishing between the groups compared to transient local causality.

Conclusions:

• The transient nature of brain activity, particularly intermittent causal coupling, is crucial for understanding neural dynamics, especially in pathological conditions.
• The developed methodology offers a more refined assessment of brain connectivity, highlighting the significance of short-lived interactions.
• Focusing on intermittent causality provides a more sensitive measure for differentiating neurological conditions during cognitive tasks.