|
The time-varying property of neural circuits is genuinely true due to their underlying biophysical plasticity, but has not been fully studied in retrieving effective connectivity by Granger causality for fMRI time-series data. The impacts of the time-varying property on the results given by classical Granger causality were discussed rigorously and demonstrated numerically. Based on Granger causality with signal-dependent noise, a novel method was proposed to measure the time-varying information flow between brain regions based on the fMRI time-series data. The method was validated on numerical toy model and applied to the fMRI time-series data collected in a two party bargaining game. The information flow between the left rostral prefrontal cortex [rPFC, approximating the Brodmann area 10 (BA10)] to the right temperoparietal junction (rTPJ) were recovered by the proposed time-varying Granger causality with signal-dependent noise (GCSDNtv). The top-down information flow from BA10 to rTPJ was negatively correlated with a behavioural characteristic indicating the involvement of the executive function processes, while significantly greater bottom-up information flow was observed for the subjects who switched between different strategies being aware of the external changes in the bargaining game than those who employed a single strategy in all trails ignoring the environmental changes. |
|
Keywords:Mathematical biology, signal-dependent noise, time-varying Granger causality, information flow |
|