Temporal connectivity patterns of the cortico-limbic learning and rewards system

The human learning and rewards system is comprised of a number of cortical and subcortical neural regions, including the orbitofrontal cortex, striatum, and anterior cingulate. While modern neural imaging methods such as functional magnetic resonance imaging (fMRI) and functional positron emission tomography (PET) can successfully detect the activity of these regions, they cannot discern temporal activation patterns, due to the slow onset of the blood oxygen level dependent (BOLD) effect. Magnetoencephalographic imaging (MEG) is able to capture these temporal patterns but traditionally has been unable to detect activity originating from the deeper regions of the brain due to signal attenuation and high noise levels. The recently published exSSS method has shown significant promise extracting deep signals from MEG data. To elicit appropriate subcortical activity we utilized a previously published gambling task. This paradigm has been shown to differentially activate a number of subcortical regions within the rewards system, including the orbitofrontal cortex (OFC), striatum, and anterior cingulate cortex (ACC), based on reward-related feedback. MEG analysis using source localization methods in conjunction with source signal reconstruction techniques yielded neural activation time courses for each of the regions of interest. Granger causality was used to identify the temporal relationships between each of these regions, and a possible functional connectivity map is presented. The behavioral paradigm was replicated using functional magnetic resonance imaging. fMRI activity patterns were similar to those previously reported in the literature using this paradigm. Additionally, the fMRI activation patterns were similar to those obtained via MEG source reconstruction of the exSSS-processed data. Our results support the literature finding that the rewards network is differentially activated based on feedback. Additionally, these results demonstrate the efficacy of the exSSS signal processing method for extracting deep activity, and suggest a possible use for MEG in the imaging of deep activity using other behavioral paradigms.