Independent Components Analysis of Prisoners Dilemma Task-Based Data

The iterated Prisoner’s Dilemma (iPD) task is an economic-exchange game that illustrates how people achieve stable cooperation and how cooperation breaks down over repeated interactions. It provides a framework for examining the various phases that constitute a reciprocal exchange between two people. Functional magnetic resonance imaging (fMRI) studies examine social interactions within the iPD that typically are focused on limited snapshots of brain activity during the process of interaction between two people. There are three prospective events of the “social decision cascade” associated with a reciprocal exchange (Thompson et al., 2021): 1) Decision, 2) Anticipation, and 3) Feedback. The purpose of the study was to characterize changes in task-based brain connectivity during various phases during the iPD using a data-driven analytic approach (group ICA). Thirty-one adult volunteers were recruited to participate against a computerized confederate in the iPD task. However, participants were deceived into believing that they were playing with real people. fMRI data for each participant was collected at CABI, preprocessed, and analyzed using SPM12. Group ICA (implemented using GIFT software) was utilized as a functional connectivity technique to identify neural components that best explain variability in our fMRI data. These components represent networks of brain regions that were recruited during the task. Furthermore, we conducted a temporal regression to determine whether any of these neural components were temporally linked to certain periods of the task, such as decision-making, anticipation, and feedback processing. The top components we identified when exploring the decision-making period of the game were 11 and 32. Figure. The strength of the regression of Component 11 was at 0.050839 during feedback processing. The strength of the regression of Component 32 was at 0.035635 during feedback processing. However, the strength of the whole was weaker than we anticipated. Component 11 had major feedback processing in the bilateral temporal parietal junction, superior parietal lobe, and the right superior frontal gyrus. Component 32 had major feedback processing in the visual cortex. While we cannot make a definitive conclusion due to the weak data. We will need to recruit a larger subject pool to establish clear statistically significant connections among brain regions.