Resting-state connectivity in action video game players
How do some cognitive training programs affect brain networks?
In the final part of my PhD dissertation, I explored how some cognitive enhancement regimes impact our brain, using resting-state fMRI (when the brain does not perform any task).
It is somehow known that improving cognition using simple tasks is challenging (like the many times we try brain game apps and fail). But some research suggests that playing action video games may lead to measurable benefits.
To study this, I developed an explainable machine learning pipeline that was able to classify individuals as either action gamers or non-gamers (based on their resting-state brain connectivity):
A surprise side-finding emerged when I systematically tested various metrics and parcellations: choices of connectivity and parcellation atlas can significantly influence conclusions about brain functions; especially those higher-level constructs like cognitive control.
The key result, however, was that the difference between gamers and non-gamers did not lie within one specific brain network. Rather, it emerged from the interaction between three control networks: the cingulo-opercular network (CON), the frontoparietal network (FPN), and the sensorimotor network (SMN)–each associated with different kinds of control.
Despite tremendous support from my collaborators, we never published this chapter as a standalone manuscript; the sample was small and results were not conclusive enough to support or reject the effects of playing video games. Still, it gave me valuable insight into the importance of methodological decisions in cognitive control research:
- Cognitive control is not a collection of isolated parts but a complex system of interacting pieces. Any analysis, model, or cognitive training program that does not capture this complexity risks missing the bigger picture.