New paper published in Journal of Cognition

We adopt a computational approach to investigate the impact of learning and synchronization on both behavioral (reaction time, error rate) and neural (θ power) measures. We train four models varying in their ability to learn and synchronize for an extended period of time on three seminal action control paradigms varying in difficulty. Learning, but not synchronization, proved essential for behavioral improvement. Synchronization however boosts performance of difficult tasks, avoiding the computational pitfalls of catastrophic interference. At the neural level, θ power decreases with practice but increases with task difficulty.

Reference:

  • Huycke, P., Lesage, E., Boehler, C. N., & Verguts, T. (2022). Computational investigations of learning and synchrony in cognitive control. Journal of Cognition