Human brain systems have their own functional expertise. As they interact, these systems allow
for efficient goal-directed behavior. Yet, the interaction between systems is poorly understood.
The “action” system represents the sequence of actions needed to perform some task; for
instance, the sequential actions a drummer performs to hit the drums. The “when” system
represents the time at which actions need to be performed, i.e. the time at which drum strokes
are executed to follow the rhythm. I propose to build a neurocomputational model about how the
brain implements the interaction between these two systems. I present 3 work packages (WP). WP
1 explains the model and the tools needed to build it. WP 2 proposes two experimental studies to
test the model predictions in distinct action sequences contexts. Various model versions will be
compared through model comparison techniques. WP 3 further tests the functional role of the
subthalamic nucleus (STN), the brain area suggested to subtend the interaction between both
systems. To treat their tremor symptoms, implanted electrodes in Parkinson’s disease patients
(PDP) stimulate the STN but simultaneously disrupt its functional role. In collaboration with a
specialized lab, I will stimulate STN in PDP while they solve my experimental tasks. I will investigate
which model version best accounts for PDP data when I simulate STN stimulation. Overall, I aim to
elucidate how humans learn to produce time-dependent action sequences.