Neural mechanisms of sensorimotor integration measured with fMRI

It has been proposed that the brain combines prior knowledge of movement dynamics and sensory information from peripheral receptors, to obtain optimal force estimation. The present study investigated the neural mechanisms underlying this function. Subjects (N=20) trained in a torque-matching task to acquire knowledge of the dynamics and immediately after training, performed the same task under two conditions. Under pulse*c condition, torque pulse magnitude was always the same pulse*v condition, torque pulse magnitude changed pseudo-randomly according to a probability distribution (mean of 4.6 Nm ± 0.4). Hemodynamic responses corresponding to torque pulse compensation under the two conditions were contrasted. Results showed that the subjects compensated more accurately under pulse*c condition (4.20 ± 0.50 Nm), relative to pulse*v condition (4.00 ± 0.66 Nm). Pulse*v-pulse*c contrast revealed significant activation (P <0.001, uncorrected) of the right BA 6 and right BA 7. These results suggest that activity in right BA 6 was related to a computational process that was guided by the statistical properties of the perturbation magnitude, which led to the determination of an optimal response based on accumulation of prior knowledge. Activity of right BA 7 was related to retrieval of previously acquired neural representation from memory.