In this work, we seek to characterize the mesoscopic circuitry of the sensorimotor system across layers, columns and brain areas. We use a blood volume and BOLD signal sensitive fMRI methodology during task and rest with 0.78 mm resolution. We find ‘columnar’-dependent connectivity, representing the same body parts (e.g. digits) across different sensory and motor areas. In the hand knob, we find ‘columnar’ networks with a finer spatial scale than the regime of individual digit representations. Layer-dependent differences of task-induced and resting-state fMRI are indicative of different input-output-driven activity.