During the control of reaching and grasping, motor commands must be integrated with sensory feedback to enable robust movements, though it remains unclear how sensorimotor cortical areas achieve this integration. In recent years, modern analytical approaches based on population dynamics have increased our understanding of the cortical control of reaching. Yet, these methods have largely treated motor cortical activity as a feedforward generator of movement. In my talk, I will explore the effect of sensory feedback on primary motor (M1) and somatosensory (S1) cortical activity. First, I will show a surprising impact of sensory stimulation on motor cortical dynamics. Intriguingly, this effect had little behavioral consequence, highlighting the natural robustness of the movement-related cortical dynamics and a possible independence of movement- and sensory-related information. I will then explore the hypothesis that such dissociation between sensory and motor dynamics is enabled by interactions within low-dimensional “communication subspaces” relating the M1 and S1 population activity. In an unconstrained reaching and grasping task, these subspaces exhibit distinct dynamical features, transitioning between largely input-driven sensory-related dynamics in S1 to the smooth, rotational dynamics commonly observed within M1. Using this framework, I will illustrate that inputs to M1 influence the cortical dynamics to enable corrections to behavioral errors. This work provides a novel approach to disentangling specific dynamical features within cortical population activity and highlights the importance of sensory feedback for motor cortical dynamics.