A central problem in neuroscience is to understand how areas in the brain communicate. Inputs to brain areas largely influence the intrinsic dynamics of each region, which shapes ongoing computations. Hence, characterizing how inputs drive different population responses is key to understanding how brain circuits are controlled to generate different behaviors. To this end, we developed an inter-areal patterned microstimulation (uStim) protocol that allowed us to finely manipulate the activity of a neural population in one brain area while simultaneously recording the activity of a second population in a different area. In macaques implanted with dual 96-channel Utah arrays, we manipulated the activity of different brain regions by electrically stimulating combinations of electrodes in one of the arrays while recording the effect on the other. We were able to generate a rich repertoire of activity patterns and to identify the dimensions along which different inputs drove the neural population. We used our protocol to study how the prefrontal cortex (PFC), a high-order cognitive area that displays robust memory encoding, is influenced by inter-hemispheric inputs. For this, we assessed the impact that different uStim patterns applied to the right-hemisphere PFC (RH-PFC) had on the contralateral PFC (LH-PFC) during a memory task. We found that the stimulation patterns selectively affected PFC activity in different dimensions and biased performance and reaction times. Our approach provides a causal tool to link brain activity and behavior at greater granularities, and paves the way towards data-driven models that explain how brain areas dynamically interact to produce computations.