PROJECT SUMMARY: Neuroimaging methods such as functional MRI and magneto- / electroencephalography (MEG/EEG) cannot directly reveal causal relationships between regional brain activity and behavior. To allow causal inference, transcranial magnetic stimulation (TMS) has been used to perturb local cortical activity to create temporary "virtual lesions”. However, even simple behavioral tasks employ widely distributed brain networks with multiple nodes activated at different millisecond-level latencies, whereas today’s TMS technology is mainly limited to single-channel devices that target only one brain area at a time. The mismatch between the large number of network nodes and the small number of cortical areas we can target with present TMS devices forms a critical barrier in exploring network-level causal inferences in the human brain. To remove this barrier, we need TMS technology that can perturb multiple cortical locations at specific processing stages. Removing this barrier would open entirely new avenues for gaining insight into how complex cognitive functions emerge from cortical networks. For the identification of neural networks underlying cognitive operations, two steps are necessary. First, network nodes are identified as locations where the strength of the TMS-induced electric field and relevant behavioral or neuroimaging-based variables maximally correlate. This can be achieved efficiently using the unprecedented ability of the multichannel TMS array to generate different customized electric field patterns in rapid succession. Next, information flow between the network nodes is explored by stimulating the nodes in rapid temporal succession. For this, the ability of the multichannel array to switch between electric field patterns in milliseconds is crucial. The network-level mapping capabilities of the instrument will be first verified in a testbench experiment in a known network (motor system). We then proceed to studies that identify network nodes underlying language comprehension. Finally, we will use the multifocal stimulation approach to investigate information flow in the language comprehension network and develop a neural mass model of the underlying neuronal circuits for a theoretical basis. In the long term, the wider society may benefit from all applications of the proposed research through network-level TMS therapies that are optimized for modulating functional connectivity between brain regions involved in critical functions such as hearing, speech, and language processing.