Neuronal ensembles, groups of simultaneously active neurons, are considered to be one of basic forms of information representation in the mammalian cortex. However, the mechanisms underlying the modulation of neuronal ensembles yet remain to be fully understood. Here, we investigated external input-dependent modulation of neuronal ensembles in cultured neuronal networks with modular connectivity mimicking the mammalian cortex. The modular neuronal networks were fabricated by growing cultured rat cortical neurons in microfluidic devices. A fraction of the neurons was stimulated using optogenetics, and the network activity was recorded by fluorescence calcium imaging [1,2]. From the recorded data, we extracted collective activity events, defined as the synchronized activity of multiple neurons, and found that collective activity events modulated before and after the stimulation. Using DBSCAN to cluster collective activity events for detecting neuronal ensembles, we found stimulation changed the frequency of neuronal ensembles. Furthermore, in the networks with low modularity, the variation of intrinsic neuronal ensembles was too low to be modulated by the input. Our findings suggest that brain-like neuronal networks may have the characteristic of being susceptible to changes in activity patterns in response to external inputs.The work was partly supported by MEXT Grant-in-Aid for Transformative Research Areas (B) "Multicellular Neurobiocomputing", JSPS KAKENHI, JST CREST, and Tohoku University RIEC Cooperative Research Project Program.[1] Sumi et al., Proc. Natl. Acad. Sci., 120, e2217008120 (2023).[2] Yamamoto et al., Sci. Adv., 9, eade1755 (2023).