Neuronal cultures and human stem-cell-derived organoids are fundamental building blocks of neuroscientific research and future personalized medicine. However, in-vitro networks show considerably more synchronous bursting activity than networks in-vivo [1]. One leading theory proposes that under a lack of external input, maladapted network’s plasticity leads to a pathological collective activity pattern: long periods of quiescence are interspersed with bursts of strong, synchronous activity [2, 3]. In theoretical simulations, weak external input can reduce this bursting behavior. However, experimental proof is still missing [4, 5, 6]. To address the challenge of missing external input, we developed a system providing month-long arbitrary 2D light stimulations with a 16x16 LED array to optogenetically light-sensitized neurons whilst enabling a simultaneous redout with an incubator-resistant multi-electrode setup (MEA2100-Mini, MultiChannelSystems) [7]. Custom made protocols enable us to perform an extensive range of theory-driven experiments, including detecting receptive fields via spike-triggered averages, pattern completion experiments after prolonged stimulus exposure, or measuring the cell culture’s cell composition after the experiments. With this platform we exposed neuronal cell cultures (N~15) to temporally and spatially-uncorrelated stimulation (Poisson Noise) of varying input strength for a period of one week and compared their activity patterns with that of unstimulated control cultures (N~15). Preliminary results indicate that external input does indeed change the collective activity regime. Inter-spike interval distributions shifted from multi-modal to gamma-like, reflecting the reduction of bursts and a transition to Poisson-like firing. Correlations in the activity between units markedly decreased compared to unstimulated controls. These preliminary results suggest that chronic stimulation can indeed harness the intrinsic plasticity and homeostasis to shape the functional connectivity within the cultures. The cultures conditioned in this way are promising model systems for all research areas from computational neuroscience to high-throughput screening.