The brain consists of many types of neurons which make a complex network. The brain’s network performance has been investigated through the analysis of its input-output relationship using electrodes. Recently, optical stimulation methods using a photosensitive ion channel, such as channelrhodopsin-2 (ChR2) from Chlamydomonas reinhardtii, has been developed as a powerful tool for the investigation of neural circuits in vivo and in vitro. Neurons can also be inactivated by light-driven Cl- or H+ pumps, such as halorhodopsin (NpHR) and archaerhodopsins (Arch, ArchT). To manipulate neuronal activity positively or negatively with desired spatiotemporal patterns, the irradiation system has to be optimized. Spatial light modulators based on the digital mirror device (DMD) is an ideal tool for patterned photostimulation because each micro-mirror can be turned on and off at high frequencies up to several kHz (Sakai et al., 2012). Here, we improved the multicolor irradiation system using a DMD-based mercury lamp projector and a multiband-pass optical filter in combination. It is expected that ChR2 is activated by the filtered blue light whereas ArchT is activated by the filtered yellow light. To test this, we recorded the activity of a neuron, which expressed both ChR2 and ArchT, in the hippocampal slice under whole cell patch clamp. The activity was switched up and down simply with the changes of irradiating colors. Our multicolor irradiation system enables one to manipulate neuronal activity with desired spatiotemporal patterns. In the dentate gyrus (DG) of hippocampus, a hilar mossy cell (HMC) receives excitatory inputs convergently from a group of granule cells (GCs) and sends excitatory output divergently to another group of GCs. Based on this anatomical feature HMCs would regulated the dynamic state of hippocampus through positive feedback association with GCs. Here, using optogenetics and the DMD-based irradiation system, we tested whether the magnitude of HMC activity is dependent on the spatiotemporally patterned inputs. Hippocampal slices were prepared from a Thy1.2-ChR2-Venus transgenic rat line in which GCs are expressing ChR2 (Ji et al., 2012). Whole cell patch clamp recordings were made from HMCs, which is identified by the electrophysiological characteristics and the morphology. Under current clamp, spontaneous action potentials were observed at low frequency. However, bursts of high-frequency action potentials were often generated in synchronous with rhythmical irradiation of whole DG at 10 Hz. The bursting activity continued for 10-40 s after cessation of irradiation. On the other hand, the activity of HMC was rather inhibited when the granule cell layer was selectively irradiated with the same rhythm. It is suggested that the bursting activity of HMC is dependent on the spatiotemporal pattern of GC inputs.