This study reports the direct optical control of oscillatory neurons based on volatile threshold switching in V3O5, a new bolometric material. The devices exhibit electroforming-free operation with tunable switching parameters, including threshold and hold voltages, through optical illumination. Conductive atomic force microscopy (C-AFM), in situ thermal imaging, and lumped element modeling reveal that the changes in switching parameters arise from an increase in overall oxide film conductivity due to photoconductive and bolometric effects. V3O5 is identified as a new bolometric material with a high temperature coefficient of resistance (TCR) of −4.6% K−1 at 423 K. The utility of these devices is demonstrated by their application in in-sensor reservoir computing and spike frequency tuning for image segmentation using simulated arrays of devices. The study highlights the potential of V3O5 for neuromorphic computing, offering a scalable, compact, and power-efficient solution for direct optical input in spiking neural networks (SNNs).This study reports the direct optical control of oscillatory neurons based on volatile threshold switching in V3O5, a new bolometric material. The devices exhibit electroforming-free operation with tunable switching parameters, including threshold and hold voltages, through optical illumination. Conductive atomic force microscopy (C-AFM), in situ thermal imaging, and lumped element modeling reveal that the changes in switching parameters arise from an increase in overall oxide film conductivity due to photoconductive and bolometric effects. V3O5 is identified as a new bolometric material with a high temperature coefficient of resistance (TCR) of −4.6% K−1 at 423 K. The utility of these devices is demonstrated by their application in in-sensor reservoir computing and spike frequency tuning for image segmentation using simulated arrays of devices. The study highlights the potential of V3O5 for neuromorphic computing, offering a scalable, compact, and power-efficient solution for direct optical input in spiking neural networks (SNNs).